Time-Series & Sensors

AIM-M: Multiview IMU Assessment of Friedreich Ataxia

Enhancing the Objective Assessment of Friedreich Ataxia Severity: A Multiview IMU-Based Approach

Pearson r = 0.94

One severity score from three instrumented devices: cup, spoon and pendant, each seeing a different side of the disease.

The problem

Most measures assess limb movement and balance separately, but clinical trials need a single score that covers both.

The approach

A multiview model that fuses IMU data from all three Ataxia Instrumented Measures into one Aggregated IMU Multiview Score (AIM-M).

The outcome

Correlation with clinical severity rose from 0.83, 0.75 and 0.77 for the single devices to 0.94 for the combined score.

In detail

Developed a multiview framework that combines movement information from three Ataxia Instrumented Measures: AIM-C, AIM-S and AIM-P. While each device captures a different functional domain, the integrated model combines them into the AIM-M score, producing a stronger association with mFARS than the individual-device models. The study reported a Pearson correlation of 0.94 for the combined multiview score.

Built with

Source

More in Time-Series & Sensors

Next project: DTW-Based Outlier Removal for Reliable FRDA Measurement

By Kanishka Ranaweera, PhD, Machine Learning Engineer in Melbourne, Victoria, Australia. hello@kanishkabuilt.me