JMIR Publications

ABSTRACT

Background:

Amyotrophic lateral sclerosis (ALS) is a neurodegenerative disease with an active trial landscape that relies on the sensitivity of selected clinical trial endpoints. Traditional clinical outcome assessments perform well in trials, but lack strong psychometric properties and may not detect small but clinically meaningful disease progression. Digital health technologies (DHTs) offer a promising alternative for tracking ALS disease progression.

Objective:

This study assessed the feasibility of remote digital monitoring in ALS using a comprehensive battery of prescribed home-based assessments via smartphone, a wearable device, and a computer-based mouse-clicking task.

Methods:

Participants completed weekly remote assessments including motor and speech tasks via a smartphone application and a computer mouse-clicking task for 24 weeks. They also participated in three remote telephone visits in weeks 1, 13, and 25. Reliability, minimal detectable change, and correlations with self-reported ALS Functional Rating Scale Revised (ALSFRS-RSE) subdomain scores were calculated for eight features across the speech, fine motor, and gross motor smartphone application tasks and for all 32 features from the computer mouse-clicking task. Sensitivity to longitudinal change was assessed for the eight smartphone-derived features and for a representative subset of eight computer mouse-clicking features.

Results:

Forty-two participants (19 with ALS, 23 controls) completed 10,237 smartphone assessments and 459 computer mouse-clicking sessions. Baseline discriminative models differentiated ALS from controls with AUC values of 0.75–0.92. Digital measures correlated strongly with self-reported ALS Functional Rating Scale Revised (ALSFRS-RSE) subdomain scores. Both participants with ALS and controls demonstrated improvement in fine motor and speech measures, with the exception of non-dominant pegboard performance, which declined in the ALS group. Improvements were smaller in participants with ALS, leading to increasing group differences over time, although only one feature showed a statistically significant separation over the 24 weeks. Gait and balance performance declined in both groups, with greater but non-significant separation observed for balance measures.

Conclusions:

These findings support the feasibility of digital remote assessments in ALS, demonstrate the ability to discriminate between ALS and controls based on certain features collected from speech, fine and gross motor tasks, and in some cases quantify functional decline over time. Further research is necessary to explore the natural history of these features longitudinally in larger cohorts of participants with ALS over extended periods to enable their potential integration into clinical trials.