Scoping Review of Computational Drug Repurposing Methods Using Electronic Health Records and Genomic Data
Abstract
Background: Computational drug repurposing increasingly combines electronic health records (EHRs), genomic data and other related omics resources, but the methods used for this purpose and particular datasets have not been fully assessed. Objective: To retrieve and characterise computational approaches that use EHR, genomic or related omics data to identify and prioritise medicines for drug repurposing. Eligibility criteria: We included original research articles, preprints and conference abstracts that described or applied computational methods for drug repurposing using eligible clinical, genomic or related omics data. Literature reviews were used for supplementary reference searching but were not included in the analysis. Publications not written in English were excluded. Sources of evidence: Embase, MEDLINE, Web of Science Core Collection and Epistemonikos were searched from database inception to 30 September 2025. References of relevant reviews were also examined.Charting and synthesis: One reviewer screened the records and extracted publication, data source, methodological, validation and accessibility information using a structured data-charting framework. Findings were summarised with counts, percentages, tables and graphical mapping of methods and data-source combinations. Results: After deduplication and screening, 117 documents were included. Genomic and transcriptomic data were the most common data types, while EHR-based studies commonly used observational causal inference or large-scale target-trial emulation approaches. Functional annotation and integration with drug-target databases were the most frequent methodological strategies. Independent validation of prioritised targets and complete open-source code availability were uncommon. Among studies reporting cohort location, Europe and North America were most frequently represented. Conclusions: Computational drug-repurposing research is expanding and increasingly integrates multiple data domains. Important gaps remain in geographical representation, validation, reproducibility and open-source analytical code availability.
Abstract as supplied by the indexing source. © the authors / publisher.