A Field-Based Predictive Model for Evidence-Based Classification in Male Footballers With Cerebral Palsy

Over a 6-year research period, the classification of footballers with cerebral palsy (CP) has identified the best activity limitation tests that best capture the influence of eligible impairments on performance, leading to the implementation of a new classification system (i.e., FT1-FT3 sport classes). This study aimed to (i) assess the sensitivity of these tests to discriminate among sport classes; and (ii) estimate predictors of activity limitation outcomes to enhance accuracy in class allocation. A total of 177 international-level male CP footballers (25.90 ± 6.13 years.; 67.70 ± 11.49 kg; 173.90 ± 8.07 cm; 156 outfield players and 21 goalkeepers) completed 11 field-based tests assessing activity limitation, including balance, coordination, jumping, change of direction ability, and dribbling capacity, and a 2 versus 2 small-sided game. Classification and regression tree analysis were used to develop predictive models of sport classes with probabilities calculated at each node, considering a final number of 7 tests. Significant differences were observed between sport classes in all variables, except in one-leg stance acceleration. Specific activity limitation test outcomes for dribbling skills, inter-limb coordination, and static balance predict the sport class allocation with a success rate of 72.1% for a valid sample of 111 CP footballers, and 75.2% for a subsample of 101 players excluding goalkeepers. The dribbling speed test is relevant for classifying those with severe (FT1) and moderate forms of impairment (FT2), while the one-leg stance test better predicts the class allocation of those with less impairment (FT3). This study provides empirical support for classifiers' decision-making and suggests distinct patterns of sport-specific activity limitations across current CP football sport classes.