Presentation (CTI)

OBJECTIVES: Despite the potential of generative artificial intelligence (gen-AI) in health economics outcomes research (HEOR), concerns remain about data privacy, transparency, and methodological integrity. This frequently asked questions (FAQ)-style abstract aims to consolidate common questions related to Gen-AI risks and propose mitigation strategies for its safe and effective adoption in HEOR.
METHODS: A targeted literature review (2019-2024) was conducted, focusing on academic publications, industry white papers, and real-world implementation that addressed Gen-AI applications in HEOR. Key themes were extracted and categorized as “FAQs,” each detailing specific risks and outlining potential mitigation measures.
RESULTS: o FAQ #1: How to handle data privacy and confidentiality? Employ secure data pipelines, anonymization protocols (de-identification, pseudonymization), and role-based access controls. o FAQ #2: How to address model bias and inaccuracy? Use diverse data, regular model auditing, and alignment with clinical and health-economic guidelines. o FAQ #3: How to ensure transparency and reproducibility? Incorporate explainable AI frameworks (e.g., SHAP, LIME), maintain open documentation, and conduct sensitivity analyses. o FAQ #4: What about regulatory and ethical considerations? Engage early with institutional review boards, adhere to established regulatory guidance, and implement dynamic ethical checklists. o FAQ #5: What to do if the AI model makes critical errors? Adopt a human-in-the-loop approach for model validation and interpretation.
o FAQ #6: How to handle the integration of AI with existing HEOR tools?
Ensure compatibility through API integration, use modular AI tools, and provide training to team. o FAQ #7: How to mitigate the risk of overfitting? Use cross-validation, regularization and diverse datasets to improve model generalization and prevent overfitting.
CONCLUSIONS: By implementing robust data governance, transparency, and human expert oversight, organizations can mitigate critical risks while fully capitalizing on the rewards of Gen-AI. This FAQ-styled research provides a useful base for HEOR practitioners to navigate the evolving Gen-AI landscape responsibly.