Smart Mix Saves Lives: New AI Helps Stop Bleeding Mistakes

Anticoagulant drugs carry major risks of severe bleeding. These blood-thinning medicines cause frequent prescription errors. Despite existing digital decision support tools, doctors often get swamped by too many alerts. This overload means vital warnings get ignored. Scientists wanted a smarter approach to keep patients safe.

The team built a hybrid clinical decision support system. It blends traditional rule-based logic with modern machine learning. The system scans every anticoagulant prescription. It marks orders that need pharmacist attention. Testing happened first at one site, then at two others for external checks.

Training data came from over 75 thousand prescriptions across three hospitals. The finished model has custom rules and drug-interaction checks. A CatBoost classifier helps rank which alerts matter most. Internal tests showed most alerts were accurate. Outside validation proved the system worked well at other locations. The result suggests this tool could become a practical aid for pharmacy teams.

actions