Towards Explainable and Privacy-Preserving LLM Frameworks for Personalized Healthcare Part 2

The project is the direct continuation of application #IT47869, extending the work initiated in the first 24-week phase toward a more scalable, robust, and deployment-ready Large Language Model framework for healthcare. While the initial unit focused on designing the system architecture, implementing explainability and privacy mechanisms, and establishing preliminary evaluation methods, this second phase will advance the prototype into an optimized, rigorously validated, and audit-ready system suitable for real-world clinical contexts. The research will include performance optimization, cost–efficiency benchmarking, expanded user studies with stakeholders, and fairness-oriented audits to ensure equitable, transparent, and reliable model behaviour across different patient and clinician profiles. Through this work, the student will gain advanced skills in system profiling, scalable evaluation, bias assessment, and privacy-aligned optimization, competencies that are increasingly essential for deploying trustworthy AI in healthcare. This continuation will strengthen the student’s academic development, support high-impact publications, and consolidate their preparation for future leadership roles in human-centred and responsible AI research.

Faculty Supervisor:

Glaucia Melo dos Santos

Student:

Partner:

Universidade Federal do Rio de Janeiro

Discipline:

Computer science

Sector:

Education

University:

Toronto Metropolitan University

Program: