Predicting and Explaining Hearing Aid Usage Using Encoder-Decoder with Attention Mechanism and SHAP
Qiqi Su, Eleftheria Iliadou
TL;DR
This work tackles the challenge of forecasting daily hearing aid usage while providing interpretable explanations for the drivers of usage. It introduces attn-ED, an encoder-decoder with self-attention, coupled with Kernel SHAP to predict future HAid hours and identify contributing factors from a synthetic longitudinal EVOTION-like dataset. Empirical results show that attn-ED outperforms a vanilla LSTM in both personalized and global forecasting, and SHAP identifies Usage as the primary predictor with other features contributing modestly, offering actionable insights for clinicians. The study demonstrates the practical potential of integrating explainable AI in a medical domain, while acknowledging limitations such as synthetic data and the need for objective XAI evaluation and real-world validation.
Abstract
It is essential to understand the personal, behavioral, environmental, and other factors that correlate with optimal hearing aid fitting and hearing aid users' experiences in order to improve hearing loss patient satisfaction and quality of life, as well as reduce societal and financial burdens. This work proposes a novel framework that uses Encoder-decoder with attention mechanism (attn-ED) for predicting future hearing aid usage and SHAP to explain the factors contributing to this prediction. It has been demonstrated in experiments that attn-ED performs well at predicting future hearing aid usage, and that SHAP can be utilized to calculate the contribution of different factors affecting hearing aid usage. This framework aims to establish confidence that AI models can be utilized in the medical domain with the use of XAI methods. Moreover, the proposed framework can also assist clinicians in determining the nature of interventions.