Dipartimento di Scienze Statistiche

We investigate automatic classification procedures for the diagnosis of the Carpal Tunnel Syndrome, a disease frequently observed in occupational medicine. We apply different classification techniques to a medical data set of patients reporting the typical symptoms of this syndrome and exploit the predictive power of such data to classify subjects as “sick” or “healthy”, according to the information obtained from the electromyography and the ultrasound imaging tests. Particular attention is paid to the “Box-Clustering” methodology which, among the tested techniques, is the most recent one. We show that all the automatic classification methods have a comparable diagnostic performance but, in some cases, Box-Clustering performs better than the others. Even if for the diagnosis of Carpal Tunnel Syndrome electromyography cannot be completely replaced by ultrasound imaging, our results show that ultrasound scan can be a valuable screening tool to detect the pathology.