Predicting Inpatient Medication Orders From Electronic Health Record Data.

Area of Science:

• Medical Informatics
• Machine Learning in Healthcare
• Clinical Decision Support

Background:

• Electronic health records (EHRs) contain vast amounts of patient data.
• Predicting medication orders can optimize treatment and reduce errors.

Purpose of the Study:

• To develop and evaluate machine learning models for predicting patient-specific medication orders.
• To assess the performance of deep learning and logistic regression models against a baseline.

Main Methods:

• Trained deep learning sequence and logistic regression models on over 3 million medication orders.
• Utilized structured EHR data for prediction of 990 possible medications.
• Compared model performance against a baseline of frequently ordered medications.

Main Results:

• The sequence model ranked 55% of ordered medications in its top-10 predictions (logistic model: 49%).
• 75% of ordered medications were in the sequence model's top-25 predictions (logistic model: 69%).
• 93% of the sequence model's top-10 predictions included a medication ordered within the next day.

Conclusions:

• Machine learning models can effectively predict medication orders using EHR data.
• These models show potential for enhancing clinical workflows and patient care.
• Predictive analytics in healthcare can leverage EHR information for improved outcomes.