New York – September 19, 2026 -- Medisights, an AI-native platform for life sciences, has completed a study showing purpose-built AI agents can compress months of qualitative physician research into days while tracking the fast-shifting multiple myeloma treatment landscape.
AI moderators sustain 45-to-60-minute physician interviews to map CAR-T referral logic
Medisights' clinical-grade AI conducted adaptive, humanized conversations with specialists, with interviews routinely running 45 to 60 minutes -- longer engagement than traditional researchers typically secure with physicians. The extended dialogue captured prescribing rationale and geographic variation in CAR-T referral pathways for multiple myeloma, a disease marked by rapidly evolving treatment practices.
Insurance authorization emerges as the primary barrier to in vivo CAR-T adoption
The study identified insurance authorization as the leading obstacle blocking access to in vivo CAR-T therapies, pointing to a market access gap that commercial teams must address before broader uptake occurs.
Community oncologists conflate in vivo CAR-T with off-the-shelf ex vivo products
Intelligence gathered by the AI agents revealed a mechanistic knowledge gap: community oncologists frequently confuse in vivo CAR-T with allogeneic or off-the-shelf ex vivo therapies. Medisights flagged this as an urgent case for targeted medical education clarifying viral-vector and LNP-mediated T-cell engineering.
Specialists rank speed-to-treatment above bispecifics for rapidly progressing patients
Specialists confirmed that removing manufacturing wait times opens a meaningful treatment-free interval, positioning "speed-to-treatment" as the defining value narrative for in vivo therapies over bispecific alternatives in patients with rapidly progressing disease.
Clinical adoption depends on matching in vivo efficacy to established ex vivo benchmarks
Results showed that uptake of in vivo therapies hinges on anchoring efficacy data to established ex vivo benchmarks, including MRD negativity rates, depth of response, and overall survival metrics.
"Imagine a world where you have AI acting as a real-time intelligence layer, continuously engaging with specialists to keep you up to date on outlier innovative practices as they happen," said Frank Seo, CEO of Medisights. He said the approach maps regional CAR-T reference patterns and identifies specific barriers to adoption for life sciences companies.
Medisights said its AI agents operate against curated clinical taxonomies designed to maintain fidelity to clinical facts, aiming to give commercial, medical affairs, and field teams a single verifiable data source.