A support method for the contextual interpretation of biomechanical data

In the clinical field, biomechanical data provided by advanced technical devices are still underexploited. Data analysis usually consists of extracting attributes or computing synthetic values from temporal data and exploiting them by means of a monovariable statistical method. This article proposes a method to support clinicians, especially those in orthopedics, in the contextual interpretation of biomechanical data. We propose to characterize temporal biomechanical data by means of fuzzy space-time windows and to induce fuzzy decision trees to map the biomechanical and clinical data related to patients. Then, we present a method for objectively explaining a given clinical characteristic of a particular patient; this method is derived using the fuzzy rule base generated from the trees and a satisfiability measure. We have applied our method to real data in order to provide an objective explanation of the subjective self-evaluation of the functional status of patients with a shoulder prosthesis, and evaluate it by means of the stratified tenfold cross validation method. The mean explanation rate-which corresponds to the mean proportion of the patients belonging to test sets whose functional state is explained by the proposed method-exceeds 80% for more than half of the decision trees, and exceeds 70% for 94% of the trees. By supporting clinicians during the biomechanical data interpretation process, our method helps them take the objective biomechanical measurements in the medical practice into account, particularly in orthopedics. It can also make subjective evaluations more objective by mapping subjective and objective data