By using foundation models to analyze vast datasets, companies can create a synthetic 'standard of care' arm for single-arm Phase 1 trials. The AI matches patients based on deep clinical and genomic parameters, providing insights comparable to a much larger Phase 3 trial.

The CREST trial showed benefit driven by patients with carcinoma in situ (CIS), while the Potomac trial showed a lack of benefit in the same subgroup. This stark inconsistency demonstrates that subgroup analyses, even for stratified factors, can be unreliable and are a weak basis for regulatory decisions or label restrictions.

For accelerated designations, a clean clinical signal from a small, homogenous patient sample is more valuable than a weaker signal from a larger, more diverse group. Early cohorts should be narrowed to a uniform population representing the true unmet medical need to ensure consistency of results.

The traditional drug-centric trial model is failing. The next evolution is trials designed to validate the *decision-making process* itself, using platforms to assign the best therapy to heterogeneous patient groups, rather than testing one drug on a narrow population.

Contrary to the belief that AI needs massive datasets, Dr. Joseph Juraji's approach with NetraAI focuses on finding small, specific patient subpopulations within small trials. This allows the identification of a drug's 'superpower' without the need for big data, transforming trial economics.

Current patient selection for radioligand therapies often relies on a simple binary 'positive' or 'negative' assessment, which fails to predict response 55% of the time. AI moves beyond this by providing quantitative, whole-body tumor assessments to predict who will actually respond to treatment, a far more critical factor for clinical and commercial success.

Even when trials like LITESPARK 022 and Keynote 564 use identical eligibility criteria, outdated staging systems result in patient populations with different underlying risks. This makes direct comparison of outcomes between trials, even for the same drug, an unfair and statistically flawed analysis that ignores the function of a control arm.

Despite meeting its primary endpoint, the PROTEUS trial provides no validated biomarkers to identify which patients actually benefit from the intensified therapy. This lack of a predictive signature means applying the results in the clinic amounts to uniform escalation, likely overtreating many patients for an uncertain benefit, making it difficult to implement.

Experts caution against making definitive clinical decisions based on subgroup analyses, such as the apparent lack of benefit for upper tract disease in an adjuvant immunotherapy trial. These analyses are hypothesis-generating, and conflicting results between different endpoints (e.g., DFS vs. OS) demonstrate they can be misleading.