JMIR Publications

ABSTRACT

Background:

Anticoagulants are high-alert medications with substantial risk of serious bleeding, yet dosing and prescribing errors remain common. Although clinical decision support systems (CDSS) can mitigate these errors, their impact is constrained by alert fatigue and limited interpretability

Objective:

We developed and validated a hybrid CDSS that integrates rule-based logic with machine learning to improve the safe use of anticoagulants.

Methods:

This multicenter study used electronic health record data from three tertiary hospitals. Data from one hospital were used for system development and internal validation, and data from two additional hospitals were used for external validation. The study included patients prescribed anticoagulants. The hybrid CDSS combined a knowledge-based rule engine with a machine learning model trained to predict whether an anticoagulant prescription would require pharmacist intervention. The system was iteratively refined through pilot testing, internal validation, and external validation.

Results:

A total of 75,200 anticoagulant prescriptions were used for model development. The final hybrid CDSS comprised 38 patient-specific rules and 1,129 drug–drug interaction rules, combined with a CatBoost classifier for alert prioritization. During internal validation, 88 (18.9%) alerts were generated, all were technically correct, with 88.6% deemed clinically relevant, 86.4% were considered clinically useful, and 13.6% required pharmacist intervention. In external validation across two hospitals, alert rates ranged from 22.6% to 32.1%, with 18.8%-57.1% of alerts requiring pharmacist intervention. The hybrid CDSS showed strong discrimination (AUROC, 0.871-0.963). No false negatives were identified, indicating that all prescriptions requiring intervention were successfully detected.

Conclusions:

A hybrid CDSS integrating rule-based logic with machine learning demonstrated high technical accuracy, clinical relevance, and robust performance across multiple institutions, suggesting its potential as a practical tool for enhancing medication safety in anticoagulant therapy.