Predictive modelling of health trajectories

Start Date

1 January 2020

End Date

8 February 2023

Project Participants

HMS, Southern Methodist University, Western Sydney University

Tags: acute care, data analytics, decision support, health planning, health system efficiency, hospital services, predictive modelling, risk stratification

Opportunity

Preventable hospital readmissions and chronic disease progression represent significant cost and quality‑of‑care challenges for health systems. There is a strong need for robust, data‑driven tools that can identify individuals at high risk early enough to enable targeted interventions and better resource allocation. This project explored the use of Health Management System’s (HMS) extensive data collections to create algorithms for predicting an individual’s risk of chronic disease.

Project objective

This project aimed to measure and understand hospital readmission risk using administrative and claims data, and develop algorithms for predicting risk at an individual level across major chronic conditions.

Publications

Refereed Journal ArticlePublished 1st September 2023Predicting Hospital Readmission in Medicaid Patients with Heart Failure Using Administrative and Claims Data
Full Publication
Refereed Journal ArticlePublished 22nd April 2024Predicting Hospital Readmission in Medicaid Patients With COPD Using Administrative and Claims Data
Full Publication
Refereed Journal ArticlePublished 22nd August 2023Predicting Hospital Readmission in Medicaid Patients With Diabetes Using Administrative and Claims Data
Full Publication
ThesisPublished 3rd August 2022Dynamic Prediction for Alternating Recurrent Events Using a Semiparametric Joint Frailty Model
Full Publication

Final Outcomes

The project successfully delivered validated predictive models for hospital admission risk in diabetes, heart failure, and chronic obstructive pulmonary disease (COPD), drawing on routinely-collected Medicaid claims data and looking at demographics, admission and discharge characteristics, and comorbidities.

The project also supported the successful completion of a PhD program.

Impact and Benefits

These algorithms were transferred to the industry partner to be applied in practice for relevant populations.

The research advanced novel statistical methods for predicting recurring events, demonstrating an ability to enable healthcare providers to identify patients at high risk of readmission using administrative health data, to support earlier, more targeted interventions. This can also assist governments with financial planning and allocation of resources, supporting health system efficiency by improving population stratification and planning.

Media Appearances

Integrity, Excellence,
Teamwork and Authenticity

Learn More