Algorithm for Diagnosis of Metabolic Syndrome and Heart Failure Using CPET Biosignals via SVM and Wavelet Transforms

Early diagnosis of diseases is essential to avoid health complications and costs to the health system. For this purpose, algorithms have been widely used in the medical field to assist in the diagnosis of diseases. This work proposes an algorithm with a new approach to analyze biosignals from cardiopulmonary exercise testing (CPET) to identify metabolic syndrome (MS) and heart failure (HF). The algorithm uses the support vector machine (SVM) as a classification technique and wavelet transforms for extraction of the features. For training, CPET data from 30 volunteers were used, of which 15 are diagnosed with MS and 15 with HF. The SVM-L-W approach, which uses wavelet transforms, has been shown to have better accuracy (93%) compared to some other approaches found in the literature. In addition, the SVM-L-W algorithm can be applied to identify other diseases, and is also adaptable to modifications in order to obtain better performance, as suggested in future work to continue this research.