Objective Assessment of Friedreich Ataxia in Children: Accounting for Developmental Deficits.

The immature nervous system of children creates challenges for assessing the severity of Friedreich's Ataxia (FRDA). Erratic movements and postural sway from a maturing nervous system are difficult to disentangle from the impact of FRDA yet both are reflected in clinical rating score. To address this issue, we propose a novel correction framework for scores of severity derived from regression models trained on Inertial Measurement Unit (IMU) data. These are collected with three specialized devices: a cup-shaped IMU (Ataxia Instrumented Measure (AIM)-C), a spoon-shaped IMU (AIM-S), and a pendant-shaped IMU (AIM-P). A tailored algorithm applied to the data from each device is designed to isolate and remove developmental effects: a reinforcement learning-based technique for AIM-C, Bayesian optimization for AIM-S, and a multi-layer perceptron framework for AIM-P. Using these correction methods, ataxia severity scores of the movement deficit related to FRDA alone were produced, which were free from development related confounders. These scores provide greater precision for clinicians when assessing FRDA progression in children. This work demonstrates the capacity of advanced signal processing and machine learning techniques to enhance the clinical utility of severity assessments in paediatric FRDA participants.Clinical relevance- By accounting for developmental effects, clinicians can more accurately measure the ataxia in Friedreich ataxia and more reliably track disease progression and evaluate treatment efficacy.