South Asian countries would account for more than half of global CVD burden. Acute Coronry syndrome is the fifth biggest cause of death and disability in this region. In Malaysia, 20-25 percent of hospital mortality were due to coronary artery disease.
ACS can be treated at any stage. Cardiac catheterization is a popular treatment. In-hospital cardiac catheterization lowers mortality following STEMI in high-risk patients. This is important in rural places when cardiology expertise is scarce. TIMI and GRACE are used to predict STEMI patient death. It is based on Western Caucasian data with limited Asian data. In Asians, MI occurs later, with greater diabetes, hypertension, and renal failure. Hence these risk scores my not reflect our diversity and only relate to select populations. TIMI and GRACE were also established using ACS patients who did not get reperfusion therapy or drug-eluting stents.
An improved algorithm that predicts STEMI patient mortality will enhance prognosis. Automated risk scoring systems can compensate for the information loss caused by conventional risk scores. Machine learning (ML) surpasses the traditional risk score model in post-STEMI mortality studies in China, Israel, and Korea.
For clinical care, especially in distant places with limited resources, identifying mortality risk factors among Asian populations is crucial.
A diversified Asian population has not been studied using ML algorithms to predict and identify cardiac catheterization risk. So a paradigm like this is essential in Southeast Asia. Hence We developed an online calcultor for this purpose.
The calculator predicts short and long term mortality in Asians. The online tool has a database for algorithm validation data archiving. The tool also determines mortality rate for patients undergoing cardiac catheterization. It is useful for early screening and resource allocation of high risk patients who will benefit from cardiac cauterization. A diverse Asian population has not been studied using ML algorithms to predict and identify cardiac catheterization risk. So a model like this is essential in Southeast Asia. Because the tool is online, it can be used in rural or urban hospitals where cardiologist expertise is limited, especially in identifying high-risk patients who benefit from cardiac catheterization.
The calculator is developed using the National Cardiovascular Disease Database for Malaysia registry, of a multi-ethnic, heterogeneous Asian population for in-hospital (6299 patients), 30-days (3130 patients), and 1-year (2939 patients). Hence the tool can be used for other Asian countries a well.
Support Vector Machine (SVM) learning algorithm was used for the algorithm development and compared to Thrombolysis in Myocardial Infarction (TIMI) score. Invasive management of varying degrees was selected as important variables that improved mortality prediction in Asian population.
Model performance were measured using area under the receiver operating characteristic curve (AUC) for SVM in-hospital, 30 days, and 1-year that outperformed TIMI risk score (AUC = 0.88 vs AUC = 0.81 ; AUC = 0.90 vs AUC = 0.80 ; AUC = 0.84 vs AUC = 0.76). TIMI score underestimates patients’ risk of mortality. 90% of non-survival patients are classified as high risk (>50%) by machine learning algorithm compared to 10-30% non-survival patients by TIMI. In a multi-ethnic population, machine learning classified STEMI patients better than TIMI scoring. Machine learning can identify distinct factors in Asian populations to improve mortality prediction.
Dr. Sorayya Malek - Senior Lecturer at Bioinformatics unit, Institute of Biological Sciences, Faculty of Science, Universiti Malaya, sorayya@um.edu.my
Prof. Sazzli Kasim - Consultant Cardiologist at UiTM Specialist Center.
Dr. Khairul Shafiq Inin Ibrahim - Cardiologist at UiTM Specialist Center.
Mr. Firdaus Aziz - Postgraduate student at Bioinformatics unit, Institute of Biological Sciences, Faculty of Science, Universiti Malaya.
Comments