Demonstration on Privacy-Preserving Validation of an Overseas-Trained Lung Cancer Prediction Model on Singapore Population Data
Main Applicant – Dr Khin Mi Mi Aung, Senior Principal Scientist, Institute of Infocomm Research, Agency for Science, Technology and Research (A*STAR)

This project aims to demonstrate how privacy-enhancing techniques (PETs) can be applied to securely validate a health risk prediction model within the MoH TRUST environment.

A lung cancer risk model previously trained using Taiwan administrative claims data will be evaluated on Singapore administrative health records without transferring individual-level data outside the secure platform. The focus of this study is to test and document privacy-preserving workflows for cross-population model validation.

All computations will be conducted within TRUST’s controlled environment. No identifiable data will be accessed, exported, or used for individual-level decision-making. Only aggregate performance statistics will be generated.

The project serves as a technical demonstration of how predictive models trained overseas can be evaluated locally under strict privacy governance controls, supporting TRUST’s mission in enabling secure and responsible use of health data.