Development of a hip osteoarthritis index for gait quality assessment: a data-driven comparative study

Scientific evidence demonstrates a strong relation between gait abnormalities and osteoarthritis, highlighting the need for a gait quality index for rapid osteoarthritis monitoring. Using a dataset on hip osteoarthritis, including 80 healthy individuals and 106 patients classified by Kellgren-Lawrence (KL) grading scale with pre- and post-arthroplasty surgery data, we introduced a linear model and a Hip Osteoarthritis Index (HOI) based on two kinematic features: (1) hip-knee angular velocity area and (2) hip maximum angular velocity. Additionally, we employed multi-layer perceptron (MLP) and deep recurrent neural network (RNN) models to evaluate the efficacy of our linear model. For healthy-patient, affected-unaffected leg, and severity grades classifications the accuracy of all models ranged around [Formula: see text], [Formula: see text], and [Formula: see text], respectively, with no significant differences ([Formula: see text]). HOI is highly correlated with KL grades ([Formula: see text]) and affected-unaffected leg of patients ([Formula: see text]). HOI differences ([Formula: see text]) were significant for (1) healthy and patients (pre-/post-surgery and affected/unaffected legs), (2) pre- and post-surgery of affected legs across severity grades, and (3) affected and unaffected legs, pre-surgery. Post-surgery, the gait pattern became more symmetric, with no significant HOI difference between affected and unaffected legs ([Formula: see text]). The affected legs' HOI moves toward healthy individuals ([Formula: see text]), indicating a more natural gait. The comparable results of the models, along with significant HOI differences between healthy and patients, pre- and post-surgery, and affected and unaffected legs, as well as the high correlation of HOI with KL grades, suggest HOI as an ideal candidate for hip osteoarthritis assessment.