Artificial Intelligence Model for Cardiovascular Incidence Prediction: A Practical Deep Learning Approach Based on Isfahan Cohort Study
محل انتشار: اولین کنگره بین المللی هوش مصنوعی در علوم پزشکی
سال انتشار: 1402
نوع سند: مقاله کنفرانسی
زبان: انگلیسی
مشاهده: 176
نسخه کامل این مقاله ارائه نشده است و در دسترس نمی باشد
- صدور گواهی نمایه سازی
- من نویسنده این مقاله هستم
این مقاله در بخشهای موضوعی زیر دسته بندی شده است:
استخراج به نرم افزارهای پژوهشی:
شناسه ملی سند علمی:
AIMS01_366
تاریخ نمایه سازی: 1 مرداد 1402
چکیده مقاله:
Background: Over the last decade, the dramatic improvements in Artificial Intelligence (AI)and Machine Learning (ML) approaches have had remarkable achievements in CardiovascularDisease (CVD) predictions. However, Studies have indicated that recent Artificial Intelligencedevelopments such as Deep Learning strategies could be more effective than both classic ML andclassic statistical models in estimating the risk of CVD incidence in large-scale cohort studies. Infact, Deep Learning approaches are much more efficient in dealing with high dimensional largedatasets. Thus, this cutting-edge technology is capable of considering various factors includingparameters with indirect impacts on development of cardiovascular diseases.Method: By utilizing of Deep Learning algorithm and a Deep Neural Network (DNN), we developeda risk prediction model for the incidence of cardiovascular diseases based on a ۱۳-yeardataset garnered from Isfahan Cohort Study (ICS). ICS is an ongoing population-based longitudinalcohort study performed in central areas of Iran, starting from ۲۰۰۱. The primary aim of ICSwas to evaluate the risk factors of CVD including myocardial infarction, stroke, unstable anginaand sudden cardiac death in a large Iranian population. The learning data frame current modelincludes ۵۸ diverse variables containing information about clinical, laboratory, socio-economic,demographic and lifestyle components of each participant.Results: Among ۶۵۰۴ participants at the baseline, ۴۳۷۰ individuals were chosen with no historyof CVD background. The target population and had was completely followed up data up until theend of ICS’s first phase and of the ICS. during the period, ۴۱۳ CVD events were recorded. TheANN proposed model fully connected neural network demonstrated considerable precision inpredicting ۱۳ years CVD incidences with accuracy classification score of ۹۰.۵۲% (the accuracyscore has been used as a metric and it is calculated by dividing the number of correct predictionsby the total prediction number).Conclusion: In this study we developed a practical risk assessment model for predicting the incidenceof CVD in Iranian population using Deep Learning and AI algorithms.
کلیدواژه ها:
نویسندگان
Hamed Etesampour
Heart Failure Research center, Isfahan Cardiovascular Research Institute, Isfahan, Iran
Mohammad Fakhrolmobasheri
Heart Failure Research center, Isfahan Cardiovascular Research Institute, Isfahan, Iran
Farzad Parvaresh
Heart Failure Research center, Isfahan Cardiovascular Research Institute, Isfahan, Iran
Davood Shafie
Heart Failure Research center, Isfahan Cardiovascular Research Institute, Isfahan, Iran
Sayed Jalal Zahabi
Heart Failure Research center, Isfahan Cardiovascular Research Institute, Isfahan, Iran