Detecting the Pre-Disease State of Single Sample Through the Change in Local Network Enrichment Level.

Background: In complex biological processes, there exists a tipping point (pre-disease state) when the system undergoes a sudden and dramatic shift to a contrasting state. Accurate detection of the pre-disease state is critical for preventive medicine. However, precise detection of the pre-disease state proves challenging due to the clinical single-sample problem. Methods: To address this challenge, in this study, we introduce a novel single-sample pre-disease state detection method based on the change in local network enrichment level. Results: We validated the proposed method on five independent real datasets, including one influenza virus infection time-course dataset and four tumor datasets. Experimental results confirmed that the proposed method can accurately identify the pre-disease state prior to overt disease onset. Further analysis verified key genes identified by the proposed method in pre-disease state are associated with viral infection and immune dysregulation for the influenza dataset, and tumor metastasis for the tumor datasets. Conclusions: These results demonstrate that this method is a robust and biologically interpretable tool for single-sample pre-disease state detection, with great potential for clinical translation in individualized preventive medicine.