Enhancing Accessibility to Primary Healthcare Centres through the Development and Validation of a Machine Learning-based Gravity Model: Strengthening Public Health Coverage

This paper introduces a novel approach, a Machine Learning-based Gravity Model, to address this issue by predicting and mitigating disparities in healthcare access based on geographical location, population distribution, transportation infrastructure, and healthcare workforce availability. The research focuses on the Chikkaballapur district in Karnataka, Southern India, as a proof of concept.The methodology involves developing an optimization model for healthcare facility locations by analyzing the geospatial and demographic characteristics of the district. The goal is to maximize accessibility to primary health units while minimizing the burden on these healthcare facilities. To achieve this, the study considers various factors from the demand side (village population, literacy rates, agricultural and non-agricultural workers, OPD and IPD data, among others), supply side (healthcare staff), and distance and travel (including travel friction). Composite indexes are constructed to represent demand, supply, and travel factors, contributing to the overall accessibility and burden calculations.The study classifies villages based on their accessibility scores, separating them into categories of poor, average, and high accessibility. Similarly, Primary Health Centers (PHCs) are classified as underutilized, average, or overutilized based on their burden scores. Using these classifications, the paper presents case scenarios and a rule-based algorithm to recommend actions, such as upgrading or downgrading PHCs and building new ones, aimed at improving accessibility and addressing burden disparities.In conclusion, the Gravity-Based Optimization Model offers a promising solution to the complex challenge of healthcare accessibility and facility optimization in India.