A Novel Deep Learning Approach for Forecasting Myocardial Infarction Occurrences with Time Series Patient Data

Area of Science:

• Cardiology and Public Health
• Artificial Intelligence in Healthcare
• Time Series Forecasting

Background:

• Myocardial Infarction (MI) is a critical global health issue, with significant impact in the Chittagong Metropolitan Area.
• Challenges in MI prevention and treatment are exacerbated by an aging population and complex healthcare demands.
• Early warning systems are vital for timely epidemic management and resource allocation.

Purpose of the Study:

• To develop and evaluate a deep learning model for early forecasting of Myocardial Infarction incidence.
• To predict the prevalence of heart attacks using a novel dataset from the Chittagong Metropolitan Area.
• To improve public health strategies through accurate MI prediction.

Main Methods:

• Utilized a novel dataset of daily heart attack incidence Time Series Patient Data (January 2020 - December 2021).
• Compared traditional models (ARIMA, ETS, TBATS) and LSTM with a proposed Myocardial Sequence Classification (MSC)-LSTM model.
• Employed deep learning techniques for time series analysis and prediction.

Main Results:

• The novel MSC-LSTM model demonstrated superior performance in forecasting MI incidence.
• Achieved a minimum Mean Percentage Error (MPE) of 1.6477, outperforming other evaluated models.
• The model provides accurate predictions for future heart attack occurrences.

Conclusions:

• The MSC-LSTM model offers a promising tool for early detection and prediction of Myocardial Infarction.
• Accurate MI forecasting supports enhanced resource allocation, policy development, and public health preparedness.
• This research contributes to proactive management of heart attack burden in the Chittagong Metropolitan Area and beyond.