Abstract
AbstractObjectivesThis paper aims to elaborate a decision tree for the early detection of adolescent swimmers at risk of presenting low bone mineral density (BMD), based on easily measurable fitness and performance variables.MethodsBone mineral status of 78 adolescent swimmers was determined using DXA scans at the hip and subtotal body. Participants also underwent physical fitness (upper and lower body strength, running speed and cardiovascular endurance) and performance (swimming history, speed and ranking) assessments. A gradient boosting machine regression tree was built in order to predict BMD of the swimmers and to further develop a simpler individual decision tree, using a subtotal BMD height-adjusted Z-score of −1 as threshold value.ResultsThe predicted BMD using the gradient boosted model was strongly correlated with the actual BMD values obtained from DXA (r=0.960, p<0.0001) with a root mean squared error of 0.034 g/cm2. According to a simple decision tree, that showed a 73.9% of classification accuracy, swimmers with a body mass index (BMI) lower than 17 kg/m2 or a handgrip strength inferior to 43kg with the sum of both arms could be at higher risk of having low BMD.ConclusionEasily measurable fitness variables (BMI and handgrip strength) could be used for the early detection of adolescent swimmers at risk of suffering from low BMD. The presented decision tree could be used in training settings to determine the necessity of further BMD assessments.Summary boxWhat are the new findings?Adolescent swimmers with a low BMI or handgrip strength seem more likely to be at higher risk of having low BMD.Subtotal BMD values predicted from our regression model are strongly correlated with DXA measurements.How might it impact on clinical practice in the futureHealthcare professionals could easily detect adolescent swimmers in need of a DXA scan.The computer-based regression tree could be included in low BMD management and screening strategies.
Publisher
Cold Spring Harbor Laboratory