Comparison of Gender Prediction Accuracy between Regression Models Derived from Hand, Foot Measurements and Long Bone Measurements in a Sample of Kolhapur Population

Abstract

Introduction: Here, we aim to compare the accuracy of the regression formula derived to predict genderusing data on measurements of foot and hand with the formula derived to predict gender using the lengthmeasurements data of the long bones (tibia and ulna).Methods: Patients attending the outpatient services at the Orthopedic Department, and between the agerange 18 to 50 years were recruited (n=1000; 500 males and 500 females). Subjects suffering from any kindof bone deformity were excluded. Vallois method was used to estimate the measurements of hand, foot,tibia, and ulna. Regression formulas were obtained from the hand, foot-long bones measurements; to predictgender, using multiple logistic regression.Results: Differences between male’s and female’s measurements of hand (P<0.001), long bones (P<0.001),and foot (P<0.001) were significant. The accuracy of the model used to predict gender, which was calculatedfrom the dimensions of hand and foot was 81.5%. The accuracy of the model used to predict gender calculatedfrom the long bone measurement was 78.3%.Conclusions: Dimensions of hand and foot are a better predictor (81.5%) of gender vs the length of longbones (tibia and ulna) (78.3%) in the Kolhapur population.