THE ROLE OF HMG COA REDUCTASE INHIBITORS ON THE PROGRESSION OF CORONARY ARTERY DISEASE: FOCUS ON PREDICTION MODEL

Abstract

Currently, an integrated site-specific and patient-specific comprehensive predictive model of plaque progression in various CVD is not available. In this study, we considered medical records of 256 patients obtained within the EU H2020 SMARTool project which is carefully designed to collect the features from various domains relevant for disease which are used in everyday clinical practice. The database contains detailed information of patients with suspected CAD disease regarding the clinical status, risk factors, routine blood analyses, CAD morphology and progression and current therapy. Results showed that there was statistically significant difference of values of this parameter for the SMARTool patients with and without disease progression, measured at the follow-up, F(1,250)=33.39, p < 0.001, while the CAD Score in the follow-up is significantly different from the score measured at the initial time point, F(1,254)=76.244, p < 0.001. The significant interaction of statins is achieved with aspirin F(1,252)= 3.921, p=0.049, while interactions with other medicaments are insignificant for CAD Score. The results showed that there was no significant interaction of statins and dyslipidemia, F(1,251)=0.877, p = 0.350. Also, there was no significant interaction of statins and hypertension, F(1,245)=0.283, p=0.596. The CAD score in the baseline was significantly different among patients who were further prescribed with therapy than those who were not, and this trend remained unchanged after a given time period, i.e. those patients who were at risk had progression in addition to statins, but the combination of statins and aspirin was shown as effective in decreasing the CAD Score. The Random Forest classifier applied on 24 selected features is the most reliable among all tested ML algorithms for the prediction of CAD progress.