Hybrid time series and machine learning models for forecasting cardiovascular mortality in India: an age specific analysis

Abstract Cardiovascular disease (CVD) is a primary cause of death in India, accounting for a significant portion of the global CVD burden. This study looks at statistics on heart disease mortality from the Institute for Health Metrics and Evaluation (IHME) from 1990 to 2021, divided into five age gr...

Full description

Saved in:
Bibliographic Details
Main Authors: M Darshan Teja, G Mokesh Rayalu
Format: Article
Language:English
Published: BMC 2025-06-01
Series:BMC Public Health
Subjects:
Online Access:https://doi.org/10.1186/s12889-025-23318-7
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Abstract Cardiovascular disease (CVD) is a primary cause of death in India, accounting for a significant portion of the global CVD burden. This study looks at statistics on heart disease mortality from the Institute for Health Metrics and Evaluation (IHME) from 1990 to 2021, divided into five age groups: 0–5, 6–15, 16–49, 50–69, and 70 + . We used both classic ARIMA and hybrid models that combined ARIMA with machine learning techniques such as Random Forest, Support Vector Machine (SVM), XGBoost, and GARCH to anticipate mortality trends. Model performance was assessed using the Root Mean Square Error (RMSE) and Mean Absolute Percentage Error (MAPE). Across several age groups, the ARIMA + SVM model outperformed standalone ARIMA in terms of accuracy, with RMSE improvements of up to 15.6%. The 70 + population has the greatest mortality rates, highlighting the urgent need for focused healthcare treatments. These hybrid models are valuable tools for healthcare legislators in developing preventative programs, allocating resources effectively, and prioritizing treatment for high-risk age groups, especially the elderly, since they improve forecasting accuracy and offer interpretive insights. Given India's growing cardiovascular disease load, our results highlight how predictive analytics may support data-driven public health planning.
ISSN:1471-2458