Reliable Objective Assessment of Friedreich Ataxia Through Isolation Forest-Based Anomaly Detection

Area of Science:

• Biomedical Engineering
• Neurology
• Data Science

Background:

• Objective assessment of motor impairments in Friedreich Ataxia (FRDA) is crucial for disease monitoring and treatment.
• The Ataxia Instrumented Measurement Cup (AIM-C) offers precise motor function quantification but is susceptible to data anomalies.
• Data anomalies can compromise the reliability of FRDA severity predictions derived from AIM-C measurements.

Purpose of the Study:

• To develop a robust anomaly detection framework for AIM-C data to improve the reliability of Friedreich Ataxia assessments.
• To enhance signal quality by implementing an adaptive median filtering approach guided by anomaly scores.
• To validate the improved accuracy and consistency of motor impairment severity predictions in FRDA.

Main Methods:

• Implemented an anomaly detection framework utilizing the Isolation Forest algorithm to identify and exclude anomalous data segments.
• Developed an adaptive median filtering technique that dynamically adjusts denoising based on anomaly scores.
• Evaluated the impact of anomaly detection and filtering on the reliability of kinematic and kinetic features extracted from AIM-C data.

Main Results:

• The proposed framework significantly improved the reliability of extracted kinematic and kinetic features from AIM-C data.
• Anomaly detection and adaptive filtering led to higher inter-class correlation (ICC) values for motor impairment severity predictions.
• The optimized AIM-C data analysis enhances the consistency and accuracy of objective assessments in Friedreich Ataxia.

Conclusions:

• The integration of anomaly detection and signal optimization techniques significantly enhances the reliability of the AIM-C device for FRDA assessment.
• This approach provides a more consistent and accurate tool for clinicians and researchers to monitor disease progression and treatment efficacy.
• The findings support the use of AIM-C, augmented with advanced data processing, as a valuable tool in Friedreich Ataxia research and clinical practice.