Artificial Neural Network for Classification of Possible Cardiovascular Risk Using Indexes of Heart Rate Variability

Prevention is a key factor to avoid chronic diseases and premature death. The assessment of risk of cardiovascular disease through the Framingham score could help in taking action on time. This paper proposes the use of an Artificial Neural Network as a classifier of risk based on the indexes of Heart Rate Variability. 60 electrocardiographic records from the database of the PhysioBC® project are used to calculate time domain, frequency domain and nonlinear indexes. These parameters, in addition to age and body mass index, will be used to classify 4 levels of risk. These levels are established using the Framingham score. The proposed architecture has a training efficiency of 91.7 %, 100 % using test vectors and 95 % with validation vectors.