Confidence Estimation for Automatic Detection of Depression and Alzheimer's Disease Based on Clinical Interviews

TL;DR

This paper presents a Bayesian confidence estimation method for automatic detection of Alzheimer's and depression from clinical interviews, improving trustworthiness and accuracy of diagnoses.

Contribution

A novel Bayesian approach with dynamic Dirichlet prior for confidence estimation in speech-based disease detection is introduced.

Findings

Outperforms baseline methods in classification accuracy

Enhances confidence estimation reliability

Effective on ADReSS and DAIC-WOZ datasets

Abstract

Speech-based automatic detection of Alzheimer's disease (AD) and depression has attracted increased attention. Confidence estimation is crucial for a trust-worthy automatic diagnostic system which informs the clinician about the confidence of model predictions and helps reduce the risk of misdiagnosis. This paper investigates confidence estimation for automatic detection of AD and depression based on clinical interviews. A novel Bayesian approach is proposed which uses a dynamic Dirichlet prior distribution to model the second-order probability of the predictive distribution. Experimental results on the publicly available ADReSS and DAIC-WOZ datasets demonstrate that the proposed method outperforms a range of baselines for both classification accuracy and confidence estimation.