Neuro fuzzy based ascvd risk prediction Model using non traditional image Markers

Abstract

Cardiovascular diseases (CVDs) are among the trending top newlinecauses for the high mortality rate in India. Early CV risk estimation and newlinepreventive measures can benefit to reduce CV events. India lacks an newlineethnic specific customized CV risk estimation model. The sub-ethnic newlinespecific studies with local data recommend the need for sub-ethnic newlinespecific CV risk calculator model for improved preventive measures. newlineThe currently used Traditional Risk Factor (TRF)-based models was newlinestudied to contribute substantial gap with under/over estimation in the newlineCV risk estimation. The ultrasound (US)-based non-Traditional Risk newlineFactors (non-TRF) markers with its merits of being non-invasive and newlineaffordable can update/improve the TRF-based CV risk estimation. This newlinechoice is suitable for research in healthy young, middle-aged adult newlinepopulation to develop a superior model design. The non-TRF based newlineCarotid artery image markers such as carotid intima-media thickness newline(cIMT) and carotid plaque (cP) is clinically recommended to improve newlinethe CV risk estimation beyond the TRF based risk estimation. newlineThis proposed research has developed an efficient, intelligent newlinehybrid Machine Learning (ML) framework for improved Kerala subethnic newlinespecific Atherosclerotic Cardiovascular (ASCV) risk prediction newlineand stratification by integrating the TRFs and US-based non-traditional newlinecarotid image markers. The methodology has used 17 TRF-based newlineclinical markers pertaining to 1029 subjects were collected and analysed newlineusing SPSS. In the second phase, early ASCV risk prediction and its newlineclassification was modifiedand#8223; by integrating the quantified cIMT and cP newlinescores with the TRFs values. Statistical evaluation and accuracy of the newlinevi newline modifiedand#8223; estimation was assessed using an ML-based Decision Tree newline(DT) model. newlineThe proposed final model has used an intelligent attribute newlinereduction, prediction, and classification implemented using the proposed newlineweight adapted SVMPSO-ANFISPSO-Multi-SVM and a SVMGWOANFISGWO- newlineMulti-SVM optimized framework. This proposed newlineframework consists of three main stage