Session (CTI)

Issue AI-enabled literature review platforms have rapidly expanded across HEOR, regulatory, and medical affairs workflows. The first wave of pilots demonstrated substantial gains in screening speed and reviewer efficiency, but also revealed variability in accuracy, transparency, and reproducibility across review types and organizational settings. Now, as multiple teams have completed successive reviews with the same platforms, the field is entering a second phase, one where outputs can be compared across projects, procurement expectations are more structured, and human, AI workflows are stabilizing. Overview The panel will not compare product features. Instead, it will synthesize cross-platform learnings and debate what the HEOR community should expect—and require—from AI-enabled evidence synthesis in its next phase of adoption. Attendees will leave with: (1) a data-informed understanding of AI SLR maturity, (2) practical evaluation considerations for real-world implementation, and (3) a clearer view of how automation and human judgment must align to preserve scientific integrity. Panelists will examine how platform capabilities have evolved since 2025 and where divergence persists in real practice. Discussion topics will include: • How human–AI integration has changed across repeated projects • Which elements of transparency, version control, and auditability have proven essential • What patterns are emerging in validation, error tolerance, and sensitivity trade-offs • Where AI-enabled tools still face methodological gaps • What “good” should look like as the field moves toward standardization