A primary estimation of the cardiometabolic risk by using artificial neural networks

Abstract

Estimation of the cardiometabolic risk (CMR) has a leading role in the early prevention of atherosclerosis and cardiovascular diseases. The CMR estimation can be separated into two parts: primary estimation (PE-CMR) that includes easily-obtained, non-invasive and low-cost diagnostic methods and secondary estimation (SE-CMR) involving complex, invasive and/or expensive diagnostic methods. This paper presents a PE-CMR solution based on artificial neural networks (ANN) as it would be of great interest to develop a procedure for PE-CMR that would save time and money by extracting the persons with potentially higher CMR and conducting complete SE-CMR tests only on them. ANN inputs are values obtained by using PE-CMR methods, i.e. primary risk factors: gender, age, waist-to-height ratio, body mass index, systolic and diastolic blood pressures. ANN output is cmr-coefficient obtained from the number of disturbances in biochemical indicators, i.e. secondary risk factors: HDL-, LDL- and total cholesterol, triglycerides, glycemia, fibrinogen and uric acid. ANN training and testing are done by dataset that includes 1281 persons. The accuracy of our solution is 82.76%.