This Patient Looks Like That Patient: Prototypical Networks for Interpretable Diagnosis Prediction from Clinical Text

TL;DR

ProtoPatient is a new deep learning approach using prototypical networks and attention mechanisms to provide accurate and interpretable diagnosis predictions from clinical text, aiding doctors with understandable justifications.

Contribution

The paper introduces ProtoPatient, a novel model combining prototypical networks and label-wise attention for interpretable clinical diagnosis prediction from text.

Findings

Outperforms existing baseline models on clinical datasets

Provides valuable, understandable explanations for diagnoses

Receives positive qualitative feedback from medical doctors

Abstract

The use of deep neural models for diagnosis prediction from clinical text has shown promising results. However, in clinical practice such models must not only be accurate, but provide doctors with interpretable and helpful results. We introduce ProtoPatient, a novel method based on prototypical networks and label-wise attention with both of these abilities. ProtoPatient makes predictions based on parts of the text that are similar to prototypical patients - providing justifications that doctors understand. We evaluate the model on two publicly available clinical datasets and show that it outperforms existing baselines. Quantitative and qualitative evaluations with medical doctors further demonstrate that the model provides valuable explanations for clinical decision support.