AI in Real-World Evidence: Insights from Mitch Higashi

In this update, we review the Q&A with Mitch Higashi, PhD, ISPOR’s Associate Chief Science Officer published on Pharmalive. It discusses AI in Real-World Evidence (RWE) and health technology assessments (HTA). Key themes include AI-driven early disease detection and NLP tackling unstructured clinical data. HTA bodies are evolving RWE methods for rare diseases, but barriers like data standardization remain. Higashi emphasizes augmented intelligence over autonomous AI. He also predicts quantum computing’s future role in RWE analytics.

Key Insights

Early detection algorithms use supervised learning to classify high-risk patients. They leverage diverse data types, reducing delayed diagnoses and costs. NLP processes unstructured EMR data (80% of healthcare data) to improve RWE research. HTA adoption is increasing as agencies use RWE for therapies with limited trial data, like gene therapies. They are developing external control arms and standardized methods. However, gaps in metadata standards hinder RWE integration. FAIR data principles (Findable, Accessible, Interoperable, Reusable) are prioritized to address this. Augmented intelligence should enhance clinician decision-making, not replace it. It provides real-time evidence synthesis while preserving patient-provider trust. Case studies show AI/ML in radiology and pathology optimizes image analysis. For example, it improves breast cancer screening efficiency by reducing false positives.

Implications

AI in Real-World Evidence could revolutionize clinical research efficiency, especially for rare diseases. However, data standardization is critical for interoperability and validity. ISPOR’s push for metadata frameworks will maintain trust in RWE-driven decisions. Quantum computing may analyze high-dimensional data in real time, enabling novel predictive models. This shift requires rethinking causal inference and patient categorization. Multidisciplinary collaboration will balance innovation with ethical rigor.