Machine Learning Improves Functional Upper Extremity Use Capture in Distal Radius Fracture Patients

Area of Science:

• Orthopedics
• Biomedical Engineering
• Rehabilitation Science

Background:

• Current outcome measures for distal radius fracture (DRF) repair, such as strength testing and patient-reported outcomes (PROs), may lack the precision to fully capture functional upper extremity (UE) use.
• Accelerometry offers potential for objective UE activity assessment, but existing methods struggle to differentiate between functional and nonfunctional movements.

Purpose of the Study:

• To assess the accuracy of machine learning (ML) algorithms in identifying functional UE movements using accelerometry data in patients post-DRF repair.
• To explore the potential of ML-driven accelerometry for enhancing the evaluation of UE function during recovery from DRF.

Main Methods:

• A prospective study involving six patients who underwent open reduction and internal fixation (ORIF) for DRF.
• Patients performed standardized activities while wearing a wrist accelerometer; data were analyzed using a novel ML algorithm and validated against visual inspection of videotaped movements.

Main Results:

• The ML algorithm achieved high accuracy in predicting functional UE movements from accelerometry data.
• Within-subject modeling demonstrated 90.4% ± 3.6% accuracy, while between-subject modeling showed 79.8% ± 8.9% accuracy.
• The ML analysis successfully captured functional UE activity in patients after DRF ORIF.

Conclusions:

• ML analysis of accelerometry data can accurately differentiate functional UE use in patients recovering from DRF.
• This approach offers a promising tool to improve the understanding of UE functional recovery and enhance postoperative rehabilitation protocols for DRF patients.