# Comparison of pharmacist evaluation of medication orders with predictions of a machine learning model

**Authors:** Sophie-Camille Hogue, Flora Chen, Geneviève Brassard, Denis Lebel, Jean-François Bussières, Audrey Durand, Maxime Thibault  
**arXiv:** [2011.01925](https://arxiv.org/abs/2011.01925)  
**Date:** 2020-11-03  
**Summary page:** https://www.summarizepaper.com/en/paper/2011.01925v1/

## Key points

- Objective: Evaluate clinical performance of unsupervised machine learning model for identifying unusual medication orders and pharmacological profiles
- Study conducted between April 2020 and August 2020
- Involving 25 clinical pharmacists rating 12,471 medication orders and 1,356 pharmacological profiles as typical or atypical
- Model's performance assessed using area under precision-recall curve (AUPR)
- Model's performance poor for medication orders but satisfactory for pharmacological profiles
- Despite limitations, considered a useful screening tool by pharmacists
- Provides insights into application of machine learning models in clinical settings
- Highlights areas for improvement in identifying atypical medication orders

## Summary

The objective of this study was to evaluate the clinical performance of an unsupervised machine learning model designed to identify unusual medication orders and pharmacological profiles. A prospective study was conducted between April 2020 and August 2020 involving 25 clinical pharmacists who rated 12,471 medication orders and 1,356 pharmacological profiles as either typical or atypical. The performance of the model was assessed using the area under the precision-recall curve (AUPR). The results showed that the model's performance was poor for medication orders but satisfactory for pharmacological profiles. Despite its limitations in accurately identifying unusual medication orders, the pharmacists considered it to be a useful screening tool. This research provides valuable insights into the application of machine learning models in clinical settings and highlights areas for improvement in identifying atypical medication orders.