Computational Mapping of Drug–Disease Relationships for Geriatric Medication Safety Using Beers Criteria

Abstract

Introduction — Geriatric patients are particularly vulnerable to adverse drug events due to age-related physiological changes and polypharmacy. The Beers Criteria provides explicit guidance for identifying potentially inappropriate medications (PIMs) in older adults; however, its application is often performed manually and inconsistently across healthcare settings. This study aims to develop a computational framework for mapping drug–disease relationships to support systematic identification of PIMs for geriatric medication safety.

Methods — A computational cross-sectional knowledge-based analysis was conducted using the publicly available Indonesian Pharmaceutical Dataset. Drug entities were normalized and mapped against the 2023 Beers Criteria using rule-based string matching. Identified PIMs were categorized into geriatric risk domains, and drug–disease relationships were extracted to construct structured mappings. Descriptive statistical analysis was performed to quantify the proportion and distribution of PIMs within the dataset.

Results — From 1,984 unique drug entities analyzed, 143 medications (7.2%) were classified as potentially inappropriate for geriatric use. Anticholinergic burden (28.7%) and sedative/CNS depressant risk (23.1%) represented the largest PIM categories. Drug–disease mapping identified 1,126 relationship pairs, with neurological and cardiovascular therapeutic domains showing prominent clustering.

Conclusion — The proposed computational mapping framework demonstrates the feasibility of transforming structured pharmaceutical knowledge into a scalable geriatric medication safety model. This informatics-based approach supports transparent, rule-driven identification of PIMs and provides a foundation for future integration into digital clinical decision-support systems.