Factors that predict walking ability with a prosthesis in lower limb amputees.

Area of Science:

• Rehabilitation Medicine
• Biomedical Engineering
• Clinical Prediction Modeling

Background:

• Predicting walking ability post-lower limb amputation is vital for setting realistic rehabilitation goals.
• Identifying early predictive factors aids in tailoring patient-specific treatment plans.

Purpose of the Study:

• To investigate the predictive capability of initial rehabilitation variables for final walking ability.
• To utilize support vector machines (SVMs) for developing a predictive model of ambulation outcomes.

Main Methods:

• Retrospective clinical case series involving 263 lower limb amputees.
• Support vector machines (SVMs) were employed to build predictive models.
• A genetic algorithm optimized variable selection for model accuracy.

Main Results:

• Six SVM models were constructed using varying numbers of input variables (4 to 11).
• Model accuracy ranged from 72.5% to 82.5%.
• The most accurate model incorporated four variables: age, Functional Comorbidity Index (FCI), amputation level, and admission mobility.

Conclusions:

• A predictive model using age, FCI, amputation level, and admission mobility accurately forecasts ambulation ability in lower limb amputees.
• This model can guide clinical decision-making and rehabilitation planning for improved patient outcomes.