Discrimination between hereditary spastic paraplegia and cerebral palsy based on gait analysis data: A machine...

Area of Science:

• Neurology
• Biomedical Engineering
• Machine Learning

Background:

• Distinguishing hereditary spastic paraplegia (HSP) from spastic cerebral palsy (CP) based on clinical presentation remains challenging.
• Previous research utilized kinematic gait analysis, but lacked a comprehensive approach combining multiple gait and physical exam measures.
• This study addresses the need for improved diagnostic discrimination between these neurological conditions.

Purpose of the Study:

• To investigate the efficacy of a machine learning approach in differentiating HSP from bilateral spastic CP.
• To leverage clinical gait analysis data for improved diagnostic accuracy.
• To develop a tool that assists in the diagnosis of undifferentiated gait disorders.

Main Methods:

• A retrospective analysis of a gait database included 179 variables from 28 individuals with HSP and 678 individuals with bilateral spastic CP.
• Bayesian additive regression trees (BART) were employed for classification, utilizing a 10-fold cross-validation.
• A diagnostic probability cutoff threshold was established to balance classification errors.

Main Results:

• The machine learning algorithm achieved 91% specificity and 90% sensitivity in classifying HSP versus bilateral spastic CP.
• The model demonstrated high accuracy in distinguishing between the two diagnostic groups.
• Contingency tables were used to analyze predicted versus actual diagnoses.

Conclusions:

• A machine learning model utilizing clinical gait analysis data can effectively differentiate HSP from bilateral spastic CP.
• This algorithm can aid in evaluating individuals with gait disorders and suspected HSP.
• The findings support the use of this model to inform decisions regarding genetic testing for HSP.