Assessment of corticospinal tract dysfunction and disease severity in amyotrophic lateral sclerosis

TL;DR

This study developed a machine learning model using neurophysiological data to accurately predict ALS disease severity, offering an objective tool for diagnosis and monitoring disease progression.

Contribution

It introduces a novel random forest model combining neurophysiological measurements to predict ALS severity with high accuracy, improving diagnostic objectivity.

Findings

Predicts ALS severity with 97% accuracy

Uses triple stimulation and TMS techniques

Provides a new objective diagnostic method

Abstract

The upper motor neuron dysfunction in amyotrophic lateral sclerosis was quantified using triple stimulation and more focal transcranial magnetic stimulation techniques that were developed to reduce recording variability. These measurements were combined with clinical and neurophysiological data to develop a novel random forest based supervised machine learning prediction model. This model was capable of predicting cross-sectional ALS disease severity as measured by the ALSFRSr scale with 97% overall accuracy and 99% precision. The machine learning model developed in this research provides a new, unique and objective diagnostic method for quantifying disease severity and identifying subtle changes in disease progression in ALS.