Exploring the characteristics of conversational agents in chronic disease management interventions: A scoping review.

Introduction

In recent years, there has been a significant rise in the prevalence of chronic diseases, resulting in a growing number of individuals living with at least one chronic health condition. 1 Chronic conditions have enduring and persistent effects, necessitating patients and healthcare professionals to navigate complex lifestyle and behavioral adjustments and engage in long-term management. 2 Moreover, these conditions reduce quality of life and life expectancy and can escalate personal healthcare expenses due to disability, frequent hospitalizations, and the need for multiple treatment procedures.

In healthcare, the continuum of care 3 is the concept of an integrated care system that guides and tracks patients over time through every subsequent step of health services. The care system contains a range of healthcare and social services, from preventive care to acute treatment, rehabilitation, and long-term care. Over time, the patient goes through various parts of the system, depending on the stage of the treatment or the need for care. The continuum of care aims to provide patients with the right care at the right time and in the most appropriate setting (e.g., home, rehabilitation centers, or hospitals), promoting better health outcomes and patient satisfaction.

Alongside work performed by healthcare professionals, there are often health-related activities that patients and their informal caregivers undertake in managing their health conditions, sometimes referred to patient work. 4 The activities include various processes, from cognitive to physical, performed individually or collaboratively with others, including family and community members, and can be classified as visible when they are acknowledged and valued by others or invisible when they are taken for granted by others and consequently undervalued. 5

Digital technologies have the potential to empower people, giving them a sense of agency and control, allowing them to extend their skills and knowledge and giving them access to experiences and functions that people did not have before. 6 Technologies can reduce the burden of patient work or self-management of health conditions and serve as accessible alternatives to in-person support and supervision. One such technology is conversational agents (CAs), computer systems that imitate human-like conversations using natural language user interfaces involving images, text, and voice. 7 CAs potentially offer the advantages of scalability, reduced costs, lowered stigma, and personalized health support available at any time. 8 , 9 CAs can be delivered through text or speech, making them versatile for different target groups, including children and older individuals. CAs can address various healthcare needs, such as mental health management support, 10 aid in chronic disease self-management, 11 and lifestyle change facilitation, e.g., physical activity and dietary modifications. 12

However, existing digital technologies designed for patients with chronic conditions face challenges adapting to changing health needs and goals. Specific subgroups significantly differ in healthcare preferences and goals and require different information and recommendations. 13 Moreover, patients evolve and change the apps they use throughout the trajectory of their condition, from diagnosis to long-term care, as their health goals change over the course of their disease. 13 The chronic illness trajectory model 14 describes how the course of illness varies for each patient and changes over time. Patients may shift between different illness phases repeatedly in unpredictable and inconsistent ways, as their conditions fluctuate.

Digital systems that are not only focused on one specific stage of the disease but also adapt to the changing needs of users can enable sustained app usage by reducing the necessity to search for more appropriate apps. Such systems would facilitate continuous tracking of healthcare data over time, supporting individuals in effectively managing their health and reducing the disruption caused by switching between different apps.

Despite the extensive research on the application of CAs in healthcare, to our knowledge, there has been no scoping review of different types of CAs in chronic disease management with no constraints on demographics. Existing reviews often restricted their focus to specific health areas, such as mental health, 15 – 17 smoking cessation, 18 physical activity 19 or body weight management 20 , 21 ; population, e.g., young (25 years and younger), 17 or adults (18 years and older) 22 ; or agent types, e.g., embodied, 23 , 24 voice-based, 25 artificial intelligence-based 2 , 21 , 26 , 27 CAs or CAs with unconstrained natural language input, 28 e.g., free text or speech. Other reviews report solely on the evaluation outcomes, e.g., effectiveness and acceptability, 18 , 27 , 29 or on one of the CAs’ aspects, e.g., personalization 30 or design features. 24

This paper provides the results of a literature analysis aimed at addressing several research questions related to the use of CAs in the context of chronic disease management. The research questions explored in this article are:

What are the health domains, and the characteristics of the end users, targeted for CA interventions?

What are the characteristics of the CA studies?

Do CAs address continuum of care and patient work concepts for self-management? Do CAs adapt to the changing needs of users?

What are the CAs’ characteristics? How do different types of CAs map with patients’ profiles and health domains?

By addressing these research questions, this study aims to contribute valuable insights into the design, implementation, and optimization of CAs, as supportive tools in managing chronic health conditions. The findings presented in this paper seek to inform healthcare professionals and technology developers about the potential of these digital solutions to empower patients, improve health outcomes, and enhance the overall quality of life for different groups of people living with chronic diseases.

Methods

This study employs a scoping review approach, 31 which provides a comprehensive overview of the evidence on the chosen research topic. Our approach, therefore, is to give a relative breadth in our review rather than to focus on evidence around a single clinical or systematic question.

Search strategy

The review followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews (PRISMA-ScR) guidelines 32 with the PRISMA-ScR checklist available as a Supplemental Material. We conducted a search of the literature using the electronic databases PubMed, ACM Digital Library, Scopus, and IEEE Xplore. These databases were chosen as they cover relevant aspects of health, technology, and interdisciplinary research and have also been used in other reviews covering similar topics. The main keywords used were “conversational agent” and “health” which were searched in titles and abstracts ( Table 1 ).

Study selection criteria

The review is based on literature dated between the beginning of January 2018 and the end of January 2023 published in English with full text available. We included articles on chronic disease management (e.g., treatment or monitoring) and prevention of specific chronic diseases. Articles that addressed general well-being were not included. Articles must provide some description of the theoretical basis, choice of the intervention components, or CA development process. We included both quantitative and qualitative studies, without any constraints on approach. Studies employing experimental designs that involve group comparisons must include details about the comparators used. Studies focusing only on technical aspects and design features of CAs (e.g., language models, systems, chatbot's personality) and studies using the Wizard of Oz method weren’t included.

Conversational agents

The review considers studies that involve interventions provided by a CA. This review does not include studies where CA focuses on video- and image-based diagnosis (e.g., skin cancer, gait, correct execution of exercises) and whose aim is screening before appointments, filling hospital forms, checking doctors’ availability, and giving diagnoses from the symptoms provided by the user.

Intervention

This review considers studies that evaluate intervention programs that include strategies to provide educational materials, help achieve health goals, and monitor health conditions. Interventions must not be intended for a hospital setting but can be tested in a laboratory environment. The intervention must be directly targeted toward the patients (e.g., not for emergency medical services, clinicians, or medical students).

Validity

This review considers studies that were tested on targeted user groups and have a report on the interventions’ impact on participants and/or participants’ experiences with the CA. Additionally, articles must have evaluation or validation of outcomes (e.g., acceptability, effectiveness) measured by reliable tools.

Results

The literature search process from the four databases resulted in a total of 3151 research papers. The steps of the screening methodology are presented in Figure 1 , along with the number of studies that were excluded in each step. The first step consisted of removing duplicate records of common studies that were found across all databases using the reference manager Zotero, which resulted in 2406 unique articles. All remaining unique studies were then considered in the initial title- and abstract-based screening. Following the proposed exclusion criteria described above, 409 articles remained. Full-text assessment of the remaining papers was then performed, resulting in a final list of 42 articles. The most relevant exclusion factor was validity, as many studies included only preliminary methods and results or presented CAs without any evaluation results (e.g., studies at the pre-pilot stage), or CAs that were evaluated with subjects different to the intended population.

Characteristics of included studies

The full list of included studies is provided in Tables 2 – 7 . Article publication dates ranged from 2018 to 2023. Eighteen studies were conducted in the USA, 8 , 10 , 33 – 48 three each in Switzerland 49 – 51 and China, 52 – 54 two each in Australia, 55 , 56 France, 57 , 58 and the UK, 59 , 60 and one each in Canada, 61 Japan, 62 Singapore, 63 Saudi Arabia, 64 South Africa, 65 India, 66 Germany, 67 Italy, 68 Norway, 69 Spain, 70 the Netherlands, 71 and across Europe. 72 The total number of participants ranged from 6 to 4737 and one study 69 didn’t provide information on participant numbers. Participants aged between 5 and 86 years but not all articles provided information on age ranges.

What are the health domains and the characteristics of the users, targeted for CA interventions?

When creating technology for a specific health domain, it is essential to consider symptoms, treatment requirements, and challenges associated with that domain that people may encounter at different stages. Moreover, understanding users’ characteristics, such as demographics, medical history, cultural background, goals, and preferences, allows the content and interaction strategies to be tailored. This enhances engagement, adherence, and overall health outcomes. 75 , 76 Table 2 shows information about the health domains addressed by the CAs in the selected articles, providing the context of the disease, including phase, e.g., prevention, after-care or during (between onset and end of the disease), as well as information about the target users.

Health domains

Among the various types of chronic conditions, different types of diabetes (type 1, type 2, gestational diabetes, and prediabetes) 47 , 55 , 56 , 63 , 65 , 66 , 68 , 69 , 71 and mental health issues (depression, depressive symptoms, anxiety, bipolar disorder, and other complex, difficult-to-treat mental disorders) 38 , 45 , 46 , 53 , 54 , 60 received the most attention. Other types of chronic conditions included six studies on cancer (multiple myeloma, breast cancer, cancer genetic evaluation, and after cancer treatment), 37 , 39 , 48 , 57 , 58 , 67 four studies on addictions and substance abuse (smoking cessation, alcohol and methamphetamine use disorder, and substance misuse), 33 , 34 , 36 , 52 four studies on cardiovascular disease (heart failure and atrial fibrillation), 8 , 35 , 40 , 61 three studies on obesity, 44 , 51 , 59 two studies on chronic pain, 49 , 62 and one study each on asthma, 50 autoimmune (celiac disease), 64 genetic condition (sickle cell disease), 72 and functional bowel disorder (irritable bowel syndrome). 42 Additionally, there were four studies focused on comorbid diseases: two on musculoskeletal conditions and mental health, 10 , 43 one on type 2 diabetes and depressive disorder, 70 and another one on obesity and cancer. 40

Most of the systems were aimed at supporting people who already have chronic diseases (37 studies). There were only three studies that highlighted their focus on prevention: one study focused on patients without cancer who were eligible for cancer genetic evaluation, 37 one on communicating breast cancer risk and the recommended medical guidelines to healthy women, 48 and another on promoting healthy lifestyle behavior changes with the focus on diabetes and prediabetes knowledge in the general population. 63 Two studies focused on after-disease care: one study concentrated on overweight or obese cancer survivors 41 and another on young people after cancer treatment. 39

Targeted users’ characteristics

In the selected studies, people with chronic conditions, or the risk of acquiring one were the most common final targeted interaction recipients. Only three studies targeted the interaction of patient–parent dyads. 47 , 50 , 59

The age groups of target users varied. There were six studies focused on children and teenagers. 45 , 47 , 50 , 51 , 59 , 71 Four studies targeted young adults. 38 , 39 , 53 , 54 The remaining 32 studies were for different groups of adults.

There was no information about the duration of illness, medication usage, number and type of comorbidities, employment and marital status or any other socio-demographic information of target users that might be important for intervention development. Only a few studies were targeted recipients additionally specified; two studies targeted university students, 38 , 54 one study focused on veterans, 34 one specifically targeted African American/Black women, 44 and one had a focus on workers. 62

What are the characteristics of the CA studies?

When conducting studies, it can be meaningful to test interventions with multiple study arms to compare them, evaluate their efficacy and inform decision-making. Conducting studies with participants representative of the target group enhances applicability and generalizability that improves the validity and relevance of findings for the interventions’ intended users. Knowing the intervention duration can be important for the understanding of the impact and effectiveness of the interventions. Moreover, chosen measure types guide the evaluation of health apps, ensuring evidence-based interventions and objective outcome assessment. Thus, effectiveness, a fundamental aspect of intervention assessment, can be measured through various outcome indicators chosen by researchers, such as symptom reduction, changes in health behavior and health indicators (e.g., blood pressure, glucose levels), improvements in quality of life, and others. These measures serve as quantifiable evidence of an intervention's success in achieving its intended goals. The characteristics of the studies are summarized in Table 3 . If there were multiple race groups participating in the study, we mention only those above 20%. We reported gender and sex as explicitly stated in the studies. If the totals do not add up to 100, it means this information was not provided in the studies.

Participants

The age of the research participants corresponded to the targeted users’ descriptions since this was a criteria for inclusion in the review. There were six studies focused on children and teenagers aged from 5 years to 18 years old. 45 , 47 , 50 , 51 , 59 , 71 Four studies targeted young adults, 17–29 years old. 38 , 39 , 53 , 54 One study had a wide range age of targeted users and participants from 13 to 65+ years old. 58 The remaining 31 studies involved different groups of adults. Three studies didn’t provide any information about participants, 64 , 67 , 69 except that they corresponded to the target users, e.g., already had the health condition or risk of acquiring the disease. Moreover, one of these studies, Baertsch et al., 67 besides patients, had healthcare professionals and caregivers among the users since the program was advertised and had open access.

Apart from age, gender, or sex was frequently mentioned in the articles as a demographic characteristic of the participants. There were 17 articles that used the term “gender,” 12 that used the term “sex,” 2 that used both terms, and 11 did not specify. Unfortunately, it is often impossible to determine exactly what researchers meant when they used a particular term. Furthermore, some of the articles provided additional information on specific health indicators, such as body mass index (BMI) or disease duration, employment and marital statuses, race, and ethnicity. Having a more detailed description provides valuable insights for understanding the diverse demographics and health profiles of the participants, allowing conducting in-depth analysis of the interventions’ effectiveness and relevance across different populations.

Duration

There were two studies with unspecified duration. 67 , 68 Of those with specified domains, there was one study 69 that collected data over 20 weeks (there were two periods: 8 and 12 weeks), though it is not clear for how long and how many times each user engaged with the chatbot since the data were anonymous and the identification of users was not permitted. Five studies took only 1 day to finish the planned intervention. 37 , 48 , 52 , 57 , 72 Two studies each were for 1 week and 2 weeks, nine studies were about 1 month long, seven studies had a 2 month duration, and eight studies had a 3 month duration. Several studies lasted more than 3 months: two studies for about 4 months, 54 , 66 one study for 5.5 months, 51 one study for 9 months, 70 and 3 studies for 1 year. 55 , 56 , 58

Study arms

Most of the studies were single-armed (22/42). There were 15 two-arm studies, 11 of which compared intervention and control groups (see Table 3 ), including five that compared intervention and usual care, 40 , 44 , 51 , 56 , 62 four that compared immediate intervention and waitlist control groups, 33 , 39 , 42 , 45 and two compared intervention and control groups with no treatment. 59 , 64 Another setup encountered is a comparison of two alternative interventions: responses from chatbot versus responses from physicians, 57 interaction with a chatbot concerning pain management versus reception motivational messages unrelated to chronic pain, 49 bibliotherapy versus chatbot interaction, 54 and smart speaker versus visually animated and voice-enabled avatar interaction. 35 There were also five three-arm studies, that compared an e-book, a general chatbot, and mental health chatbot 53 ; control group that only received a link to a book, 2 and 4 weeks intervention groups 38 ; control version, adaptive, and non-adaptive embodied conversational agent (ECA) 48 ; control group that received printed written information, smart speaker, and text-based intervention 41 ; and usual orthopedic care without any specific mental health intervention, usual orthopedic care with in-person psychological counseling and usual orthopedic care with digital mental health intervention. 43

Measures

A diverse range of measures was used to assess the interventions in the selected studies. One of the most common measures were health outcomes (23/42) which were measured by physical assessments like BMI, glycated hemoglobin level, blood pressure, and heart rate, 44 , 51 , 56 , 59 , 66 , 70 and questionnaires, such as Generalized Anxiety Disorder-7, Patient Health Questionnaire-9, International Physical Activity Questionnaire, Pittsburgh Sleep Quality Index, and Perceived Stress Scale. 33 , 38 , 39 , 42 – 46 , 49 , 51 – 54 , 56 , 59 , 60 , 62 , 63 , 65 , 68 , 70 – 72 All the studies demonstrated improvement, but not always statistically significant.

The same metrics are often used to evaluate engagement, adherence, and interaction with apps and CAs, e.g., number, lengths and context of messages sent, and this can create challenges in distinguishing the specific aspect being assessed. Among the selected studies, 26 evaluated chats with CAs and activity in the apps (we called this measure Adherence, Engagement, and Interaction in Table 3 ), which was measured by a variety of methods, including analysis of interview transcripts, 8 chat transcripts and metrics, e.g., number and length of messages, 35 , 37 – 40 , 45 , 49 , 51 , 53 , 56 , 58 , 60 , 62 – 64 , 69 , 70 app usage, e.g., number of accesses to the app and videos watched, 42 , 44 , 45 , 56 , 70 different questionnaires, e.g., Godspeed for assessing human-like traits of CA, 36 the Usefulness Scale for Patient Information Material for evaluating perceived usefulness 72 and the User Engagement Scale–Short Form questionnaire for measuring self-reported user engagement, 68 and other surveys to evaluate various aspects, e.g., the level of confidence, attitude, and perceived quality of the answers. 39 , 47 , 54 , 57 , 58 Many metrics based on chat and app usage provided quantitative values that were evaluated relative to each other, such as the number of messages sent weekly. The results indicated that participants found the CAs to be useful, reported overall positive attitudes, and showed high engagement and adherence rates. All other 26 measurements identified were used in fewer than 13 articles.

Effectiveness

Most of the studies didn’t talk explicitly about the effectiveness. It can be assumed that effectiveness of the intervention is determined by the results of the selected measures, as noted in the previous paragraph, such as health outcomes, usability, and engagement. However, there are several articles that mentioned why they didn’t provide information about effectiveness; some of the studies weren’t designed to test the effectiveness, 63 other couldn’t establish it for different reasons, e.g., the small sample size and/or insufficient intervention duration. 45 , 46 , 51 , 59 , 60

Do CAs address continuum of care and patient work concepts for self-management? Do CAs adapt to the changing needs of users?

CAs play a multifaceted role in chronic disease management, addressing the continuum of care by providing, for example, education, remote monitoring, and seamless health records integration that can enable access to comprehensive patient data that contributes to creating personalized and tailored interventions. CAs contribute to patient work through symptom tracking and coaching, empowering patients and caregivers to self-manage their health. Moreover, human involvement, e.g., relatives, peers, and healthcare professionals, in chronic condition management can provide social support and motivation for patients, enhancing adherence to the interventions and overall well-being. 77 , 78 Furthermore, the inclusion of medical records in health applications can facilitate the development of personalized interventions, resulting in enhanced health outcomes, greater user engagement, and heightened satisfaction.

The purpose of the reviewed CAs are presented in Table 4 and the factors impacting CA's role in self-management are shown in Table 5 for any study that mentioned the CA's factors.

CA purposes

In the selected studies, CAs had a diverse range of purposes. The most common ones were providing information and education (33/42), delivering cognitive therapy (12), and providing mental health and emotional support (11) (see Table 4 ). All other purposes, such as providing feedback, collecting data, sending reminders, and motivational messages and monitoring symptoms, were mentioned fewer than six times.

CA personalization

Twenty-five CAs didn’t adopt message content to users’ characteristics at all (see Table 5 ). Five studies mentioned that CAs sent “personalized text messages” and “tailored replies depending on users’ responses” without further explanation. 49 , 56 – 58 , 62 In one study participants can choose to receive more or less details on the topics based on their preferences. 37 Based on participants’ goals, habits, and data entered during a conversation with CA, four CAs provided personalized feedback, 36 , 44 , 50 , 71 and four CAs sent personalized educational, motivational, and coaching messages. 41 , 55 , 61 , 66 One CA tailored the conversation to the user's present situation and to the reported need at that moment, 45 one CA referred to earlier data entered, tasks, or activities performed and added user's name 72 and another CA added user's name and appropriate time context to the messages, e.g., “Good Afternoon John.” 40

Medical records

We have only considered the usage of medical records in conjunction with the CAs. For example, cases when medical records were used for recruiting participants are not demonstrated in Table 5 . Thirty studies didn’t mention any connection to patients’ medical records (see Table 5 ). Nine studies used medical records data for the studies. 10 , 35 , 37 , 40 , 41 , 43 , 56 , 61 , 66 One study recorded information after the study to participants’ electronic health record (EHR) 37 and one study used participants’ medical charts to confirm any additional mental health diagnoses and record their initial treatment plan. 45 Moreover, one study aimed to integrate the findings in existing patient web portals that support the care of patients with type 1 diabetes, 71 and one study noted the possibility of integrating the chatbot's administration panel into the existing EHR system. 38 One study collected medical history and stored it in the app when the patient was onboarded to the program 61 and one study stored personal health record files with additional diabetes-related information (e.g., laboratory reports and details of treating physician) that users logged in the mobile application. 66 Among these 12 studies, there were three on diabetes, 56 , 66 , 71 three on cardiovascular conditions, 35 , 40 , 61 two on mental health, 38 , 45 two on comorbid chronic pain and mental health, 10 , 43 one on cancer, 37 and one on comorbid cancer and obesity. 41

Human involvement

Despite the evidence that human involvement, e.g., family members, peers, and healthcare professionals, in the management of chronic conditions can offer patients social support and motivation, 77 , 78 in 31 studies there was no human involvement in the intervention groups (see Table 5 ). Four studies allowed the participants to communicate with health experts, two studies only through chat, 10 , 43 one through chat and face-to-face interaction 50 and one through chat and voice calls. 66 In three studies, healthcare professionals monitor participants’ engagement and results. 35 , 52 , 70 In one study, healthcare professionals could intervene by telephone (Mental Assistance Hotline) when the participants reported that they needed emergency psychological assistance 54 and in another study, the app alerted healthcare professionals in case of a lack of chat interaction during more than 2 days and four standardized counseling on-site visits, two via phone and 10 minutes of interaction through chat app was included. 51 Family members were included in one study by Kowatsch et al., 50 CA could send SMS text messages to them, and healthcare professionals had a possibility to communicate with family members through the app. In Stasinaki et al. 51 patients were able to chat with each other and in Klaassen et al. 71 there was a message exchange between patients, their peers, and their caregivers. In two of the studies without human involvement the CA motivated users to join online peer-support patient communities or request health coaching from a human 72 or search for local support groups. 67

The targeted health domains for the studies with human involvement were diabetes, 66 , 71 comorbid chronic pain and mental health, 10 , 43 mental health, 54 comorbid diabetes and mental health, 70 asthma, 50 cardiovascular conditions, 35 addictions and substance abuse, 52 cancer, 67 and obesity. 51

What are the CAs’ characteristics? How do different types of CAs map with patients’ profiles and health domains?

Choosing or designing CAs suitable to target users requires an understanding of their characteristics, such as type of CA, input and output modalities, as these influence the user experience and engagement. The input modality refers to how users interact with the CA, such as through text, speech, or by choosing from the responses menu, while the output modality relates to the CA's responses, whether in text, speech, or visual forms. Moreover, different types of CAs have unique capabilities and advantages. For instance, relational agents focus on building and maintaining long-term, social–emotional relationships with users, fostering a sense of trust and empathy 79 ; ECAs have the same properties as humans in face-to-face conversation, including producing and responding to verbal and nonverbal communication 80 ; and task-oriented dialogue agents are designed to perform a specific function or deliver a particular service, such as providing educational tutorials. These agents are different from general-purpose CAs, which are more versatile and capable of handling a wide range of tasks and conversations. 81

Theoretical frameworks are essential when developing health interventions and applications as they provide a structured and evidence-based foundation, guiding the design, implementation, and evaluation process to ensure effectiveness and alignment with established health behavior principles.

Underpinning frameworks are presented in Table 6 and CAs’ technical characteristics are in Table 7 .

Theoretical frameworks

One-third of all the studies (14/42) did not mention any theoretical base on which the interventions and the application design were based (see Table 6 ). The rest of the studies had a lot of variety in the theories, therapies, techniques, and frameworks that were considered when developing the studies. A total of 40 different theories were mentioned, with cognitive behavioral therapy, mindfulness (includes training and all mindfulness-based therapy and techniques), motivational interviewing (includes brief motivational interviewing), dialectical behavior therapy, behavior change techniques, and self-determination theory being referenced most frequently (10, 6, 5, 4, 3, and 3 times, respectively), with the remaining 33 theories having no more than two mentions each. All the frameworks can be categorized into several domains: psychotherapy and counseling theories (acceptance and commitment therapy, cognitive behavioral therapy, dialectical behavior therapy, emotionally focused therapy, interpersonal psychotherapy, motivational enhancement therapy, solution-focused brief therapy), health behavior change theories (behavior change techniques, goal setting theory, self-determination theory, social cognitive theory, stress and coping theory, theory of planned behavior, transtheoretical model), educational and learning theories (adult learning theory, cognitive theory of multimedia learning, experiential learning theory), healthcare models and guidelines (American Association of Diabetes Educators (AADE) framework, chronic care model, chronic-disease extended model, obesity-related behavioral intervention trials framework, US Clinical Practice Guidelines, World Health Organization's handbooks on how to implement text-based mHealth interventions), mind-body techniques (mindfulness techniques, sleep meditations, deep breathing techniques), technology and digital health frameworks (digital persuasion model, persuasive system design model, technology acceptance model).

CAs’ characteristics

The majority of CAs in the reviewed studies were text-based (28 out of 42) (see Table 7 ). Seven CAs only allowed users to choose an answer from the options provided 46 , 51 , 62 , 63 , 65 , 70 , 72 and three CAs had several free text questions, e.g., provide the user's name. 37 , 49 , 50 Among these CAs, four sent media, like audio lessons and pictures with exercises, to users. 49 , 62 , 63 , 65 In three studies, participants could type their responses freely 39 , 57 , 60 and, in Liu et al., 54 participants could also send voice messages to the chatbot. In six CAs, users could both choose responses from a menu and send free text messages or choose from the responses menu. 10 , 38 , 43 , 58 , 68 , 69 Six studies didn’t specify users’ input format. 33 , 42 , 45 , 53 , 64 , 66 There were three studies with unidirectional coaches where the CAs sent messages to the participant, but the participant could not communicate with the agent. 41 , 44 , 71 Four studies based their intervention on Amazon's Alexa, 35 , 59 , 61 , 67 including one that compared Amazon's Alexa with a visually animated and voice-enabled avatar. 35 Additionally, one study compared unidirectional coach with Amazon voice assistant. 41 Ten studies used an ECA to provide interventions, 8 , 34 – 36 , 40 , 47 , 48 , 52 , 55 , 56 the users’ input modes varied: most systems allowed interaction with the ECA only with predefined answer options, 8 , 48 , 52 while others allowed voice and/or free text messages. 36 , 55 , 56

Among the selected studies, there were five with relational agents designed to build and maintain long-term emotional relationships with their users. 33 , 40 , 45 , 47 , 60 Three of them were text-based for mental health and addictions and substance abuse 33 , 45 , 60 and two were ECA targeting diabetes and cardiovascular issues. 40 , 47

Discussion

Advancements in technologies have facilitated a plethora of mobile apps specifically designed to support and improve health habits that give patients more control and a sense of agency, empowering them to take charge of their health. 82 Existing digital technologies for patients with chronic conditions face challenges in adapting to changing health needs and goals, requiring diverse information and recommendations for specific subgroups. 13 Implementing adaptable digital systems can enable sustained app usage, supporting individuals in effectively managing their health and reducing disruption caused by switching between different apps.

Relevant studies were obtained from the PubMed, ACM Digital Library, Scopus, and IEEE Xplore databases. The methodology adopted in this review aligned well with the PRISMA-ScR guidelines and checklist 32 process, see the Supplemental Material for further details. The review included 42 studies reporting interventions delivered by CAs, targeting chronic diseases. Diabetes, mental health, and cancer are the diseases most commonly targeted by CA interventions, as opposed to other chronic conditions, such as autoimmune, genetic conditions, functional bowel disorders, and asthma. The review shows that current CAs address a wide variety of chronic diseases. Although it provides valuable insights, the extensive range of conditions complicates the ability to compare them within and between conditions. To address this, future research should focus on evaluating CAs specifically tailored to particular chronic diseases.

The overall trends can be summarized as follows:

Focus on stable chronic condition management

The analysis reveals that the majority of studies have primarily targeted individuals who already have chronic diseases. There seems to be a disproportionate emphasis on stable chronic condition management rather than on other stages of the disease where people might have other needs, for example, newly diagnosed patients or patients after crises. Only one study mentioned it was designed to address current and future patients’ needs. 58 This finding highlights a potential gap in leveraging digital interventions that adapt to the changing needs of people with chronic diseases, which could significantly impact the overall burden of these conditions on healthcare systems.

Limited representation of specific demographics

The research highlights the lack of studies focusing on certain demographic groups, including children/teenagers and older adults, which aligns with previous reviews. 22 While interventions for adults aged 18 and above are relatively abundant, there is a noticeable gap in providing support for younger and older populations. Understanding the specific challenges and requirements of these groups is crucial for designing age-appropriate and inclusive digital health solutions.

Incomplete participant information

For this review, we selected studies featuring participants who matched the targeted user group. Yet, there is often limited information about the participants beyond basic demographics such as age and gender or sex. Since studies sometimes did not explicitly report all values beyond the primary gender or sex and the terms “gender” and “sex” can be used interchangeably, with “gender” increasingly used to describe biological variations traditionally assigned to “sex,” 83 comparing these studies becomes challenging. Marital status, ethnicity, and other socio-economic features are often missing. Moreover, studies failed to provide a thorough comprehension of participants’ needs, aligning with the previous reviews. 22 , 28 This lack of comprehensive participant information might hinder a thorough understanding of the interventions’ applicability and effectiveness across diverse populations.

Importance of human involvement

While CAs offer promising directions, the limited provision for human interaction within these systems is identified as a potential drawback. Human support from healthcare professionals, family members, and peers with similar health problems can provide emotional encouragement, motivation, and vital insights that are essential for successful long-term management of chronic conditions. 77 , 78

Limited integration of medical records

The review indicates that limited use of medical records by the CAs potentially hinder smooth communication and collaboration between individuals and healthcare professionals. Moreover, medical records in health apps can enable the creation of personalized and tailored interventions that lead to improved health outcomes and increased user engagement and satisfaction.

Insufficient information on theoretical framework

Several studies didn’t provide information about the theoretical framework guiding the design and implementation of interventions. This omission raises concerns about the basis for the interventions and highlights the importance of incorporating robust theoretical foundations in future research and development.

Lack of unified evaluation measures

In accordance with prior reviews, we also found that the evaluation measures used to assess the effectiveness of the interventions and their impact on chronic conditions varied widely across studies. 15 , 28 This lack of standardization makes it challenging to compare and generalize findings.

Limited use of technologies and user information

There is potential for CAs to assist people in actively managing their conditions instead of passively consuming information. However, most current CAs have dialogue management systems that do not consider user preferences, goals, or history of interaction with the system. To better serve user needs, the design of these agents should evolve. The linguistic data generated by users during interaction with CAs holds the capacity to provide insights into users’ emotional and physical states. These data can help people with self-management and provide valuable information for the patients’ support team. A scoping review on psychology-oriented ECAs revealed that most agents were still in their initial phases of development and evaluation, 23 aligning with our results. A systematic review on health-related CAs with unconstrained language input capabilities found limited use of agent-based systems capable of handling complex dialogues and allowing users to lead conversation. 28 The review found only one study evaluated such systems in health contexts, but agents weren't designed for health-related queries. The authors indicated the requirement for large training datasets as a major drawback, potentially slowing their adoption in health applications.

Another factor to consider is that, at the time of writing, advances in the application of large language models offer significant opportunities in various fields, including healthcare. However, this has not yet been reflected in the available literature. It will be interesting to observe how this technological progress could potentially address the highlighted issues.

Strengths and limitations

The main strength of this review is its novelty. To the best of our knowledge, this is the first scoping review addressing different types of CAs in chronic disease management with no constraints on demographics. Nevertheless, it is essential to acknowledge that this analysis has its limitations. First, given that the research question is very broad, a detailed review of the use of CAs in different chronic diseases was not possible. Second, the review was exclusively focused on text and voice CAs, omitting those utilizing images and video. Third, we included articles in English with full text available only and the search in other databases, such as CINAHL and Web of Science, was not conducted. Fourth, our analysis only incorporated features, e.g., theoretical frameworks, explicitly mentioned in the articles. Therefore, certain characteristics may not be fully represented in our review, potentially limiting the generalizability of the conclusions due to potential bias.

Conclusions

This paper highlights the current state of CAs for chronic condition management and raises important considerations for future research and development in this field. The findings emphasize the health application requirements to adapt to the changing needs during the course of illness that varies for each patient and changes over time, customize interventions based on specific user subgroups, and improve the reporting of study participant characteristics to enhance the applicability of the findings. It also underscores the importance of incorporating human support and medical records integration within digital health solutions to provide more effective care for individuals living with chronic conditions. Additionally, studies should clearly articulate the theoretical frameworks guiding their interventions, and efforts should be made to standardize evaluation measures to facilitate meaningful comparisons between studies. Future research could more closely explore studies on each particular disease to gain a deeper understanding of CAs’ use and potential in disease management.

Supplemental Material

sj-docx-1-dhj-10.1177_20552076241277693 - Supplemental material for Exploring the characteristics of conversational agents in chronic disease management interventions: A scoping review

Supplemental material, sj-docx-1-dhj-10.1177_20552076241277693 for Exploring the characteristics of conversational agents in chronic disease management interventions: A scoping review by Ekaterina Uetova, Lucy Hederman, Robert Ross and Dympna O’Sullivan in DIGITAL HEALTH

Acknowledgments

We would like to thank Róisín Guilfoyle, research support librarian at Technological University Dublin, for her assistance and guidance in this research.

Footnotes

Contributorship: Ekaterina Uetova researched literature and wrote the first draft of the manuscript. All authors reviewed and edited the manuscript and approved the final version of the manuscript.

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding: This work was conducted with the financial support of the Science Foundation Ireland Centre for Research Training in Digitally Enhanced Reality (d-real) under Grant No. 18/CRT/6224 and the ADAPT SFI Research Centre for AI-Driven Digital Content Technology under Grant No. 13/RC/2106_P2. For the purpose of Open Access, the author has applied a CC BY public copyright license to any Author Accepted Manuscript version arising from this submission.

Ethical approval: Ethical approval was not required as this is a review paper.

Guarantor: Ekaterina Uetova.

Reporting guidelines: Completed Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews (PRISMA-ScR) flowchart is included as a cited figure, and the PRISMA-ScR checklist is available as a Supplemental Material .

ORCID iD: Ekaterina Uetova https://orcid.org/0009-0009-9605-9616

Supplemental material: Supplemental material for this article is available online.

References

1. Anderson E, Durstine JL. Physical activity, exercise, and chronic diseases: a brief review. Sports Med Health Sci 2019; 1: 3–10.

2. Schachner T, Keller R, Wangenheim FV. Artificial intelligence-based conversational agents for chronic conditions: systematic literature review. J Med Internet Res 2020; 22: e20701.

3. American Psychological Association. Continuum of care. APA Dictionary of Psychology. https://dictionary.apa.org/ (accessed 23 June 2023).

4. Valdez RS, Holden RJ, Novak LL, et al. Transforming consumer health informatics through a patient work framework: connecting patients to context. J Am Med Inform Assoc JAMIA 2015; 22: 2–10.

5. Corbin J, Strauss A. Managing chronic illness at home: three lines of work. Qual Sociol 1985; 8: 224–247.

6. Schneider H, Eiband M, Ullrich D, et al. Empowerment in HCI – A survey and framework. In: Proceedings of the 2018 CHI Conference on Human Factors in Computing Systems , Montreal, QC, Canada, pp.1–14: ACM.

7. Hebbar A. Augmented intelligence: enhancing human capabilities. In: 2017 Third International Conference on Research in Computational Intelligence and Communication Networks (ICRCICN) , 2017, pp.251–254.

8. Bickmore TW, Kimani E, Trinh H, et al. Managing chronic conditions with a smartphone-based conversational virtual agent. In: Proceedings of the 18th International Conference on Intelligent Virtual Agents , New York, NY, USA, pp.119–124: Association for Computing Machinery.

9. Pereira J, Díaz Ó. Using health chatbots for behavior change: a mapping study. J Med Syst 2019; 43: 135.

10. Leo AJ, Schuelke MJ, Hunt DM, et al. A digital mental health intervention in an orthopedic setting for patients with symptoms of depression and/or anxiety: feasibility prospective cohort study. JMIR Form Res 2022; 6: e34889.

11. Ter Stal S, Sloots J, Ramlal A, et al. An embodied conversational agent in an eHealth self-management intervention for chronic obstructive pulmonary disease and chronic heart failure: exploratory study in a real-life setting. JMIR Hum Factors 2021; 8: e24110.

12. Maher CA, Davis CR, Curtis RG, et al. A physical activity and diet program delivered by artificially intelligent virtual health coach: proof-of-concept study. JMIR MHealth UHealth 2020; 8: e17558.

13. Yin K, Jung J, Coiera E, et al. How patient work changes over time for people with multimorbid type 2 diabetes: qualitative study. J Med Internet Res 2021; 23: e25992.

14. Corbin JM. The Corbin and Strauss chronic illness trajectory model: an update. Sch Inq Nurs Pract 1998; 12: 33–41.

15. Jabir AI, Martinengo L, Lin X, et al. Evaluating conversational agents for mental health: scoping review of outcomes and outcome measurement instruments. J Med Internet Res 2023; 25: e44548.

16. Gaffney H, Mansell W, Tai S. Conversational agents in the treatment of mental health problems: mixed-method systematic review. JMIR Ment Health 2019; 6: e14166.

17. Balan R, Dobrean A, Poetar CR. Use of automated conversational agents in improving young population mental health: a scoping review. Npj Digit Med 2024; 7: 1–9.

18. He L, Balaji D, Wiers RW, et al. Effectiveness and acceptability of conversational agents for smoking cessation: a systematic review and meta-analysis. Nicotine Tob Res 2023; 25: 1241–1250.

19. Luo TC, Aguilera A, Lyles CR, et al. Promoting physical activity through conversational agents: mixed methods systematic review. J Med Internet Res 2021; 23: e25486.

20. Noh E, Won J, Jo S, et al. Conversational agents for body weight management: systematic review. J Med Internet Res 2023; 25: e42238.

21. Chew HSJ. The use of artificial intelligence-based conversational agents (chatbots) for weight loss: scoping review and practical recommendations. JMIR Med Inform 2022; 10: e32578.

22. Griffin AC, Xing Z, Khairat S, et al. Conversational agents for chronic disease self-management: a systematic review. AMIA Annu Symp Proc AMIA Symp 2020; 2020: 504–513.

23. Provoost S, Lau HM, Ruwaard J, et al. Embodied conversational agents in clinical psychology: a scoping review. J Med Internet Res 2017; 19: e151.

24. ter Stal S, Kramer LL, Tabak M, et al. Design features of embodied conversational agents in eHealth: a literature review. Int J Hum–Comput Stud 2020; 138: 102409.

25. Bérubé C, Schachner T, Keller R, et al. Voice-Based conversational agents for the prevention and management of chronic and mental health conditions. Systematic literature review . J Med Internet Res 2021; 23: e25933.

26. Bin Sawad A, Narayan B, Alnefaie A, et al. A systematic review on healthcare artificial intelligent conversational agents for chronic conditions. Sensors 2022; 22: 2625.

27. Milne-Ives M, de Cock C, Lim E, et al. The effectiveness of artificial intelligence conversational agents in health care. Systematic Review . J Med Internet Res 2020; 22: e20346.

28. Laranjo L, Dunn AG, Tong HL, et al. Conversational agents in healthcare: a systematic review. J Am Med Inform Assoc JAMIA 2018; 25: 1248–1258.

29. Balaji D, He L, Giani S, et al. Effectiveness and acceptability of conversational agents for sexual health promotion: a systematic review and meta-analysis. Sex Health 2022; 19: 391–405.

30. Kocaballi AB, Berkovsky S, Quiroz JC, et al. The personalization of conversational agents in health care: systematic review. J Med Internet Res 2019; 21: e15360.

31. Noble H, Smith J. Reviewing the literature: choosing a review design. Evid Based Nurs 2018; 21: 39–41.

32. Tricco AC, Lillie E, Zarin W, et al. PRISMA extension for scoping reviews (PRISMA-ScR): checklist and explanation. Ann Intern Med 2018; 169: 467–473.

33. Prochaska JJ, Vogel EA, Chieng A, et al. A randomized controlled trial of a therapeutic relational agent for reducing substance misuse during the COVID-19 pandemic. Drug Alcohol Depend 2021; 227: 108986.

34. Abdullah AS, Gaehde S, Bickmore T. A tablet based embodied conversational agent to promote smoking cessation among veterans: a feasibility study. J Epidemiol Glob Health 2018; 8: 225–230.

35. Apergi LA, Bjarnadottir MV, Baras JS, et al. Voice interface technology adoption by patients with heart failure: pilot comparison study. JMIR MHealth UHealth 2021; 9: e24646.

36. Boustani M, Lunn S, Visser U, et al. Development, feasibility, acceptability, and utility of an expressive speech-enabled digital health agent to deliver online, brief motivational interviewing for alcohol misuse: descriptive study. J Med Internet Res 2021; 23: e25837.

37. Chavez-Yenter D, Kimball KE, Kohlmann W, et al. Patient interactions with an automated conversational agent delivering pretest genetics education: descriptive study. J Med Internet Res 2021; 23: e29447.

38. Fulmer R, Joerin A, Gentile B, et al. Using psychological artificial intelligence (TESS) to relieve symptoms of depression and anxiety: randomized controlled trial. JMIR Ment Health 2018; 5: e64.

39. Greer S, Ramo D, Chang Y-J, et al. Use of the chatbot “Vivibot” to deliver positive psychology skills and promote well-being among young people after cancer treatment: randomized controlled feasibility trial. JMIR MHealth UHealth 2019; 7: e15018.

40. Guhl E, Althouse AD, Pusateri AM, et al. The atrial fibrillation health literacy information technology trial: pilot trial of a mobile health app for atrial fibrillation. JMIR Cardio 2020; 4: e17162.

41. Hassoon A, Baig Y, Naiman DQ, et al. Randomized trial of two artificial intelligence coaching interventions to increase physical activity in cancer survivors. Npj Digit Med 2021; 4: 168.

42. Hunt M, Miguez S, Dukas B, et al. Efficacy of Zemedy, a mobile digital therapeutic for the self-management of irritable bowel syndrome: crossover randomized controlled trial. JMIR MHealth UHealth 2021; 9: e26152.

43. Leo AJ, Schuelke MJ, Hunt DM, et al. Digital mental health intervention plus usual care compared with usual care only and usual care plus in-person psychological counseling for orthopedic patients with symptoms of depression or anxiety: cohort study. JMIR Form Res 2022; 6: e36203.

44. Napolitano MA, Harrington CB, Patchen L, et al. Feasibility of a digital intervention to promote healthy weight management among postpartum African American/Black women. Int J Environ Res Public Health 2021; 18: 2178.

45. Nicol G, Wang R, Graham S, et al. Chatbot-delivered cognitive behavioral therapy in adolescents with depression and anxiety during the COVID-19 pandemic: feasibility and acceptability study. JMIR Form Res 2022; 6: e40242.

46. Schroeder J, Wilkes C, Rowan K, et al. Pocket skills: a conversational mobile web app to support dialectical behavioral therapy. In: Proceedings of the 2018 CHI Conference on Human Factors in Computing Systems , New York, NY, USA, pp.1–15: Association for Computing Machinery.

47. Thompson D, Callender C, Gonynor C, et al. Using relational agents to promote family communication around type 1 diabetes self-management in the diabetes family teamwork online intervention: longitudinal pilot study. J Med Internet Res 2019; 21: e15318.

48. Zhou S, Bickmore T. A virtual counselor for breast cancer genetic counseling: adaptive pedagogy leads to greater knowledge gain. In: Proceedings of the 2022 CHI Conference on Human Factors in Computing Systems , pp.1–17. New York, NY, USA: Association for Computing Machinery.

49. Hauser-Ulrich S, Künzli H, Meier-Peterhans D, et al. A smartphone-based health care chatbot to promote self-management of chronic pain (SELMA): pilot randomized controlled trial. JMIR MHealth UHealth 2020; 8: e15806.

50. Kowatsch T, Schachner T, Harperink S, et al. Conversational agents as mediating social actors in chronic disease management involving health care professionals, patients, and family members: multisite single-arm feasibility study. J Med Internet Res 2021; 23: e25060.

51. Stasinaki A, Büchter D, Shih C-HI, et al. Effects of a novel mobile health intervention compared to a multi-component behaviour changing program on body mass index, physical capacities and stress parameters in adolescents with obesity: a randomized controlled trial. BMC Pediatr 2021; 21: 308.

52. Chen T, Chen L, Li S, et al. Virtual digital psychotherapist app-based treatment in patients with methamphetamine use disorder (Echo-APP): single-arm pilot feasibility and efficacy study. JMIR MHealth UHealth 2023; 11: e40373.

53. He Y, Yang L, Zhu X, et al. Mental health chatbot for young adults with depressive symptoms during the COVID-19 pandemic: single-blind, three-arm randomized controlled trial. J Med Internet Res 2022; 24: e40719.

54. Liu H, Peng H, Song X, et al. Using AI chatbots to provide self-help depression interventions for university students: a randomized trial of effectiveness. Internet Interv 2022; 27: 100495.

55. Baptista S, Wadley G, Bird D, et al. Acceptability of an embodied conversational agent for type 2 diabetes self-management education and support via a smartphone app: mixed methods study. JMIR MHealth UHealth 2020; 8: e17038.

56. Gong E, Baptista S, Russell A, et al. My diabetes coach, a mobile app-based interactive conversational agent to support type 2 diabetes self-management: randomized effectiveness-implementation trial. J Med Internet Res 2020; 22: e20322.

57. Bibault J-E, Chaix B, Guillemassé A, et al. A chatbot versus physicians to provide information for patients with breast cancer: blind, randomized controlled noninferiority trial. J Med Internet Res 2019; 21: e15787.

58. Chaix B, Bibault J-E, Pienkowski A, et al. When chatbots meet patients: one-year prospective study of conversations between patients with breast cancer and a chatbot. JMIR Cancer 2019; 5: e12856.

59. Carlin A, Logue C, Flynn J, et al. Development and feasibility of a family-based health behavior intervention using intelligent personal assistants: randomized controlled trial. JMIR Form Res 2021; 5: e17501.

60. Gaffney H, Mansell W, Tai S. Agents of change: understanding the therapeutic processes associated with the helpfulness of therapy for mental health problems with relational agent MYLO. Digit Health 2020; 6: 2055207620911580.

61. Barbaric A, Munteanu C, Ross H, et al. A voice app design for heart failure self-management: proof-of-concept implementation study. JMIR Form Res 2022; 6: e40021.

62. Anan T, Kajiki S, Oka H, et al. Effects of an artificial intelligence-assisted health program on workers with neck/shoulder pain/stiffness and low back pain: randomized controlled trial. JMIR MHealth UHealth 2021; 9: e27535.

63. Dhinagaran DA, Sathish T, Soong A, et al. Conversational agent for healthy lifestyle behavior change: web-based feasibility study. JMIR Form Res 2021; 5: e27956.

64. Alghamdi E, Alnanih R. Chatbot design for a healthy life to celiac patients: a study according to a new behavior change model. Int J Adv Comput Sci Appl 12. Epub ahead of print 2021. DOI: 10.14569/IJACSA.2021.0121077.

65. Mash R, Schouw D, Fischer AE. Evaluating the implementation of the GREAT4Diabetes WhatsApp chatbot to educate people with type 2 diabetes during the COVID-19 pandemic: convergent mixed methods study. JMIR Diabetes 2022; 7: e37882.

66. Krishnakumar A, Verma R, Chawla R, et al. Evaluating glycemic control in patients of south Asian origin with type 2 diabetes using a digital therapeutic platform: analysis of real-world data. J Med Internet Res 2021; 23: e17908.

67. Baertsch M-A, Decker S, Probst L, et al. Convenient access to expert-reviewed health information via an Alexa voice assistant skill for patients with multiple myeloma: development study. JMIR Cancer 2022; 8: e35500.

68. Bassi G, Giuliano C, Perinelli A, et al. A virtual coach (Motibot) for supporting healthy coping strategies among adults with diabetes: proof-of-concept study. JMIR Hum Factors 2022; 9: e32211.

69. Sagstad MH, Morken N-H, Lund A, et al. Quantitative user data from a chatbot developed for women with gestational diabetes Mellitus: observational study. JMIR Form Res 2022; 6: e28091.

70. Roca S, Lozano ML, García J, et al. Validation of a virtual assistant for improving medication adherence in patients with comorbid type 2 diabetes mellitus and depressive disorder. Int J Environ Res Public Health 2021; 18: 12056.

71. Klaassen R, Bul KCM, Op den Akker R, et al. Design and evaluation of a pervasive coaching and gamification platform for young diabetes patients. Sensors 2018; 18: 402.

72. Issom D-Z, Hardy-Dessources M-D, Romana M, et al. Toward a conversational agent to support the self-management of adults and young adults with sickle cell disease: usability and usefulness study. Front Digit Health 2021; 3: 600333.

73. Weir CB, Jan A. BMI classification percentile and cut off points. In: StatPearls. Treasure Island (FL): StatPearls Publishing, http://www.ncbi.nlm.nih.gov/books/NBK541070/ (2023, accessed 20 October 2023).

74. Kroenke K, Spitzer RL, Williams JBW. The PHQ-9. J Gen Intern Med 2001; 16: 606–613.

75. Kroeze W, Werkman A, Brug J. A systematic review of randomized trials on the effectiveness of computer-tailored education on physical activity and dietary behaviors. Ann Behav Med 2006; 31: 205–223.

76. Noar SM, Grant Harrington N, Van Stee SK, et al. Tailored health communication to change lifestyle behaviors. Am J Lifestyle Med 2011; 5: 112–122.

77. Peñarrieta MI, Flores-Barrios F, Gutiérrez-Gómez T, et al. Self-management and family support in chronic diseases. J Nurs Educ Pract 2015; 5: 73.26504502

78. Fisher EB, Boothroyd RI, Elstad EA, et al. Peer support of complex health behaviors in prevention and disease management with special reference to diabetes: systematic reviews. Clin Diabetes Endocrinol 2017; 3: 4.

79. Bickmore TW. Relational agents: effecting change through human-computer relationships . Thesis, Massachusetts Institute of Technology, https://dspace.mit.edu/handle/1721.1/36109 (2003, accessed 28 July 2023).

80. Cassell J. Embodied conversational interface agents. Commun ACM 2000; 43: 70–78.

81. Allouch M, Azaria A, Azoulay R. Conversational agents: goals, technologies, vision and challenges. Sensors 2021; 21: 8448.

82. McKay FH, Wright A, Shill J, et al. Using health and well-being apps for behavior change: a systematic search and rating of apps. JMIR MHealth UHealth 2019; 7: e11926.

83. Torgrimson BN, Minson CT. Sex and gender: what is the difference? J Appl Physiol 2005; 99: 785–787.

Associated Data

Supplementary Materials

sj-docx-1-dhj-10.1177_20552076241277693 - Supplemental material for Exploring the characteristics of conversational agents in chronic disease management interventions: A scoping review

Supplemental material, sj-docx-1-dhj-10.1177_20552076241277693 for Exploring the characteristics of conversational agents in chronic disease management interventions: A scoping review by Ekaterina Uetova, Lucy Hederman, Robert Ross and Dympna O’Sullivan in DIGITAL HEALTH