Profiling patterns in healthcare using an ensemble model framework to predict employee health risks

Chan, Nicholas Kin Whai (2024) Profiling patterns in healthcare using an ensemble model framework to predict employee health risks. Masters thesis, Sunway University.

Preview

Text Final Hardbound Thesis.pdf - Accepted Version Download (5MB) | Preview

Abstract

In the current evolution of the digital world, data has become the cornerstone of decision-making processes, shaping industries and societies alike. The exponential growth of data, commonly referred to as big data, has sparked a surge in interest in advanced analytics techniques to harness its potential. Among these techniques, big data analytics, particularly in healthcare, holds immense promise for understanding overall population health and predicting high-risk and high-cost individuals. This thesis delves into the realm of healthcare analytics in Malaysia, focusing on the analysis of extensive medical data to identify patterns and insights that can aid in the identification of high-risk and high-cost individuals. The objectives of this research are: first, to uncover and comprehend usage patterns within healthcare claims data, elucidating factors contributing to the identification of high-risk individuals; second, to propose an innovative ensemble stacking model approach; and third, to demonstrate the efficacy of this approach in enhancing predictive accuracy. The proposed ensemble stacking model integrates the Stacking technique with hybrid feature selection and feature engineering methodologies. By amalgamating multiple predictive models into a cohesive framework, the ensemble model offers superior predictive accuracy compared to traditional single-model approaches. Furthermore, the model's versatility enables its application across various classification tasks within the healthcare domain. Through empirical analysis, this research highlights the enhanced predictive accuracy and efficacy of the ensemble model framework. Notably, key features such as ICD Category, TotalRemainingAmt, and TotalAmtInsured emerge as significant contributors to determining an individual's risk profile based on their medical claim patterns and behaviours. By leveraging big data analytics and ensemble modelling techniques, this research contributes to the advancement of predictive analytics in healthcare, offering valuable insights for decision-makers and stakeholders in the industry.

Actions (login required)

View Item