Statistical Pattern Recognition Techniques for Early Diagnosis of Diabetic Neuropathy by Posturographic Data

The goal of this chapter is to describe the use of statistical pattern recognition techniques in order to build a classification model for the early diagnosis of peripheral diabetic neuropathy. In particular, the authors present two experimental methodologies, based on linear discriminant analysis and Bayes vec- tor quantizer algorithms respectively. The former algorithm has demonstrated the best performance in distinguish between non-neuropathic and neuropathic patients, while the latter is able to build models that recognize the severity of the neuropathy.