Toward Automated Detection of Biased Social Signals from the Content of Clinical Conversations

TL;DR

This study develops an automated pipeline using speech recognition and NLP to detect social signals in clinical conversations, revealing biases in provider communication towards different racial groups.

Contribution

It introduces a novel automated method for detecting social signals in clinical interactions and demonstrates its effectiveness and fairness in identifying racial biases.

Findings

Automated pipeline achieved 90.1% accuracy in predicting social signals.

Significant differences found in provider behaviors towards white and non-white patients.

Providers showed more warmth and engagement with white patients.

Abstract

Implicit bias can impede patient-provider interactions and lead to inequities in care. Raising awareness is key to reducing such bias, but its manifestations in the social dynamics of patient-provider communication are difficult to detect. In this study, we used automated speech recognition (ASR) and natural language processing (NLP) to identify social signals in patient-provider interactions. We built an automated pipeline to predict social signals from audio recordings of 782 primary care visits that achieved 90.1% average accuracy across codes, and exhibited fairness in its predictions for white and non-white patients. Applying this pipeline, we identified statistically significant differences in provider communication behavior toward white versus non-white patients. In particular, providers expressed more patient-centered behaviors towards white patients including more warmth, engagement, and attentiveness. Our study underscores the potential of automated tools in identifying subtle communication signals that may be linked with bias and impact healthcare quality and equity.