Automatic Prediction of Amyotrophic Lateral Sclerosis Progression using Longitudinal Speech Transformer

TL;DR

This paper introduces ALST, a neural network model that predicts ALS progression from longitudinal speech data, achieving high accuracy and interpretability, thus offering an efficient alternative to manual assessments.

Contribution

The paper presents a novel neural network model that leverages pretrained speech features and longitudinal data for accurate ALS progression prediction.

Findings

Achieved 91.0% AUC on ALS TDI dataset.

Improved prediction accuracy by 5.6% over previous models.

Demonstrated interpretability in distinguishing severe ALS cases.

Abstract

Automatic prediction of amyotrophic lateral sclerosis (ALS) disease progression provides a more efficient and objective alternative than manual approaches. We propose ALS longitudinal speech transformer (ALST), a neural network-based automatic predictor of ALS disease progression from longitudinal speech recordings of ALS patients. By taking advantage of high-quality pretrained speech features and longitudinal information in the recordings, our best model achieves 91.0\% AUC, improving upon the previous best model by 5.6\% relative on the ALS TDI dataset. Careful analysis reveals that ALST is capable of fine-grained and interpretable predictions of ALS progression, especially for distinguishing between rarer and more severe cases. Code is publicly available.