Integrating Artificial Intelligence, Circulating Tumor DNA, and Real-World Evidence to Optimize Hematologic Clinical Trials: Toward Adaptive and Learning Trial Designs.

1. Introduction

Hematological malignancies represent the fifth most common group of cancers and are characterized by considerable molecular and clinical heterogeneity, which limits the sensitivity and generalizability of traditional, rigid trial designs. However, the integration of artificial intelligence (AI), circulating tumor DNA (ctDNA), and real-word evidence (RWE) is now creating new opportunities as “learning trials”. In this model, AI functions as the analytic core, employing machine learning (ML) and natural language processing (NLP) to mine high-dimensional electronic health records (EHR) and multi-omics data, efficiently phenotyping patients and identifying a molecular subtype suitable for highly targeted studies [ 1 , 2 , 3 , 4 , 5 ]. Simultaneously, circulating tumor DNA (ctDNA) serves as a minimally invasive “liquid biopsy” and dynamic biomarker of minimal residual disease (MRD), capable of detecting molecular relapse and treatment resistance months before conventional imaging or clinical progression, and increasingly being incorporated into adaptive and dose-finding trial designs to shorten study duration and optimize dosing decisions [ 6 , 7 , 8 ]. RWE derived from routine clinical practice, when standardized and analyzed using AI, can generate external or synthetic control arms (SCAs). This approach enhances trial feasibility and ethical acceptability for rare, relapsed, or molecularly defined patient populations. It also improves generalizability compared to the restrictive eligibility criteria of traditional randomized controlled trials (RCTs) [ 1 , 3 , 5 , 9 ]. Taken together, RWE provides essential context by using data from routine clinical practice to generate SCAs, thereby, reducing the ethical and logistical burden of recruiting placebo groups in rare or relapsed populations.

The integrated framework aims to replace rigid, uniform trial structures with hybrid adaptive designs. In these designs, AI-derived RWE refines eligibility criteria and cohort selection, while ctDNA provides dynamic surrogate endpoints and MRD assessments. Traditional clinical endpoints maintain the rigor necessary for regulatory validation [ 10 , 11 ]. Emerging data from AI-facilitated, biomarker-adaptive, and hybrid RWE-RCT approaches suggest a potential reduction in development timelines. These gains arise from expedited surrogate endpoints combined with external or synthetic controls, without compromising statistical validity or regulatory compliance [ 10 , 12 , 13 , 14 ]. This effectively reframes drug development from a static, protocol constrained experiment into a continuously learning clinical ecosystem that iteratively updates trial conduct and therapeutic decisions in parallel with each patient’s evolving disease course [ 3 , 10 , 15 ]. Despite the rapid evolution of AI, ctDNA, and RWE, these domains are often explored in isolation. However, their true transformative potential emerges when they are integrated within a unified clinical trial framework. AI provides the computational infrastructure for predictive modeling and adaptive decision-making, ctDNA enables real-time molecular monitoring of disease dynamics, and RWE ensures external validity by capturing outcomes across diverse patient populations. Together, these technologies enable the development of adaptive, continuously learning clinical trial systems that bridge the gap between controlled experimental settings and real-world clinical practice. Therefore, this review aims not only to summarize these advances individually but to highlight their synergistic integration as a foundation for next-generation hematologic clinical trials.

2. Challenges in Conventional Hematology Clinical Trials

2.1. Diseases and Trial Design Complexity

The field of hematology has made significant advances in the biological understanding and therapeutic management of hematological malignancies; however, traditional clinical trial frameworks often fail to keep pace with this rapid progress. Trial methodologies are challenged by marked disease heterogeneity, multiple prior therapeutic exposures, and growing complexity of targeted and immunotherapeutic agents; collectively, these factors necessitate larger patient cohorts, more granular outcome measures, and longer study durations to detect clinically meaningful treatment effects [ 16 ].

2.2. Recruitment, Representativeness and Equity

Restrictive eligibility criteria, demanding visit schedules, and the centralization of trials in specialized academic centers limit recruitment and retention, particularly among older adults, patients with comorbidities, and individuals from geographically rural or socioeconomically disadvantaged communities. These limitations reduce both statistical power and generalizability of trial findings [ 16 , 17 , 18 ]. Additional barriers, including travel demands, caregiving responsibilities, insurance constraints, and the centralization of hematologic malignancy and cellular therapy trials at comprehensive cancer centers further exacerbate enrolment inequities [ 16 , 18 ].

2.3. Outcome Selection, Reporting, and Patient-Centredness

Selective outcome reporting and underreporting of patient-reported outcomes (PROs) further constrain the interpretability and patient-centeredness of hematology trials. Systematic reviews have documented frequent discrepancies between registered and published endpoints in hematology RCTs, including, the addition, omission, or demotion of primary outcomes that often favor statistically significant result [ 19 ]. In phase III trials of hematologic malignancies, although nearly three-quarters prespecify PROs, only a minority report them in primary publications, typically when findings are neutral or favorable [ 17 , 19 ].

2.4. Regulatory, and Modality Specific Challenges

Traditional clinical trial frameworks struggle to incorporate innovative therapies such as cellular treatments, NK-cell and CAR-T products, and therapeutic vaccines into regulatory systems originally designed for small molecule chemotherapies and conventional biologics [ 20 , 21 , 22 ]. This creates uncertainty in endpoint selection, safety monitoring, and long-term follow-up requirements [ 20 , 22 , 23 ]. As a result, interest is rapidly increasing in decentralized clinical trials (DCTs) and hybrid designs that incorporate telemedicine, home-based procedures, wearable technologies, and community-based sites to reduce travel burden, expand access, and better align trial conduct with real-world practice. Evidence from cross-disease evaluations of DCTs suggests that, when carefully implemented, decentralized methodologies can enhance accessibility, participant diversity, and patient engagement. However, these approaches also introduce challenges related to data reliability, digital literacy gaps, data privacy and cybersecurity risks, and equity considerations, all of which require targeted mitigation strategies to prevent the widening of existing disparities [ 17 , 24 , 25 , 26 , 27 , 28 ]. To maximize inclusivity, recent frameworks emphasize purposeful engagement of patients and individuals with lived experience in trial design, the use of digital tools to support community-based recruitment and bidirectional communication, and the implementation of deliberate diversity, equity, and inclusion (DEI) strategies tailored to underserved populations [ 17 , 25 , 28 ]. Moreover, the regulatory pathway for this AI-integrated framework remains unclear, especially for dynamic or continuously learning decision support models. Leading agencies such as the FDA and EMA have begun issuing high-level guidance and risk-based principles for AI/ML in drug development, clinical trials, and regulatory workflows, yet these fall short of establishing a standardized, comprehensive approval process for adaptive AI systems [ 29 ]. Dynamic AI systems, including adaptive models and in silico “digital twin” approaches, exacerbate these uncertainties by conflicting with conventional one-time validation paradigms and posing challenges in managing continuous learning, dataset shift, bias, and performance drift that should be managed within a compliant framework [ 30 , 31 ].

Finally, the integration of RWE derived from registries and routine clinical practice into the evidence hierarchy provides an important complement to traditional RCTs, enabling assessment of effectiveness, toxicity, and access across broader, more diverse hematology populations than can feasibly be enrolled on protocolized studies [ 24 , 25 , 32 ]. When combined with decentralized and adaptive trial designs, RWE-informed eligibility criteria, and robust PRO and equity frameworks, these innovations offer a pathway to address the inherent limitations of conventional hematology trials and to support more representative, patient-centered, and methodologically rigorous evaluation of emerging therapies [ 33 , 34 ].

3. Artificial Intelligence in Hematological Malignancies

Artificial intelligence (AI) is increasingly reshaping the field of hematology, with a notable acceleration in both research output and clinical-oriented applications since approximately 2019. Contemporary AI methodologies including machine learning and deep learning approaches ( Figure 1 ), are being incorporated across multiple aspects of hematologic disease management, including early disease detection, diagnostic classification, treatment planning and personalization, prognostic risk stratification, patient monitoring and management, and clinical decision support systems [ 35 , 36 , 37 ]. The expansion of AI has been driven by advances in ML and deep learning (DL) methodologies, alongside the increasing availability of large-scale datasets from digital pathology, flow cytometry, and EHRs. The COVID-19 pandemic further accelerated this digital transformation, reinforcing the central role of AI applications in the future hematology practice [ 38 ].

Hematological malignancies represent a broad and heterogeneous group of neoplasms, including leukemia, lymphoma, and multiple myeloma, and are characterized by complex underlying biology and highly heterogeneous clinical behaviors [ 39 , 40 , 41 , 42 ]. This diversity, reflected in distinct genetic, epigenetic, and microenvironmental characteristics, translates into widely variable responses to antineoplastic therapies [ 43 ]. Historically, diagnostic evaluation and treatment selection have relied predominantly on histopathologic assessment and clinical judgment, resulting in relatively standardized approaches despite substantial variability in clinical outcomes [ 44 , 45 ].

However, the advent of AI technologies including ML algorithms, NLP, and advanced imaging analytics, has the potential to enhance precision medicine in this domain. By leveraging rich multi-omics datasets (including genomics, transcriptomics, and proteomics), AI methods enable the identification of novel molecular biomarkers, support the prediction of disease evolution, and guide the personalization of treatment decisions for individual patient profiles ( Figure 2 ) [ 46 , 47 ]. The field has demonstrated remarkable progress, with measurable improvements in diagnostic accuracy, risk stratification, and patient outcomes across multiple hematological malignancies.

For instance, the integration of AI into predictive modeling can markedly improve risk stratification, enabling clinicians to implement personalized treatment strategies guided by patient-specific genomic and phenotypic characteristics, ultimately improving survival outcomes and quality of life [ 48 ]. Moreover, generative AI holds promise for accelerating the development of novel therapeutic strategies by simulating treatment responses and optimizing drug combinations, proving especially advantageous for complex conditions like acute myeloid leukemia (AML) [ 49 ]. However, the effective implementation of these advanced technologies depends on addressing challenges related to data quality, algorithmic bias, and interpretability, necessitating a collaborative approach among healthcare professionals, data scientists, and regulatory bodies to ensure equitable access and effective integration into clinical practice [ 37 , 50 ].

As summarized in Table 1 , contemporary AI-based investigations demonstrate a wide range of applications across hematologic malignancies. In leukemia, Kilic Gunes et al. (2025) applied convolutional neural networks (CNNs) and random survival forests (RSF) to peripheral blood smear and flow cytometry data, achieving over 95% accuracy for AML detection and a concordance index (C-index) above 0.77 for prognostic risk stratification, thereby facilitating earlier diagnosis in resource-limited settings [ 38 ]. Thirugnanasambandam et al. (2025) implemented Random Forest models and neural networks using clinical and mutation profiles to predict relapse and guide personalized therapeutic decisions [ 51 ]. Li et al. (2025) employed XGBoost coupled with SHAP-based explainability on biomarker data to predict 30-day mortality in high-risk patients [ 52 ]. In lymphoma, Carreras et al. (2024) applied CNNs with transfer learning on histopathology images for subtype classification, while Uppal et al. (2025) leveraged ResNet and DenseNet architectures on whole-slide images to achieve precise differentiation among lymphoma categories [ 53 , 54 ]. Carreras et al. (2022) further combined ensemble survival ML models with genomic and clinical variables to estimate patient outcomes [ 55 ]. In multiple myeloma, Eweje et al. (2021) integrated CNNs with radiomic features extracted from bone marrow imaging to detect lesions and developed DL ensemble approaches that predict treatment response and survival based on biomarker profiles [ 56 ]. Table 2 summarizes hematological malignancies studies, highlighting different AI applications and their associated clinical outcomes.

Collectively, these advances significantly enhance diagnostic accuracy, prognostic assessment, and precision-medicine strategies in hematologic malignancies.

4. Enhancing Trial Efficiency Through ctDNA in Hematologic Malignancies

As an emerging field, ctDNA is increasingly recognized as a non-invasive biomarker with substantial clinical value in hematologic malignancies [ 66 ]. ctDNA represents the proportion of cell-free DNA that originates from malignant cells, released into the bloodstream through apoptosis, necrosis or active secretion ( Figure 3 ) [ 67 , 68 , 69 , 70 ].

Studies have shown that ctDNA contains clinically relevant mutations, reflects disease burden, and can detect therapy-resistant clones that may not be identified by conventional tissue biopsies [ 70 ]. Moreover, ctDNA dynamics correlate with treatment response, the early detection of relapse, prognosis, and estimation of drug resistance, and can support MRD monitoring, targeted therapy selection, and immunotherapy surveillance [ 71 , 72 , 73 , 74 ]. For example, ctDNA characterization identified markers of relapse and refractoriness in mantle cell lymphoma (MCL), demonstrating its potential utility in capturing tumor heterogeneity, particularly alterations involving the BTK gene [ 75 , 76 ].

Liquid biopsy is a well-established method for the analysis of ctDNA [ 77 , 78 ]. However, it still requires standardization due to limited sensitivity and variability in tumor-specific analysis methods [ 79 , 80 , 81 ]. Recent technologies, including next-generation sequencing (NGS), duplex sequencing, phased variant detection, and droplet digital PCR (ddPCR), have expanded the capability to detect tumor DNA in plasma or serum, rendering ctDNA analysis increasingly clinically actionable [ 80 , 82 , 83 ]. Beyond its role in traditional diagnostics, ctDNA offers multidimensional and quantitative information that can be integrated into trials to enhance adaptive decision-making, improve outcomes assessment, and increase trial efficiency [ 84 ].

Real-time tracking of disease progression using ctDNA plays a very crucial role in hematologic malignancies, as it can identify molecular recurrence at earlier stages than conventional imaging modalities such as positron emission tomography (PET)/CT [ 85 ]. Prediction of residual metabolic disease activity followed by MRD assessment, represents a key treatment milestone at the end of therapy that allows for the design of more streamlined clinical trials [ 85 , 86 , 87 , 88 ]. Modern basket trials in hematologic malignancies increasingly integrate multi-omics biomarkers to guide treatment strategies. Moreover, adaptive trial designs are incorporating real-time longitudinal omics data to dynamically adjust treatment and dosing [ 89 ]. In clinical trials, ctDNA can be employed to select patients based on the presence of specific molecular alterations for enrolment, to stratify study populations into high-risk and low-risk cohorts, and to monitor response to anticancer treatments [ 90 ]. A comprehensive overview of clinical trials in hematologic malignancies employing ctDNA-based approaches to guide therapeutic interventions is presented in Table 3 , highlighting the transition toward molecularly driven trial designs.

Recently, in patients with peripheral T-cell lymphoma (PTCL), ctDNA analysis revealed that high pretreatment ctDNA burden and recurrent mutations, particularly TP53 , STAT3 , TET2 , DNMT3A , and RHOA , were associated with poor survival outcomes and high relapse rates. Notably, end-of-treatment (EOT) MRD negativity was rare, even among patients achieving complete or partial remission, and persistent MRD strongly predicted inferior progression-free and overall survival [ 91 , 92 , 93 , 94 ].

5. Focusing on Real-World Evidence

In hematologic malignancies, RWE offers a complementary perspective to inform optimal study design, elucidate treatment algorithms, and the personalization of patient care beyond the scope of standard RCTs [ 95 , 96 , 97 ]. RWE derived from real-world data (RWD), which includes electronic health records, disease registries, insurance claims databases, treatment assignment systems, tumor genomic profiling records, and patient outcome reports, provides a valuable complement to traditional RCT evidence [ 95 , 97 , 98 , 99 ]. These prospective cohort studies, which evaluate interventions in routine clinical settings while retaining randomization, are non-randomized interventional studies that follow prospective protocols without random allocation. In contrast, RCTs may employ external control arms constructed from RWD to augment or replace traditional randomized comparators [ 100 ].

Within the digital health data system, EHRs contain routine patient records generated by healthcare providers [ 101 ]. Clinical decision support tools and structured data reports are utilities within EHRs that help users identify high-risk drug allergy labels and correct or remove inaccurate or outdated entries in the patient’s chart [ 102 ]. Comprehensive health insurance claims databases, as a source of RWE, allow investigation of geographic variations (regional differences) in key metrics such as disease incidence, patient demographics, treatment modalities (e.g., surgery, ablation, systemic therapy), and potentially clinical outcomes [ 103 , 104 , 105 ].

RWE is particularly valuable in this setting as it can provide evidence on treatment efficacy and safety across broader patient populations. It can help detect rare adverse events and inform lifelong patient care, as well as facilitate regulatory decision-making and post-marketing surveillance [ 106 ].

For example, RWE indicates that Carfilzomib, Lenalidomide, and Dexamethasone (KRd) in relapsed/refractory multiple myeloma have an acceptable safety profile and effective clinical outcomes with a median progression-free survival (PFS) of 26 months and a 2-year overall survival (OS) rate of 70% [ 107 ]. A US-based real-world study of Chimeric antigen receptor T (CAR-T) cell therapy evaluated treatment patterns and clinical outcomes following CAR-T failure or relapse. Results showed that 44.7% of patients initiated subsequent treatment at a median of 263 days post-CAR T infusion, with systemic chemotherapy representing the most commonly used conventional approach [ 108 ].

For trials in which enrolling the randomized control groups is challenging, such as those involving rare subtypes and high-risk populations, real world data can be used to construct synthetic or external control arms (ECAs). These ECAs provide a comparator in single-arm or early-phase studies [ 109 , 110 , 111 ]. Recent data suggest that RWD-derived ECAs may have great potential, particularly when traditional RCTs are not feasible in certain hematologic settings [ 109 , 112 , 113 ]. However, synthetic control arms have some limitations, including bias due to unmeasured confounders; this bias could potentially be overcome by using careful cohort selection, covariate harmonization and sensitivity analyses. Selukar et al. described a blinatumomab case study in which investigators accounted for potential unmeasured confounders, such as the different index dates for the trial and external cohort, namely the time of blinatumomab initiation in study and the start of the last salvage therapy in the external dataset. The author sought to reduce discrepancies in disease classification and response assessment by harmonizing both datasets through standardized, shared data definitions [ 114 ]. Another investigator proposed a (Partial Bias Correction) PBC approach to mitigating bias due to unmeasured covariates by limiting the maximum tolerable amount of bias and calibrating the trade-off between bias and variance when borrowing strength from SC. In reality, statisticians and clinical partners may find it difficult to decide the largest acceptable bias [ 115 ]. While data collection is an integral part of research and drug development, post-approval therapies face ongoing regulatory safety concerns. In this context, RWE offers a pathway for post-marketing surveillance and long-term safety monitoring [ 116 , 117 , 118 ].

6. Regulatory Considerations

6.1. Artificial Intelligence: Approval, Safety, and Liability (FDA and EU AI Act)

A central legal and regulatory challenge associated with the use of AI in healthcare concerns the liability for errors, adverse outcomes and negative consequences. AI networks and systems, particularly those based on machine learning, are trained on large sets of data and they evolve dynamically over time, making their behavior difficult to fully predict in all clinical scenarios, which may lead to inappropriate diagnostic interpretations or treatment recommendations, raising concerns regarding patient safety, accountability, and regulatory oversight [ 119 , 120 ]. The examination of the legal and regulatory framework has led to the emergence of definitions for both synthetic data and RWD. At present, the EU Data Governance Act (DGA) is the primary legal instrument that explicitly references synthetic data. Following recent developments in the EU AI Act, synthetic data are increasingly associated with “general-purpose AI models”, accompanied by specific requirements regarding risk management, transparency, and methodology [ 121 ].

6.2. ctDNA: Diagnostic Tests, Clinical Validation

Gene profiling technologies provide a distinct view into tumor biology although they are accompanied by significant challenges in biomarker discovery, analysis, and validation. These challenges include technical variability, interindividual heterogeneity and the critical requirement for standardization. Such limitations can impact the reliability and reproducibility of study results, frequently delaying regulatory approval and hindering the translation of biomarkers into routine clinical practice [ 89 ]. In clinical trials, ctDNA MRD is increasingly used to select and stratify patients. MRD positivity is closely correlated with poor clinical outcomes—an opportunity for earlier therapeutic intervention in cancer management [ 122 ]. The U.S. Food and Drug Administration has already approved a series of ctDNA-based companion diagnostic tests, supporting the analytical validation of ctDNA as a biomarker for regulatory use in early-stage solid tumors. There are several MRD detection methods such as the tumor-informed method, panel-based next-generation sequencing (NGS), tumor-naïve methods or a reduced candidate gene panel [ 123 ]. ctDNA might be influenced by the histology, grade, stage, or size of the tumor so that the timing for this measurement would need to be established in consultation with the FDA during the submission process and supported by the performance characteristics of the study test relative to disease characteristics and tumor biology. The variability of results from ctDNA assays can potentially be reduced by the adoption of standardized pre-analytical protocols, through stringent analytical validation with clearly established detection limits, use of error-corrected sequencing technologies (unique molecular identifiers, UMIs), the harmonization bioinformatics pipelines and participation in external quality control programs based on regulatory and consensus recommendations [ 123 , 124 , 125 , 126 ].

6.3. RWE: Data Privacy, HIPAA/GDPR, Quality Standards

RWE may be used to evaluate the effectiveness and safety of treatments; however, its application in hematology remains carefully regulated, with relatively few approvals by the U.S Food and Drug Administration (FDA) or the European Medicines Agency (EMA) relying primarily on RWE. Key challenges include variable data quality, insufficient sample sizes, unmeasured confounding, missing data and heterogeneous endpoint definitions, all of which contribute to an increased risk of bias [ 127 ]. Secondary uses of health data are subject to stringent safeguards aimed at minimizing re-identification risk and protecting patient privacy. In the United States, such use is governed by the Health Insurance Portability and Accountability Act (HIPAA), while in the European Union, it is regulated under the General Data Protection Regulation (GDPR) [ 128 , 129 , 130 , 131 ].

7. Ethical Considerations

7.1. Artificial Intelligence: Bias, Explainability, Accountability, and Equitable Access

In healthcare, the introduction of biased AI systems raises fundamental ethical concerns related to justice, non-maleficence and patient autonomy. Algorithmic bias can disproportionately affect underrepresented populations, while limited explainability may hinder informed consent and shared decision-making, as clinicians and patients may struggle to interpret AI-generated recommendations. These ethical considerations can undermine patient confidence and clinical care and compromise decision-making. Consequently, the development and implementation of AI in healthcare must prioritized transparency, accountability and equity to ensure fair and responsible clinical adoption [ 120 , 132 ]. There may be an unintended effect on patient autonomy with AI. The opacity of a subset of AI algorithms can conceal rationales and affect autonomy and the quality of informed consent. It also may influence autonomy by limiting choice, as in deciding about human- vs. AI-led health care treatment decisions. Building out systems and recommendations for transparent reporting and organized AI model evaluation can improve both the credibility and ethical deployment of AI in healthcare [ 133 , 134 , 135 ].

7.2. ctDNA: Consent, Genetic Findings, Clinical Utility

Access to genomic and multi-omics data can contribute to precision medicine by enabling highly individualized, “custom-fit” therapeutic strategies. However, it also raises significant ethical and privacy concerns, as misuse or unauthorized disclosure of such data can compromise patient confidentiality. Germline genomic information which remains stable throughout an individual’s lifetime, constitutes a uniquely identifying biometric, and its exposure can pose substantial privacy risks [ 89 , 136 ]. Pathologists and laboratory professionals must develop a deep under-standing of the many policy dimensions that influence the molecular diagnostic considerations because they drive molecular diagnostics: their availability and accessibility. Attention needs to be paid during training to the responsible use of genomic data, laying out for trainees what are their ethical responsibilities as custodians of patient confidentiality and transparent participants in testing [ 137 , 138 , 139 ].

7.3. RWE: EHR-Derived Data, Patient Consent, Re-Identification Risk

Practical barriers, such as feasibility, data governance and long-term sustainability often restrict timely access to multi-national RWD due to diverse legal and ethical considerations, with patient consent and privacy protection frequently resulting in delays in achieving data availability, thereby constraining the ability to generate timely safety evidence and inform regulatory and clinical decision-making [ 140 , 141 ]. Indeed, many of these data can be obtained as part of standard clinical care rather than research and patients may not be aware that they are being used in a secondary capacity—eroding traditional informed consent paradigms and emphasizing the rationale for transparent, broad or dynamic consent strategies. Whilst datasets often are de-identified, developments in data linkage and genomics increase the risk that patients could be re-identified and suffer stigma, discrimination or psychological harm, particularly in rare hematologic diseases [ 142 , 143 , 144 , 145 ].

8. Limitations in Current Implementation

In practice, the implementation of AI, ctDNA, and RWE continues to face substantial difficulties that limit their effectiveness. The limitations encompass suboptimal data quality and interoperability; algorithmic biases, dataset shift, and constrained generalizability; opaque “black-box” models undermining clinician confidence; and unclear, evolving regulatory expectations for validation and monitoring. Contemporary research emphasizes multicenter prospective AI trials incorporating clinical endpoints, the creation of explainable and fairness-conscious models, and risk-stratified regulatory structures to facilitate AI integration throughout the clinical trial continuum [ 57 , 146 , 147 ]. In the context of ctDNA, the limitations involve biological heterogeneity in tumor shedding, insufficient sensitivity for low tumor burden detection, inconsistencies between circulating and tissue-based genotyping, and extensive variability in assays and bioinformatics workflows that obstruct standardization and threshold establishment. Future research trajectories involve expansive interventional studies assessing ctDNA-directed escalation/de-escalation, global harmonization of pre-analytical procedures and sequencing/bioinformatics workflows, and integrative multi-modal methodologies (e.g., fragment omics, methylation analysis, long-read sequencing, and ML signatures) to augment sensitivity and specificity [ 148 , 149 , 150 , 151 ]. The core limitations of RWE include heterogeneous EHR and claims data, issues with data missingness and miscoding, inconsistent capture methods, and serious risks of confounding, selection, and information biases that compromise causal inference. Current efforts prioritize robust, auditable real-world data (RWD) platforms; refined causal inference and trial design strategies (including external control arms, pragmatic studies, and hybrid trial-RWD integrations); and extensive application of AI/large language models to automate data abstraction and cleaning, thereby generating regulator-grade real-world evidence (RWE) for policy and practice [ 152 , 153 , 154 , 155 ].

9. Integration of AI, ctDNA, and RWE: A Combined Approach to Trial Design

While AI, ctDNA, and RWE each provide significant advancements independently, their integration creates a synergistic framework that enables a paradigm shift in clinical trial design. Rather than functioning as parallel tools, these components interact dynamically, as AI leverages ctDNA-derived molecular data to refine predictive models, while RWE continuously validates and recalibrates these models in real-world populations. This interconnected system transforms clinical trials into adaptive, feedback-driven ecosystems. The combination of AI, ctDNA, and RWE is propelling a novel era of hematologic clinical trial designs characterized by increased efficiency, adaptability, and being effectively patient-centric. This paradigm aligns the precision hematology transition from rigid, tumor-type based protocols toward biomarker-driven, master protocol designs, including basket, umbrella, and adaptive platform trials [ 3 , 156 , 157 , 158 , 159 ]. AI platforms now interrogate expansive clinicogenomic databases, EHRs, and structured master protocol datasets to inform nearly every aspect of clinical trial design. These tools are being used to refine eligibility criteria, identify biomarker-enriched cohorts, select sensitive endpoints, and determine realistic sample sizes, while also discovering predictive biomarkers thereby enhancing the probability of detecting therapeutic benefit [ 1 , 3 , 4 , 156 , 157 , 158 , 160 ]. Across multiple precision oncology programs, AI-facilitated matching of therapies to molecular profiles has been associated with better outcomes compared with non-matched treatment approaches, highlighting the importance of optimized clinical trial design ( Figure 4 ) [ 3 , 156 , 158 ].

ctDNA provides a complementary capability as minimally invasive tool for real-time assessment of MRD, tumor burden, and therapeutic response. Across diverse hematological malignancies and solid tumor settings, ctDNA-based MRD predicts relapse months before standard imaging and strongly enriches for high-risk populations [ 8 , 122 , 161 , 162 , 163 ]. This capability has enabled ctDNA-guided adjuvant trials include CIRCULATE-Japan, which incorporates concurrent escalation (intensified therapy for ctDNA-positive patients) and de- escalation (reduced or omitted chemotherapy for ctDNA-negative patients) arms, within a single adaptive platform [ 8 , 161 ]. Similar ctDNA-informed designs have been implemented in lung cancer and other settings, allowing earlier endpoint assessment, shortening trial duration, and minimizing patient exposure to ineffective therapies [ 122 , 161 , 163 , 164 ]. For instance, AI-enriched ctDNA analysis has achieved mutation-detection limits as low as 0.002% allele fraction, with relapse detection sensitivities above 90% and lead times of several months compared to conventional imaging. These advances have the potential to redefine how disease progression and treatment response are assessed in clinical trials [ 6 , 161 , 163 ].

Meanwhile, RWE derived from routine practice EHRs, registries, and master-registry trials offers the longitudinal depth and scale that traditional trials often lack. These data are increasingly used to develop external or hybrid control arms, refine inclusion and exclusion criteria, and ensure that biomarker-selected trials populations more closely resemble the diverse patients treated in real-world clinical settings [ 1 , 3 , 156 , 158 , 161 ]. When coupled with advanced NLP and large language models, radiology reports, and pathology descriptions can be converted into quantifiable variables for trial design and post-marketing surveillance [ 1 , 160 , 165 ]. Importantly, the integration of these modalities enables a continuous learning loop in which clinical trial data, molecular disease monitoring, and real-world outcomes are iteratively fed back into AI systems. This approach not only enhances trial efficiency but also supports real-time optimization of therapeutic strategies, ultimately improving patient outcomes.

This synergistic framework integrating AI, ctDNA, and RWE can seamlessly support adaptive platform trials, tumor-agonistic basket studies, N-of-1 designs, and ctDNA- guided adjuvant trials [ 3 , 8 , 157 , 158 , 159 , 161 ]. In an adaptive system, ctDNA monitoring combined with AI-based risk stratification can govern decisions regarding treatment escalation, de-escalation, or discontinuation. In N-of-1 and basket paradigms, multimodal models can identify the most promising targeted or immunotherapy combination tailored for an individual patient’s molecular profile [ 3 , 156 , 157 , 158 , 164 ]. As hematology trials widely embed routine molecular profiling and post-treatment ctDNA surveillance, and feed their results back into learning health system registries, clinical research is beginning to resemble a continuous feedback loop rather than a series of isolated experiments [ 3 , 156 , 158 , 159 , 161 ]. In this emerging model, trials evolve from drug-centric evaluations of average efficacy toward patient-oriented learning systems that continuously update hypotheses using real-time molecular dynamics and real-world evidence.

10. Future Perspective in Clinical Trials Design

This approach—AI, ctDNA, and RWE—is expected to fundamentally reshape hematologic clinical trial methodology in the coming years. Recent work also emphasizes that RWE still struggles at some stages with data quality and interpretability, but AI has the capacity to significantly improve RWD utilization, draw causal inferences, and connect the two approaches to routine clinical practice and trial evidence accumulation [ 155 , 166 ]. Using AI mathematical modeling, we are capable of simulating product performance, predicting the interaction of product attributes with patient physiology and clinical outcomes, tailoring treatments, and advancing personalized medicine. This approach will increasingly support predictive modeling, reinforcement learning and Bayesian optimization will drive adaptive trial designs that dynamically modify enrollment criteria [ 167 , 168 , 169 , 170 ].

The lack of benefit with intrathecal or high-dose methotrexate prophylaxis for prevention of CNS relapse in DLBCL among high-risk patients has not been confirmed in large retrospective studies, but it may be more targeted to treat occult disease in patients with detectable ctDNA at baseline [ 171 ]. In the future, ctDNA detection technologies are expected to be increasingly integrated into new therapeutic strategies to dynamically monitor changes in tumor burden, disease progression, and prognosis [ 172 ]. Classic clinical variables such as age, serum lactate dehydrogenase (LDH), Eastern Cooperative Oncology Group (ECOG), performance status, and extranodal involvement retain robust and independent prognostic significance within AI-based models regardless of model complexity or data integration strategy [ 173 ]. Given the high cost of novel drug development, RWD are anticipated to play an increasingly central role in regulatory decision-making and value assessment [ 5 ]. One of the most promising developments is the emergence of “self-learning trials,” in which incoming clinical, molecular, and real-world data streams are iteratively incorporated to refine trial design, patient selection, and therapeutic interventions in near real-time. Such systems may reduce development timelines, improve endpoint sensitivity through ctDNA-based surrogate markers, and enhance regulatory decision-making through the integration of high-quality real-world data. However, realizing this vision requires overcoming key challenges, including data standardization, cross-platform interoperability, regulatory acceptance of adaptive AI systems, and ethical considerations related to data privacy and algorithmic bias. Addressing these challenges will be essential to fully harness the synergistic potential of these technologies and to translate innovation into clinical benefit.

Over the coming years, integrating AI, ctDNA and RWE into routine clinical practice will require not only technological refinement but also resolution of regulatory, clinical, and economic challenges. Trial frameworks are expected to leverage AI-driven algorithms for real-time treatment modifications based on integrated data streams, supporting more adaptive and personalized therapeutic strategies. AI-guided interpretation of ctDNA kinetics will allow earlier and better-informed identification of molecular relapse, with preliminary implications for adaptive treatment approaches and a reduction in sample size required to show benefit at the clinical level.

11. Conclusions

The integration of AI, ctDNA, and RWE is reshaping the design and conduct of hematologic clinical trials. Each of these components contributes distinct value: AI enables advanced data analysis and patient stratification, ctDNA provides real-time insights into disease dynamics, and RWE extends evidence generation to broader and more representative patient populations. The integration of AI, ctDNA, and RWE represents a coherent and evidence-based advancement of hematologic clinical trial design. Across hematologic malignancies, AI enables improved patient stratification, endpoint selection, and data integration from complex clinical and molecular sources, improving trial efficiency and interpretability while preserving methodological rigor [ 1 , 3 , 5 ]. ctDNA provides a validated, minimally invasive biomarker for molecular residual disease assessment, treatment response, monitoring, and early relapse detection, supporting adaptive and biomarker-driven trial frameworks with clinically meaningful surrogate endpoints [ 6 , 7 , 8 , 49 , 75 , 76 , 77 , 78 ]. Concomitantly, RWE complements randomized evidence by extending evaluation to broader, more representative patient populations, facilitating external control arms, post-marketing surveillance, and long-term safety assessment, particularly in rare or high-risk hematologic settings [ 85 , 96 , 99 , 100 , 101 , 102 , 103 ].

When combined, these approaches support more adaptive, efficient, and patient-centered clinical trial frameworks. This integrated model allows for improved trial design, earlier detection of treatment response or failure, and better alignment between experimental research and real-world clinical practice without compromising regulatory or ethical standards [ 3 , 10 , 119 , 120 , 121 ]. Despite these advances, several challenges remain, including data standardization, validation of novel endpoints such as ctDNA, management of bias in AI models, and regulatory and ethical considerations related to data use. Addressing these issues will be essential to ensure the safe and effective implementation of these technologies. Therefore, the successful implementation of this integrated approach depends on robust data quality, standardization, transparency, and regulatory oversight, as well as careful attention to privacy, bias, and equity considerations [ 107 , 108 , 110 , 117 ]. Overall, the convergence of AI, ctDNA, and RWE represents a promising step toward more precise, flexible, and clinically relevant research in hematologic malignancies. Continued efforts to integrate and refine these approaches will be critical to translating innovation into improved patient outcomes [ 5 , 15 , 119 ].

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Author Contributions

A.M.M. Conceptualization, Methodology, Investigation, Writing—Original Draft Preparation, Writing—Review and Editing, Supervision, Project Administration. J.P.T.: Writing—Original Draft Preparation, Writing—Review and Editing. S.R.Z.: Writing—Original Draft Preparation, Writing—Review and Editing. A.A.K.: Validation, Review and Editing, Visualization. C.D.: Conceptualization, Validation, Resources, Writing—Review and Editing, Visualization. All authors have read and agreed to the published version of the manuscript.

Data Availability Statement

No new data were created or analyzed in this study.

Conflicts of Interest

The authors declare no conflicts of interest.

References

1. Bryant A.K. Zamora-Resendiz R. Dai X. Morrow D. Lin Y. Jungles K.M. Rae J.M. Tate A. Pearson A.N. Jiang R. Artificial intelligence to unlock real-world evidence in clinical oncology: A primer on recent advances Cancer Med. 2024 13 e7253 10.1002/cam4.7253 38899720 PMC11187737

2. Alhamrani S.Q. Ball G.R. El-Sherif A.A. Ahmed S. Mousa N.O. Alghorayed S.A. Alatawi N.A. Ali A.M. Alqahtani F.A. Gabre R.M. Machine Learning for Multi-Omics Characterization of Blood Cancers: A Systematic Review Cells 2025 14 1385 10.3390/cells14171385 40940796 PMC12427946

3. Fountzilas E. Tsimberidou A.M. Vo H.H. Kurzrock R. Clinical trial design in the era of precision medicine Genome Med. 2022 14 101 10.1186/s13073-022-01102-1 36045401 PMC9428375

4. Chang T.-G. Park S. Schäffer A.A. Jiang P. Ruppin E. Hallmarks of artificial intelligence contributions to precision oncology Nat. Cancer 2025 6 417 431 10.1038/s43018-025-00917-2 40055572 PMC11957836

5. Passamonti F. Corrao G. Castellani G. Mora B. Maggioni G. Gale R.P. Della Porta M.G. The future of research in hematology: Integration of conventional studies with real-world data and artificial intelligence Blood Rev. 2022 54 100914 10.1016/j.blre.2021.100914 34996639

6. Thalambedu N. Balla M. Sivasubramanian B.P. Sadaram P. Malla K.P. Vasipalli K.P. Kakadia S. Integrating artificial intelligence with circulating tumor DNA for non-small cell lung cancer: Opportunities, challenges, and future directions Front. Med. 2025 12 1612376 10.3389/fmed.2025.1612376 PMC12187728 40568188

7. Chen X. Mozgunov P. Baird R.D. Jaki T. Using circulating tumor DNA as a novel biomarker of efficacy for dose-finding designs in oncology Stat. Methods Med. Res. 2025 34 1665 1683 10.1177/09622802251350457 40590061 PMC12365363

8. Taniguchi H. Nakamura Y. Kotani D. Yukami H. Mishima S. Sawada K. Shirasu H. Ebi H. Yamanaka T. Aleshin A. CIRCULATE-Japan: Circulating tumor DNA–guided adaptive platform trials to refine adjuvant therapy for colorectal cancer Cancer Sci. 2021 112 2915 2920 10.1111/cas.14926 33931919 PMC8253296

9. Fountzilas E. Pearce T. Baysal M.A. Chakraborty A. Tsimberidou A.M. Convergence of evolving artificial intelligence and machine learning techniques in precision oncology npj Digit. Med. 2025 8 75 10.1038/s41746-025-01471-y 39890986 PMC11785769

10. Zhu M. Sridhar S. Hollingsworth R. Chit A. Kimball T. Murmello K. Greenberg M. Gurunathan S. Chen J. Hybrid clinical trials to generate real-world evidence: Design considerations from a sponsor’s perspective Contemp. Clin. Trials 2020 94 105856 10.1016/j.cct.2019.105856 31669449

11. Anderer A. Bastani H. Silberholz J. Adaptive Clinical Trial Designs with Surrogates: When Should We Bother? Manag. Sci. 2022 68 1982 2002 10.1287/mnsc.2021.4096

12. Tian F. Liu M. Zhao Y. Lin J. Liu R. BEAM: Bayesian Hybrid Design With Adaptive Sample Size Through Multisource Exchangeability Modeling Stat. Med. 2025 44 e70137 10.1002/sim.70137 40474447

13. Dermawan D. Alotaiq N. From Lab to Clinic: How Artificial Intelligence (AI) Is Reshaping Drug Discovery Timelines and Industry Outcomes Pharmaceuticals 2025 18 981 10.3390/ph18070981 40732273 PMC12298131

14. Bothwell L.E. Avorn J. Khan N.F. Kesselheim A.S. Adaptive design clinical trials: A review of the literature and ClinicalTrials.gov BMJ Open 2018 8 e018320 10.1136/bmjopen-2017-018320 PMC5829673 29440155

15. Nikanjam M. Kato S. Allen T. Sicklick J.K. Kurzrock R. Novel clinical trial designs emerging from the molecular reclassification of cancer CA Cancer J. Clin. 2025 75 243 267 10.3322/caac.21880 39841128 PMC12061631

16. Kaye D. Isidori A. Current Challenges in Hematology: Awareness, Prevention, Equity Front Oncol. 2021 11 653020 10.3389/fonc.2021.653020 33987089 PMC8111442

17. Aiyegbusi O.L. Cruz Rivera S. Kamudoni P. Anderson N. Collis P. Denniston A.K. Harding R. Hughes S.E. Khunti K. Kotecha D. Recommendations to promote equity, diversity and inclusion in decentralized clinical trials Nat. Med. 2024 30 3075 3084 10.1038/s41591-024-03323-w 39472759

18. Hantel A. Blackstone E.C. Revette A.C. E Roberts J. Mims A. DeAngelo D.J. Lathan C.S. A Abel G. Barriers and facilitators to representative acute leukemia trial participation at National Cancer Institute-designated comprehensive cancer centers Oncologist 2025 30 oyaf241 10.1093/oncolo/oyaf241 40795379 PMC12445654

19. Wayant C. Scheckel C. Hicks C. Nissen T. Leduc L. Som M. Vassar M. Evidence of selective reporting bias in hematology journals: A systematic review PLoS ONE 2017 12 e0178379 10.1371/journal.pone.0178379 28570573 PMC5453439

20. Zhao X. Xiong J. Li D. Zhang Y. Clinical trials of nanoparticle-enhanced CAR-T and NK cell therapies in oncology: Overcoming translational and clinical challenges—A mini review Front. Med. 2025 12 1655693 10.3389/fmed.2025.1655693 PMC12403998 40904365

21. Sainatham C. Yadav D. Babu A.D. Tallapalli J.R. Kanagala S.G. Filippov E. Chavez F.M. Ahmed N. Lutfi F. The current socioeconomic and regulatory landscape of immune effector cell therapies Front. Med. 2024 11 1462307 10.3389/fmed.2024.1462307 PMC11652178 39697210

22. Ciccocioppo R. Comoli P. Astori G. del Bufalo F. Prapa M. Dominici M. Locatelli F. Developing cell therapies as drug products Br. J. Pharmacol. 2020 178 262 279 10.1111/bph.15305 33140850

23. Pan K. Farrukh H. Chittepu V.C.S.R. Xu H. Pan C.-X. Zhu Z. CAR race to cancer immunotherapy: From CAR T, CAR NK to CAR macrophage therapy J. Exp. Clin. Cancer Res. 2022 41 119 10.1186/s13046-022-02327-z 35361234 PMC8969382

24. Jean-Louis G. Seixas A.A. The value of decentralized clinical trials: Inclusion, accessibility, and innovation Science 2024 385 eadq4994 10.1126/science.adq4994 39172847

25. Tan R.K.J. Wu D. Day S. Zhao Y. Larson H.J. Sylvia S. Tang W. Tucker J.D. Digital approaches to enhancing community engagement in clinical trials npj Digit. Med. 2022 5 37 10.1038/s41746-022-00581-1 35338241 PMC8956701

26. Rogers A. De Paoli G. Subbarayan S. Copland R. Harwood K. Coyle J. Mitchell L. MacDonald T.M. Mackenzie I.S. Consortium T. A systematic review of methods used to conduct decentralised clinical trials Br. J. Clin. Pharmacol. 2022 88 2843 2862 10.1111/bcp.15205 34961991 PMC9306873

27. Singh P. Burden A.M. Natanegara F. Beckman R.A. Design and Execution of Sustainable Decentralized Clinical Trials Clin. Pharmacol. Ther. 2023 114 802 809 10.1002/cpt.3009 37489911

28. Biswas A. Dalal J. Decentralized clinical trials-balancing promise and pitfalls in modern research: A systematic review Int. J. Clin. Trials 2025 12 224 232 10.18203/2349-3259.ijct20252201

29. Oualikene-Gonin W. Jaulent M.C. Thierry J.P. Oliveira-Martins S. Belgodère L. Maison P. Ankri J. Ansm T.S.A.B.O. Artificial intelligence integration in the drug lifecycle and in regulatory science: Policy implications, challenges and opportunities Front. Pharmacol. 2024 15 1437167 10.3389/fphar.2024.1437167 39156111 PMC11327028

30. Tubbs A. Vazquez E.A. Digital twins in increasing diversity in clinical trials: A systematic review J. Biomed. Inform. 2025 169 104879 10.1016/j.jbi.2025.104879 40784603

31. Mangalam M. AI-driven dynamic grouping for adaptive clinical trials: Rethinking randomization in precision medicine Artif. Intell. Med. 2025 170 103272 10.1016/j.artmed.2025.103272 40975956

32. Zhang G. Bai M. Du H. Yuan Y. Wang Y. Fan W. Zhu H. Wu D. He P. Xue B. Current Advances and Challenges in CAR-T Therapy for Hematological and Solid Tumors ImmunoTargets Ther. 2025 14 655 680 10.2147/ITT.S519616 40599347 PMC12212355

33. Peccatori N. Brivio E. Lissat A. Sirvent F.B. Salzer E. Biondi A. Fazio G. Rizzari C. Tasian S.K. Zwaan C.M. Molecularly Targeted Small Molecule Inhibitor Therapy for Pediatric Acute Lymphoblastic Leukemia: A Comprehensive Review of Clinical Trials Cancers 2025 17 3322 10.3390/cancers17203322 41154380 PMC12562350

34. Lamers-Kok N. Panella D. Georgoudaki A.M. Liu H. Özkazanc D. Kučerová L. Duru A.D. Spanholtz J. Raimo M. Natural killer cells in clinical development as non-engineered, engineered, and combination therapies J. Hematol. Oncol. 2022 15 164 10.1186/s13045-022-01382-5 36348457 PMC9644572

35. Wang X. Xi H. Geng X. Li Y. Zhao M. Li F. Li Z. Ji H. Tian H. Artificial Intelligence-Based Prediction of Lower Extremity Deep Vein Thrombosis Risk After Knee/Hip Arthroplasty Clin. Appl. Thromb./Hemost. 2023 29 10760296221139263 10.1177/10760296221139263 36596268 PMC9830569

36. Ding R. Ding Y. Zheng D. Huang X. Dai J. Jia H. Deng M. Yuan H. Zhang Y. Fu H. Machine Learning-Based Screening of Risk Factors and Prediction of Deep Vein Thrombosis and Pulmonary Embolism After Hip Arthroplasty Clin. Appl. Thromb./Hemost. 2023 29 10760296231186145 10.1177/10760296231186145 37394825 PMC10328010

37. Didi I. Alliot J.M. Dumas P.Y. Vergez F. Tavitian S. Largeaud L. Bidet A. Rieu J.-B. Luquet I. Lechevalier N. Artificial intelligence-based prediction models for acute myeloid leukemia using real-life data: A DATAML registry study Leuk. Res. 2024 136 107437 10.1016/j.leukres.2024.107437 38215555

38. Kilic Gunes E. Artificial intelligence in hematology: Current trends and application areas Ann. Hematol. 2025 104 6033 6043 10.1007/s00277-025-06706-2 41177826 PMC12672734

39. Bouziana S. Hematological Malignancies: Molecular Mechanisms and Therapy Int. J. Mol. Sci. 2025 26 2438 10.3390/ijms26062438 40141082 PMC11941954

40. Khoury R. Raffoul C. Khater C. Hanna C. Precision Medicine in Hematologic Malignancies: Evolving Concepts and Clinical Applications Biomedicines 2025 13 1654 10.3390/biomedicines13071654 40722725 PMC12292076

41. Mattioda C. Voena C. Ciardelli G. Mattu C. In Vitro 3D Models of Haematological Malignancies: Current Trends and the Road Ahead? Cells 2025 14 38 10.3390/cells14010038 39791739 PMC11720277

42. Guevara-Ramírez P. Cadena-Ullauri S. Paz-Cruz E. Ruiz-Pozo V.A. Tamayo-Trujillo R. Cabrera-Andrade A. Zambrano A.K. Gut Microbiota Disruption in Hematologic Cancer Therapy: Molecular Insights and Implications for Treatment Efficacy Int. J. Mol. Sci. 2024 25 10255 10.3390/ijms251910255 39408584 PMC11476909

43. Auberger P. Tamburini-Bonnefoy J. Puissant A. Drug resistance in hematological malignancies Int. J. Mol. Sci. 2020 21 6091 10.3390/ijms21176091 32847013 PMC7503602

44. Cordoba R. Eyre T.A. Klepin H.D. Wildes T.M. Goede V. A comprehensive approach to therapy of haematological malignancies in older patients Lancet Haematol. 2021 8 e840 e852 10.1016/S2352-3026(21)00241-6 34624238

45. Bogdanovic B. Hugonnet F. Montemagno C. Theranostics in Hematological Malignancies: Cutting-Edge Advances in Diagnosis and Targeted Therapy Cancers 2025 17 1247 10.3390/cancers17071247 40227793 PMC11987953

46. Sun D. Chen W. He J. He Y. Jiang H. Jiang H. Liu D. Li L. Liu M. Mao Z. A novel method for screening malignant hematological diseases by constructing an optimal machine learning model based on blood cell parameters BMC Med. Inform. Decis. Mak. 2025 25 72 10.1186/s12911-025-02892-1 39934810 PMC11816569

47. Cai Q. Ye B. Zheng W. Zhang S. Zhang J. Shen Y. Yao D. Zhang H. Huang Z. Hu J. Development of a screening model for APL using cell population data and deep learning-extracted WBC scattergram features BMC Cancer 2025 25 1725 10.1186/s12885-025-15034-7 41204132 PMC12593920

48. Nazha A. Elemento O. Ahuja S. Lam B. Miles M. Shouval R. McWeeney S. Sirhan S. Srisuwananukorn A. Haferlach T. Artificial intelligence in hematology Blood 2025 146 2283 2292 10.1182/blood.2025029876 40845137

49. Bahyat K. Ammour A. Cheddadi M. Fedorova V. Houari M. The Role of Artificial Intelligence in the Hematology Department Sch. J. Med. Case Rep. 2025 13 1972 1975 10.36347/sjmcr.2025.v13i09.005

50. Cuocolo R. Caruso M. Perillo T. Ugga L. Petretta M. Machine Learning in oncology: A clinical appraisal Cancer Lett. 2020 481 55 62 10.1016/j.canlet.2020.03.032 32251707

51. Thirugnanasambandam R.P. Bauer A. D’Angelo C. The Role of Artificial Intelligence in Palliative Oncology: Zeroing in on Hematologic Malignancies Oncology 2025 39 468 478 10.46883/2025.25921057 41370240

52. Li Q. Lin N. Wang Z. Chen Y. Xie Y. Wang X. Tang J. Xu Y. Xu M. Lu N. Machine learning-based prognostic model for bloodstream infections in hematological malignancies using Th1/Th2 cytokines BMC Infect. Dis. 2025 25 415 10.1186/s12879-025-10808-7 40140749 PMC11948653

53. Carreras J. Nakamura N. Artificial Intelligence, Lymphoid Neoplasms, and Prediction of MYC, BCL2, and BCL6 Gene Expression Using a Pan-Cancer Panel in Diffuse Large B-Cell Lymphoma Hemato 2024 5 119 143 10.3390/hemato5020011

54. Uppal A. Kakkar B. Johri P. Kumar Y. Koul A. Enhancing lymphoma cancer detection using deep transfer learning on histopathological images Sci. Rep. 2025 15 38042 10.1038/s41598-025-21888-5 41168354 PMC12575721

55. Carreras J. Roncador G. Hamoudi R. Artificial Intelligence Predicted Overall Survival and Classified Mature B-Cell Neoplasms Based on Immuno-Oncology and Immune Checkpoint Panels Cancers 2022 14 5318 10.3390/cancers14215318 36358737 PMC9657332

56. Eweje F.R. Bao B. Wu J. Dalal D. Liao W.-H. He Y. Luo Y. Lu S. Zhang P. Peng X. Deep Learning for Classification of Bone Lesions on Routine MRI eBioMedicine 2021 68 103402 10.1016/j.ebiom.2021.103402 34098339 PMC8190437

57. Han R. Acosta J.N. Shakeri Z. A Ioannidis J.P. Topol E.J. Rajpurkar P. Randomised controlled trials evaluating artificial intelligence in clinical practice: A scoping review Lancet Digit. Health 2024 6 e367 e373 10.1016/S2589-7500(24)00047-5 38670745 PMC11068159

58. Lam T.Y. Cheung M.F. Munro Y.L. Lim K.M. Shung D. Sung J.J.Y. Randomized controlled trials of artificial intelligence in clinical practice: A systematic review J. Med. Internet Res. 2022 24 e37188 10.2196/37188 35904087 PMC9459941

59. Zhou Q. Chen Z. Cao Y. Peng S. Clinical impact and quality of randomized controlled trials involving interventions evaluating artificial intelligence prediction tools: A systematic review npj Digit. Med. 2021 4 154 10.1038/s41746-021-00524-2 34711955 PMC8553754

60. Makutam V. Yashashwini Achanti S. Doostan M. Integration of Artificial Intelligence in Adaptive Trial Designs: Enhancing Efficiency and Patient-Centric Outcomes Int. J. Adv. Res. 2024 12 205 215 10.21474/IJAR01/19245

61. Olawade D.B. Fidelis S.C. Marinze S. Egbon E. Osunmakinde A. Osborne A. Artificial intelligence in clinical trials: A comprehensive review of opportunities, challenges, and future directions Int. J. Med. Informatics 2025 206 106141 10.1016/j.ijmedinf.2025.106141 41075423

62. Chopra H. Annu Shin D.K. Munjal K. Priyanka Dhama K. Emran T.B. Revolutionizing clinical trials: The role of AI in accelerating medical breakthroughs Int. J. Surg. 2023 109 4211 4220 10.1097/JS9.0000000000000705 38259001 PMC10720846

63. Park Y. Jackson G.P. Foreman M.A. Gruen D. Hu J. Das A.K. Evaluating artificial intelligence in medicine: Phases of clinical research JAMIA Open 2020 3 326 331 10.1093/jamiaopen/ooaa033 33215066 PMC7660958

64. Romero M. Mosquera Orgueira A. Mejía Saldarriaga M. How artificial intelligence revolutionizes the world of multiple myeloma Front. Hematol. 2024 3 1331109 10.3389/frhem.2024.1331109

65. Allegra A. Tonacci A. Sciaccotta R. Genovese S. Musolino C. Pioggia G. Gangemi S. Machine Learning and Deep Learning Applications in Multiple Myeloma Diagnosis, Prognosis, and Treatment Selection Cancers 2022 14 606 10.3390/cancers14030606 35158874 PMC8833500

66. Li J.Y. Zuo L.P. Xu J. Sun C.-Y. Clinical applications of circulating tumor DNA in hematological malignancies: From past to the future Blood Rev. 2024 68 101237 10.1016/j.blre.2024.101237 39261219

67. Duffles G. Maués JHda S. Lupinacci F. Pereira L.G. Ferreira E.N. Freitas L. Niemann F. Takahashi M.E.S. Ramos C.D. Chauffaille M.d.L.L.F. Circulating tumor DNA in diffuse large B-cell lymphoma: Analysis of response assessment, correlation with PET/CT and clone evolution Hematol. Transfus. Cell Ther. 2024 46 S241 S249 10.1016/j.htct.2024.07.005 39317576 PMC11726095

68. Stroun M. Lyautey J. Lederrey C. Olson-Sand A. Anker P. About the possible origin and mechanism of circulating DNA: Apoptosis and active DNA release Clin. Chim. Acta 2001 313 139 142 10.1016/S0009-8981(01)00665-9 11694251

69. Bettegowda C. Sausen M. Leary R.J. Kinde I. Wang Y. Agrawal N. Bartlett B.R. Wang H. Luber B. Alani R.M. Detection of circulating tumor DNA in early- and late-stage human malignancies Sci. Transl. Med. 2014 6 224 10.1126/scitranslmed.3007094 PMC4017867 24553385

70. Razavi P. Li B.T. Brown D.N. Jung B. Hubbell E. Shen R. Abida W. Juluru K. De Bruijn I. Hou C. High-intensity sequencing reveals the sources of plasma circulating cell-free DNA variants Nat. Med. 2019 25 1928 1937 10.1038/s41591-019-0652-7 31768066 PMC7061455

71. Ouyang Z. Zeng R. Wang S. Wu X. Li Y. He Y. Wang C. Xia C. Ou Q. Bao H. Genomic signatures in plasma circulating tumor DNA reveal treatment response and prognostic insights in mantel cell lymphoma Cancer Cell Int. 2025 25 172 10.1186/s12935-025-03789-9 40319323 PMC12049778

72. Haider Z. Wästerlid T. Spångberg L.D. Rabbani L. Jylhä C. Thorvaldsdottir B. Skaftason A. Awier H.N. Krstic A. Gellerbring A. Whole-genome informed circulating tumor DNA analysis by multiplex digital PCR for disease monitoring in B-cell lymphomas: A proof-of-concept study Front. Oncol. 2023 13 1176698 10.3389/fonc.2023.1176698 37333831 PMC10272573

73. Woof V.G. Lee R.J. Lorigan P. French D.P. Circulating tumour DNA monitoring and early treatment for relapse: Views from patients with early-stage melanoma Br. J. Cancer 2022 126 1450 1456 10.1038/s41416-022-01766-x 35301436 PMC8927744

74. Pessoa L.S. Heringer M. Ferrer V.P. ctDNA as a cancer biomarker: A broad overview Crit. Rev. Oncol. 2020 155 103109 10.1016/j.critrevonc.2020.103109 33049662

75. Meriranta L. Rask Kragh Jørgensen R. Pasanen A. Kolstad A. Hutchings M. Niemann C.U. El Galaly T.C. Riise J. Christensen J.H. Sonnevi K. Circulating tumor DNA determinants of response and outcome in relapsed/refractory mantle cell lymphoma Blood Adv. 2025 9 4528 4541 10.1182/bloodadvances.2025015791 40402132 PMC12447437

76. Meriranta L. Alkodsi A. Pasanen A. Lepistö M. Mapar P. Blaker Y.N. Jørgensen J.M. Karjalainen-Lindsberg M.-L. Fiskvik I. Mikalsen L.T.G. Molecular features encoded in the ctDNA reveal heterogeneity and predict outcome in high-risk aggressive B-cell lymphoma Blood 2022 139 1863 1877 10.1182/blood.2021012852 34932792

77. Nikanjam M. Kato S. Kurzrock R. Liquid biopsy: Current technology and clinical applications J. Hematol. Oncol. 2022 15 131 10.1186/s13045-022-01351-y 36096847 PMC9465933

78. Adashek J.J. Janku F. Kurzrock R. Signed in blood: Circulating tumor dna in cancer diagnosis, treatment and screening Cancers 2021 13 3600 10.3390/cancers13143600 34298813 PMC8306582

79. Tan X. Yan H. Chen L. Zhang Y. Sun C. Clinical Value of ctDNA in Hematological Malignancies (Lymphomas, Multiple Myeloma, Myelodysplastic Syndrome, and Leukemia): A Meta-Analysis Front. Oncol. 2021 11 632910 10.3389/fonc.2021.632910 33747954 PMC7970179

80. Wan J.C.M. Massie C. Garcia-Corbacho J. Mouliere F. Brenton J.D. Caldas C. Pacey S. Baird R. Rosenfeld N. Liquid biopsies come of age: Towards implementation of circulating tumour DNA Nat. Rev. Cancer 2017 17 223 238 10.1038/nrc.2017.7 28233803

81. Al-Yafie Shikaha Ahmed Y. Shyn L.W. Xiang P.B.J. Dhanusha R. Michael T.R. Yap W.H. Goh B.H. Wong R.S. Liquid Biopsy in Hematologic Malignancies: Advances, Challenges, and Future Directions Asia-Pacific J. Clin. Oncol. 2025 online ahead of print 10.1111/ajco.70038 41165026 PMC13568961

82. Sharma P. Singh S. Gopi P. Rani M.S.S. Singh P. Pandya P. Ali M.S. Evaluating the carcinogenic potential of trazodone hydrochloride via duplex DNA targeting: Molecular interaction, binding mechanism and affinity assessment via structural calculations J. Pharm. Biomed. Anal. 2025 263 116881 10.1016/j.jpba.2025.116881 40300316

83. Ramkissoon L.A. Montgomery N.D. Applications of next-generation sequencing in hematologic malignancies Hum. Immunol. 2021 82 859 870 10.1016/j.humimm.2021.02.006 33648805

84. Siravegna G. Marsoni S. Siena S. Bardelli A. Integrating liquid biopsies into the management of cancer Nat. Rev. Clin. Oncol. 2017 14 531 548 10.1038/nrclinonc.2017.14 28252003

85. Roschewski M. Kurtz D.M. Westin J.R. Lynch R.C. Gopal A.K. Alig S.K. Sworder B.J. Cherng H.-J.J. Kuffer C. Blair D. Remission Assessment by Circulating Tumor DNA in Large B-cell Lymphoma J. Clin. Oncol. 2025 43 3652 3661 10.1200/JCO-25-01534 40802906 PMC12363663

86. Landgren O. Prior T.J. Masterson T. Heuck C. Bueno O.F. Dash A.B. Einsele H. Goldschmidt H. Knop S. Li C. EVIDENCE Meta-Analysis: Evaluating Minimal Residual Disease as an Intermediate Clinical End Point for Multiple Myeloma Blood 2024 144 359 367 10.1182/blood.2024024371 38768337 PMC11418064

87. Landgren O. Devlin S.M. Minimal Residual Disease as an Early Endpoint for Accelerated Drug Approval in Myeloma: A Roadmap Blood Cancer Discov. 2024 6 13 22 10.1158/2643-3230.BCD-24-0292 PMC11707509 39630969

88. Avet-Loiseau H. Ludwig H. Landgren O. Paiva B. Morris C. Yang H. Zhou K. Ro S. Mateos M.-V. Minimal Residual Disease Status as a Surrogate Endpoint for Progression-free Survival in Newly Diagnosed Multiple Myeloma Studies: A Meta-analysis Clin. Lymphoma Myeloma Leuk. 2019 20 e30 e37 10.1016/j.clml.2019.09.622 31780415 PMC7444731

89. Pirouzbakht M. Hamzeh S. Soleimani Samarkhazan H. Beyond single biomarkers: Multi-omics strategies to predict immunotherapy outcomes in blood cancers Clin. Exp. Med. 2025 25 355 10.1007/s10238-025-01902-w 41193778 PMC12589271

90. Food and Drug Administration Use of Circulating Tumor DNA for Early-Stage Solid Tumor Drug Development Guidance for Industry Draft Guidance 2022 Available online: https://www.regulations.gov/document/FDA-2022-D-0084-0002 (accessed on 6 January 2026)

91. Liang J.H. Hua W. Yin H. Li Y. Zhang X.-Y. Liang J.-H. Zhu L.-Q. Gao R. Wang C.-X. Shao Y. Clinical implications of ctDNA-based minimal residual disease detection in newly diagnosed peripheral T-cell lymphoma: A single-center cohort study Cell Commun. Signal. 2025 23 441 10.1186/s12964-025-02438-y 41094514 PMC12522508

92. Fardella E. Zanirato G. Magni M. Caldarelli N. Chiappella A. Dodero A. Ljevar S. Orsucci L. Re A. Usai S.V. Prognostic impact of pretreatment cell-free DNA concentration in newly diagnosed peripheral T-cell lymphomas Br. J. Haematol. 2024 204 1752 1756 10.1111/bjh.19280 38176400

93. Gao Y. He H. Li X. Zhang L. Xu W. Feng R. Li W. Xiao Y. Liu X. Chen Y. Sintilimab (anti-PD-1 antibody) plus chidamide (histone deacetylase inhibitor) in relapsed or refractory extranodal natural killer T-cell lymphoma (SCENT): A phase Ib/II study Signal Transduct. Target. Ther. 2024 9 121 10.1038/s41392-024-01825-0 38755119 PMC11099117

94. Kim S.J. Kim Y.J. Yoon S.E. Ryu K.J. Park B. Park D. Cho D. Kim H.-Y. Cho J. Ko Y.H. Circulating Tumor DNA–Based Genotyping and Monitoring for Predicting Disease Relapses of Patients with Peripheral T-Cell Lymphomas Cancer Res. Treat. 2023 55 291 303 10.4143/crt.2022.017 35240014 PMC9873338

95. Passamonti F. Corrao G. Castellani G. Mora B. Maggioni G. Della Porta M.G. Gale R.P. Using real-world evidence in haematology Best Pract. Res. Clin. Haematol. 2024 37 101536 10.1016/j.beha.2024.101536 38490764

96. Derman B. Tan C. Steinfield I. Wilson F.R. Lin D. Wu B. Fernandez M. Fowler J. Paner-Straseviciute A. Kim N. Real-World Evidence Evaluating Teclistamab in Patients with Relapsed/Refractory Multiple Myeloma: A Systematic Literature Review Cancers 2025 17 1235 10.3390/cancers17071235 40227780 PMC11988155

97. Malagola M. Ohgami R. Greco R. Editorial: Real-world data and real-world evidence in hematologic malignancies Front. Oncol. 2023 13 1232980 10.3389/fonc.2023.1232980 37361576 PMC10289287

98. Girman C.J. Ritchey M.E. Re V.L. Real-world data: Assessing electronic health records and medical claims data to support regulatory decision-making for drug and biological products Pharmacoepidemiol. Drug Saf. 2022 31 717 720 10.1002/pds.5444 35471704 PMC9320939

99. US Food and Drug Administration Real-World Evidence. U.S. Food and Drug Administration (US FDA) 16 May 2024 Available online: https://www.fda.gov/science-research/science-and-research-special-topics/real-world-evidence (accessed on 6 January 2026)

100. Giacco S. Del Canonica G.W. Agache I. Price D. Roche N. Schunemann H. Allan K. Ansotegui I. Barbaglia S. Bernstein J. Expert Perspectives on Next Generation Health Guidelines: How to Integrate RWE in EBM Pragmatic Obs. Res. 2025 16 169 177 10.2147/POR.S487342 PMC12436199 40964518

101. Finnegan H. Mountford N. 25 Years of Electronic Health Record Implementation Processes: Scoping Review J. Med. Internet Res. 2025 27 e60077 10.2196/60077 40053758 PMC11914847

102. Molloy M.J. Yan A.P. Wilson A.E. Beus J. Hess L.M. Using the Electronic Health Record to Facilitate Drug Allergy Delabeling Curr. Allergy Asthma Rep. 2025 25 49 10.1007/s11882-025-01229-2 41136878 PMC12552388

103. Nakano M. Kawaguchi M. Nakano D. Sano T. Amano K. Nakamura T. Takahashi H. Tarasawa K. Fujimori K. Kawaguchi T. Regional Difference of Liver Cancer in Japan: A Real-world Evidence Using the National Database of Health Insurance Claims Intern. Med. 2025 64 2263 2269 10.2169/internalmedicine.4820-24 39924234 PMC12393965

104. Choquet S. Marchal C. Deygas F. Deslandes M. Macher N. de Pouvourville G. Levy V. A retrospective observational study of ibrutinib in chronic lymphocytic leukaemia in a real-life setting in France using the national claims database (OSIRIS) Ann. Hematol. 2024 103 2969 2981 10.1007/s00277-024-05859-w 38965145 PMC11283383

105. Kinai E. Ono M. Oh A. Ota M. Myaguchi Y. Ueda H. Analysis of nationwide hemophilia care: A cohort study using two Japanese healthcare claims databases Health Sci. Rep. 2022 5 e498 10.1002/hsr2.498 35128076 PMC8795212

106. Taylor L. Chen A. Pierre A. Utilization of real-world evidence in regulatory approvals for multiple myeloma therapies Blood Cancer J. 2025 15 210 10.1038/s41408-025-01420-8 41309560 PMC12660783

107. Garcia-Guiñon A. Charry P.A. Jimenez M. Sarra J. Delgado I. de la Torre L.S. Santaliestra M. Garcia-Pintos M. Gonzalez Y. Senin A. Real-world evidence of Carfilzomib, Lenalidomide and Dexamethasone (KRD) Scheme in patients with relapsed/refractory multiple myeloma Ann. Hematol. 2025 104 1177 1186 10.1007/s00277-025-06240-1 39954075 PMC11971156

108. Zuber M. Elshafie S. Taj S. Zapata L.V. CO114 Treatment Patterns Post-CAR T Cell Therapy in Patients with Hematologic Malignancies: A Real-World US Study Value Health 2025 28 S43 10.1016/j.jval.2025.04.199

109. Lin L. Lucassen M.J.J. van der Noort V. Egberts T.C. Beijnen J.H. Huitema A.D. The feasibility of using real world data as external control arms in oncology trials Drug Discov. Today 2025 30 104324 10.1016/j.drudis.2025.104324 40054765

110. Ventz S. Lai A. Cloughesy T.F. Wen P.Y. Trippa L. Alexander B.M. Design and evaluation of an external control arm using prior clinical trials and real-world data Clin. Cancer Res. 2019 25 4993 5001 10.1158/1078-0432.CCR-19-0820 31175098 PMC6697596

111. Russek M. Peltner J. Haenisch B. Supplementing Single-Arm Trials with External Control Arms—Evaluation of German Real-World Data Clin. Pharmacol. Ther. 2025 118 1443 1450 10.1002/cpt.3684 40237254 PMC12641071

112. Gray C.M. Grimson F. Layton D. Pocock S. Kim J. A Framework for Methodological Choice and Evidence Assessment for Studies Using External Comparators from Real-World Data Drug Saf. 2020 43 623 633 10.1007/s40264-020-00944-1 32440847 PMC7305259

113. Liu J. Yao M. Wang M. Jie W. Liu Y. Luo X. Huan J. Deng K. Deng K. Zou K. Design, Conduct, and Analysis of Externally Controlled Trials JAMA Netw. Open 2025 8 e2530277 10.1001/jamanetworkopen.2025.30277 40906478 PMC12411980

114. Selukar S. Yin V. Othus M. Synthetic control arms and other uses of external data in clinical trials for hematological malignancies Blood Rev. 2025 74 101324 10.1016/j.blre.2025.101324 40753019

115. Li L. Jemielita T. Confounding adjustment in the analysis of augmented randomized controlled trial with hybrid control arm Stat. Med. 2023 42 2855 2872 10.1002/sim.9753 37186394

116. Lee Ventola C. Big Data and Pharmacovigilance: Data Mining for Adverse Drug Events and Interactions J. Formul. Manag. 2018 43 340 351 PMC5969211 29896033

117. Berlin J.A. Glasser S.C. Ellenberg S.S. Adverse event detection in drug development: Recommendations and obligations beyond phase 3 Am. J. Public Health 2008 98 1366 1371 10.2105/AJPH.2007.124537 18556607 PMC2446471

118. US Food and Drug Administration FDA Use of Real-World Evidence in Regulatory Decision Making 26 September 2025 Available online: https://www.fda.gov/science-research/real-world-evidence/fda-use-real-world-evidence-regulatory-decision-making (accessed on 6 January 2026)

119. Gerke S. Minssen T. Cohen G. Ethical and legal challenges of artificial intelligence-driven healthcare Artificial Intelligence in Healthcare Academic Press Cambridge, MA, USA 2020 295 336 10.1016/B978-0-12-818438-7.00012-5

120. Nazer L.H. Zatarah R. Waldrip S. Ke J.X.C. Moukheiber M. Khanna A.K. Hicklen R.S. Moukheiber L. Moukheiber D. Ma H. Bias in artificial intelligence algorithms and recommendations for mitigation PLoS Digit. Health 2023 2 e0000278 10.1371/journal.pdig.0000278 37347721 PMC10287014

121. Pasculli G. Virgolin M. Myles P. Vidovszky A. Fisher C. Biasin E. Mourby M. Pappalardo F. D’AMico S. Torchia M. Synthetic Data in Healthcare and Drug Development: Definitions, Regulatory Frameworks, Issues CPT Pharmacometrics Syst. Pharmacol. 2025 14 840 852 10.1002/psp4.70021 40193292 PMC12072219

122. Vellanki P.J. Ghosh S. Pathak A. Fusco M.J. Bloomquist E.W. Tang S. Singh H. Philip R. Pazdur R. Beaver J.A. Regulatory implications of ctDNA in immuno-oncology for solid tumors J. Immunother. Cancer 2023 11 e005344 10.1136/jitc-2022-005344 36796877 PMC9936292

123. U.S. Department of Health and Human Services Food and Drug Administration Use of Circulating Tumor Deoxyribonucleic Acid for Early-Stage Solid Tumor Drug Development; Guidance for Industry; Availability. Preprint November 2024 Available online: https://www.fda.gov/regulatory-information/search-fda-guidance-documents/use-circulating-tumor-deoxyribonucleic-acid-early-stage-solid-tumor-drug-development-guidance (accessed on 16 February 2026)

124. van der Leest P. Schuuring E. Critical Factors in the Analytical Work Flow of Circulating Tumor DNA-Based Molecular Profiling Clin. Chem. 2024 70 220 233 10.1093/clinchem/hvad194 38175597

125. Lockwood C.M. Borsu L. Cankovic M. Earle J.S. Gocke C.D. Hameed M. Jordan D. Lopategui J.R. Pullambhatla M. Reuther J. Recommendations for Cell-Free DNA Assay Validations: A Joint Consensus Recommendation of the Association for Molecular Pathology and College of American Pathologists J. Mol. Diagn. 2023 25 876 897 10.1016/j.jmoldx.2023.09.004 37806433

126. Li W. Huang X. Patel R. Schleifman E. Fu S. Shames D.S. Zhang J. Analytical evaluation of circulating tumor DNA sequencing assays Sci. Rep. 2024 14 4973 10.1038/s41598-024-54361-w 38424110 PMC10904763

127. Ro S.K. Zhang W. Jiang Q. Li X.N. Liu R. Lu C.C. Marchenko O. Sun L. Zhao J. Statistical Considerations on the Use of RWD/RWE for Oncology Drug Approvals: Overview and Lessons Learned Ther. Innov. Regul. Sci. 2023 57 899 910 10.1007/s43441-023-00528-y 37179264 PMC10276785

128. US Department of Health and Human Services Health Information Privacy. n.d Available online: https://www.hhs.gov/hipaa/for-professionals/special-topics/reproductive-health/index.html (accessed on 6 January 2026)

129. Jose J.H. Hendricks-Sturrup R. A HIPAA “Expert Pathway” Method to Achieve More Efficient Access to Protected Healthcare Information. MIT Computational Law Report 2021 Available online: https://law.mit.edu/pub/ahipaaexpertpathway (accessed on 6 January 2026)

130. EU General Data Protection Regulation (GDPR) (EU) 2016/679 18 May 2018 Available online: https://gdpr-info.eu/ (accessed on 6 January 2026)

131. Jaffe D.H. Malin B.A. Hendricks-Sturrup R.M. A Real-World Data Challenge: Guidance for Aligning Data Privacy Compliance and Fit-for-Purpose Usability Health Aff. Sch. 2025 3 qxaf210 10.1093/haschl/qxaf210 41323413 PMC12661526

132. Chinta S.V. Wang Z. Palikhe A. Zhang X. Kashif A. Smith M.A. Liu J. Zhang W. AI-Driven Healthcare: A Review on Ensuring Fairness and Mitigating Bias PLoS Digit. Health 2025 4 e0000864 10.1371/journal.pdig.0000864 40392801 PMC12091740

133. Kelkar A.H. Hantel A. Koranteng E. Cutler C.S. Hammer M.J. Abel G.A. Digital Health to Patient-Facing Artificial Intelligence: Ethical Implications and Threats to Dignity for Patients With Cancer JCO Oncol. Pract. 2024 20 314 317 10.1200/OP.23.00412 37922435

134. Naik N. Hameed B.M.Z. Shetty D.K. Swain D. Shah M. Paul R. Aggarwal K. Ibrahim S. Patil V. Smriti K. Legal and Ethical Consideration in Artificial Intelligence in Healthcare: Who Takes Responsibility? Front. Surg. 2022 9 862322 10.3389/fsurg.2022.862322 35360424 PMC8963864

135. Jeyaraman M. Balaji S. Jeyaraman N. Yadav S. Unraveling the Ethical Enigma: Artificial Intelligence in Healthcare Cureus 2023 15 e43262 10.7759/cureus.43262 37692617 PMC10492220

136. Bonomi L. Huang Y. Ohno-Machado L. Privacy challenges and research opportunities for genomic data sharing Nat. Genet. 2020 52 646 654 10.1038/s41588-020-0651-0 32601475 PMC7761157

137. Horgan D. Pesapane F. Van der Buckle M. de Maria R. Dube F. Singh J. Ługowska I. Bayle A. Hofman P. Malapelle U. From theory to practice: Implementing next-generation sequencing and public health genomics in healthcare systems Crit. Rev. Oncol. 2024 201 104433 10.1016/j.critrevonc.2024.104433 38955310

138. Horgan D. Čufer T. Gatto F. Lugowska I. Verbanac D. Carvalho Â. Lal J.A. Kozaric M. Toomey S. Ivanov H.Y. Accelerating the Development and Validation of Liquid Biopsy for Early Cancer Screening and Treatment Tailoring Healthcare 2022 10 1714 10.3390/healthcare10091714 36141326 PMC9498805

139. Ilié M. Malapelle U. Alix-Panabières C. Andersen C.L. Chlebowski S. Lake V. Lacoux C. Lespinet-Fabre V. Bordone O. Heeke S. Perspectives in educating molecular pathologists on liquid biopsy: Toward integrative, equitable, and decentralized precision oncology Mol. Oncol. 2025 20 894 903 10.1002/1878-0261.70163 41235697 PMC13060639

140. Hennessy S. Atsuta Y. Hill S. Rägo L. Juhaeri J. Council for International Organizations of Medical Sciences (CIOMS) Working Group XIII Real-World Data and Real-World Evidence in Regulatory Decision Making: Report Summary from the Council for International Organizations of Medical Sciences (CIOMS) Working Group XIII Pharmacoepidemiol. Drug Saf. 2025 34 e70117 10.1002/pds.70117 40070104 PMC11897686

141. Cave A. Kurz X. Arlett P. Real-World Data for Regulatory Decision Making: Challenges and Possible Solutions for Europe Clin. Pharmacol. Ther. 2019 106 36 39 10.1002/cpt.1426 30970161 PMC6617710

142. Baines R. Stevens S. Austin D. Anil K. Bradwell H. Cooper L. Maramba I.D. Chatterjee A. Leigh S. Patient and Public Willingness to Share Personal Health Data for Third-Party or Secondary Uses: Systematic Review J. Med. Internet Res. 2024 26 e50421 10.2196/50421 38441944 PMC10951832

143. WHO Europe WHO Europe—Secondary Use of Health Data Meeting and Policy Discussions 13 December 2022 Available online: https://www.who.int/europe/news-room/events/item/2022/12/13/default-calendar/meeting-on-secondary-use-of-health-data?utm_source=chatgpt.com (accessed on 17 February 2026)

144. European Commission Reuse of Health Data and Privacy Safeguards. n.d Available online: https://health.ec.europa.eu/ehealth-digital-health-and-care/reuse-health-data_en?utm_source=chatgpt.com (accessed on 17 February 2026)

145. Gallagher T. Dube K. McLachlan S. Ethical issues in secondary use of personal health information IEEE Technol. Policy Ethic. 2018 3 1 5 10.1109/NTPE.2018.9778123

146. Kelly C.J. Karthikesalingam A. Suleyman M. Corrado G. King D. Key challenges for delivering clinical impact with artificial intelligence BMC Med. 2019 17 195 10.1186/s12916-019-1426-2 31665002 PMC6821018

147. Askin S. Burkhalter D. Calado G. El Dakrouni S. Artificial Intelligence Applied to clinical trials: Opportunities and challenges Health Technol. 2023 13 203 213 10.1007/s12553-023-00738-2 PMC9974218 36923325

148. Zeng Z. Yi Z. Xu B. The biological and technical challenges facing utilizing circulating tumor DNA in non-metastatic breast cancer patients Cancer Lett. 2025 616 217574 10.1016/j.canlet.2025.217574 39983895

149. Lu Y. Ren L. Yang M. Liu J. Clinical Management of Circulating Tumor DNA in Breast Cancer: Detection, Prediction, and Monitoring Breast Cancer Targets Ther. 2025 17 851 861 10.2147/BCTT.S542704 PMC12479222 41036092

150. Dang D.K. Park B.H. Circulating tumor DNA: Current challenges for clinical utility J. Clin. Investig. 2022 132 e154941 10.1172/JCI154941 35703177 PMC9197509

151. Pascual J. Attard G. Bidard F.C. Curigliano G. De Mattos-Arruda L. Diehn M. Italiano A. Lindberg J. Merker J. Montagut C. ESMO recommendations on the use of circulating tumour DNA assays for patients with cancer: A report from the ESMO Precision Medicine Working Group Ann. Oncol. 2022 33 750 768 10.1016/j.annonc.2022.05.520 35809752

152. Tang M. Pearson S.A. Simes R.J. Chua B.H. Harnessing Real-World Evidence to Advance Cancer Research Curr. Oncol. 2023 30 1844 1859 10.3390/curroncol30020143 36826104 PMC9955401

153. Khozin S. Blumenthal G.M. Pazdur R. Real-world Data for Clinical Evidence Generation in Oncology JNCI J. Natl. Cancer Inst. 2017 109 djx187 10.1093/jnci/djx187 29059439

154. Pfaffenlehner M. Behrens M. Zöller D. Ungethüm K. Günther K. Rücker V. Reese J.-P. Heuschmann P. Kesselmeier M. Remo F. Methodological challenges using routine clinical care data for real-world evidence: A rapid review utilizing a systematic literature search and focus group discussion BMC Med. Res. Methodol. 2025 25 8 10.1186/s12874-024-02440-x 39810151 PMC11731536

155. Pellat A. Kather J.N. Limited impact of real-world evidence in medical oncology: How can AI turn the tide Eur. J. Cancer 2025 228 115714 10.1016/j.ejca.2025.115714 40876086

156. Tsimberidou A.M. Fountzilas E. Nikanjam M. Kurzrock R. Review of precision cancer medicine: Evolution of the treatment paradigm Cancer Treat. Rev. 2020 86 102019 10.1016/j.ctrv.2020.102019 32251926 PMC7272286

157. Duan X.P. Qin B.D. Jiao X.D. Liu K. Wang Z. Zang Y.-S. New clinical trial design in precision medicine: Discovery, development and direction Signal Transduct. Target. Ther. 2024 9 57 10.1038/s41392-024-01760-0 38438349 PMC10912713

158. Fountzilas E. Tsimberidou A.M. Hiep Vo H. Kurzrock R. Tumor-agnostic baskets to N-of-1 platform trials and real-world data: Transforming precision oncology clinical trial design Cancer Treat. Rev. 2024 125 102703 10.1016/j.ctrv.2024.102703 38484408

159. Berry D.A. The Brave New World of clinical cancer research: Adaptive biomarker-driven trials integrating clinical practice with clinical research Mol. Oncol. 2015 9 951 959 10.1016/j.molonc.2015.02.011 25888066 PMC4516171

160. Wang Z. Zhang R.Y. Ji C. Zhang J.-Y. Yue B.-T. Wang F. Revolutionizing gastrointestinal cancer research with artificial intelligence: From precision patient stratification to real-world evidence World J. Gastrointest. Oncol. 2025 17 111339 10.4251/wjgo.v17.i10.111339 41114095 PMC12531783

161. Kasi P.M. Fehringer G. Taniguchi H. Starling N. Nakamura Y. Kotani D. Powles T. Li B.T. Pusztai L. Aushev V.N. Impact of Circulating Tumor DNA-Based Detection of Molecular Residual Disease on the Conduct and Design of Clinical Trials for Solid Tumors JCO Precis. Oncol. 2022 6 e2100181 10.1200/PO.21.00181 35263168 PMC8926064

162. Verschoor N. Bos M.K. Oomen-de Hoop E. Martens J.W. Sleijfer S. Jager A. Beije N. A review of trials investigating ctDNA-guided adjuvant treatment of solid tumors: The importance of trial design Eur. J. Cancer 2024 207 114159 10.1016/j.ejca.2024.114159 38878446

163. Anagnostou V. Ho C. Nicholas G. Juergens R.A. Sacher A. Fung A.S. Wheatley-Price P. Laurie S.A. Levy B. Brahmer J.R. ctDNA response after pembrolizumab in non-small cell lung cancer: Phase 2 adaptive trial results Nat. Med. 2023 29 2559 2569 10.1038/s41591-023-02598-9 37814061 PMC10579094

164. Guigal-Stephan N. Lockhart B. Moser T. Heitzer E. A perspective review on the systematic implementation of ctDNA in phase I clinical trial drug development J. Exp. Clin. Cancer Res. 2025 44 79 10.1186/s13046-025-03328-4 40022112 PMC11871688

165. Albani F.G. Alghamdi S.S. Almutairi M.M. Alqahtani T. Artificial Intelligence-Driven Innovations in Oncology Drug Discovery: Transforming Traditional Pipelines and Enhancing Drug Design Drug Des. Dev. Ther. 2025 19 5685 5707 10.2147/DDDT.S509769 40626099 PMC12232943

166. Topol E.J. High-performance medicine: The convergence of human and artificial intelligence Nat. Med. 2019 25 44 56 10.1038/s41591-018-0300-7 30617339

167. Mirakhori F. Niazi S.K. Harnessing the AI/ML in Drug and Biological Products Discovery and Development: The Regulatory Perspective Pharmaceuticals 2025 18 47 10.3390/ph18010047 39861110 PMC11769376

168. Frommeyer T.C. Gilbert M.M. Fursmidt R.M. Park Y. Khouzam J.P. Brittain G.V. Frommeyer D.P. Bett E.S. Bihl T.J. Reinforcement Learning and Its Clinical Applications Within Healthcare: A Systematic Review of Precision Medicine and Dynamic Treatment Regimes Healthcare 2025 13 1752 10.3390/healthcare13141752 40724777 PMC12295150

169. Abbas G.H. Khouri E.R. Thaher O. Taha S. Vladimirov M. Oviedo R.J. Schmidt J. Bausch D. Pouwels S. Predictive modeling for metastasis in oncology: Current methods and future directions Ann. Med. Surg. 2025 87 3489 3508 10.1097/MS9.0000000000003279 PMC12140723 40486555

170. Kourou K. Exarchos T.P. Exarchos K.P. Karamouzis M.V. Fotiadis D.I. Machine learning applications in cancer prognosis and prediction Comput. Struct. Biotechnol. J. 2014 13 8 17 10.1016/j.csbj.2014.11.005 25750696 PMC4348437

171. Cherng H.J.J. Herrera A. Circulating Tumor DNA in Diffuse Large B-Cell Lymphoma: From Bench to Bedside? Curr. Treat. Options Oncol. 2024 25 659 678 10.1007/s11864-024-01201-8 38656685

172. Ma L. Guo H. Zhao Y. Liu Z. Wang C. Bu J. Sun T. Wei J. Liquid biopsy in cancer current: Status, challenges and future prospects Signal Transduct. Target. Ther. 2024 9 336 10.1038/s41392-024-02021-w 39617822 PMC11609310

173. Popescu D.C. Găman M.A. Artificial Intelligence for Risk Stratification in Diffuse Large B-Cell Lymphoma: A Systematic Review of Classification Models and Predictive Performances Med. Sci. 2025 13 280 10.3390/medsci13040280 41440512 PMC12734444