Developing an Accumulative Assessment System of Upper Extremity Motor Function in Patients With Stroke Using Deep...

Area of Science:

• Neuroscience
• Rehabilitation Medicine
• Medical Technology

Background:

• The Fugl-Meyer assessment for upper extremity (FMA-UE) is a standard for evaluating motor function post-stroke.
• The FMA-UE's lengthy administration time poses challenges for clinical feasibility.
• There is a need for more efficient assessment tools that retain diagnostic accuracy.

Purpose of the Study:

• To develop an accumulative assessment system of upper extremity motor function (AAS-UE).
• To enhance administrative efficiency of upper extremity motor function assessment in stroke patients.
• To ensure the AAS-UE retains sufficient psychometric properties compared to the FMA-UE.

Main Methods:

• Utilized secondary data from three prior studies encompassing FMA-UE datasets from subacute and chronic stroke individuals.
• Employed deep learning algorithms within the AAS-UE to personalize item selection based on patient history.
• Incorporated prior FMA-UE scores, assessment intervals, and stroke chronicity to predict scores and optimize item sets.

Main Results:

• The AAS-UE demonstrated excellent concurrent validity (r = 0.97-0.99) and high test-retest reliability (ICC = 0.96).
• Achieved low random measurement error (15.6% minimal detectable change) and good responsiveness at both group (SRM = 0.65-1.07) and individual levels (30.5%-53.2% significant improvement).
• Psychometric properties of the AAS-UE were comparable to the established FMA-UE.

Conclusions:

• The AAS-UE represents an innovative assessment method leveraging patient data for improved efficiency.
• The system effectively balances administrative efficiency with robust psychometric performance.
• The AAS-UE shows particular promise in improving individual-level responsiveness and minimizing measurement error.