SHM Converge Scientific Abstract ePoster Gallery-Increasing Feedback on Residents' Clinical Reasoning Documentation Using Machine Learning

Pdf Summary

This project explored whether machine learning could help increase feedback on internal medicine residents’ clinical reasoning documentation. Residents often receive limited feedback because supervisors lack time and a shared framework for assessing note quality. The team had previously developed and validated a machine learning/natural language processing tool that classifies admission notes as high- or low-quality. In this pilot implementation, the tool was integrated into a dashboard that displayed note-quality results to residents.<br /><br />The intervention occurred during a two-week inpatient night block. Residents first received an introductory email explaining the shared mental model for high-quality documentation and how to use the dashboard. They could then review their data and submit action plans and feedback.<br /><br />In this small pilot of 12 residents, there was no statistically significant overall improvement in note quality from pre- to post-intervention. However, residents who viewed the dashboard more than once showed a trend toward improvement, and the correlation between number of dashboard views and improvement in note quality was moderately positive (Pearson r = 0.70). About 70% of residents accessed the dashboard at least once.<br /><br />Resident feedback suggested they wanted more specific guidance than a simple high- vs low-quality label. Their action plans focused on making differentials more explicit, including supporting evidence for diagnoses, and adding alternative diagnoses. Based on this feedback, the team created a faculty dashboard so educators can view resident data and provide more targeted feedback. The next step is to train faculty to use the tool more consistently, with the goal of improving dashboard use and note quality.

Asset Subtitle

nan

Meta Tag

Author List Verity Schaye, David Kudlowitz, Daniel Sartori, Jesse Rafel, Benedict Guzman, Ilan Reinstein, Yindalon Apinyanaphongs, Marina Marin

Category Research

Concept Clinical Reasoning Documentation

Concept Machine Learning

Concept Natural Language Processing

Concept Feedback

Concept Resident Notes

Distinguished Non-Finalist

Presenter Organization NYU School of Medicine

Presenting Author Verity Schaye

Track Technology in Hospital Medicine

Keywords

machine learning

natural language processing

clinical reasoning

residents

note quality

feedback dashboard

internal medicine

documentation

pilot study

faculty feedback

Clinical Reasoning Documentation

Machine Learning

Natural Language Processing

Feedback

Resident Notes