Despite their potential to save time and money, a large majority of commercial Phase 2 and 3 clinical trials in 2023 did not include a pre-planned interim analysis. This indicates a massive, underutilized opportunity to identify failing drugs sooner and reallocate resources more effectively.
To maintain trial integrity while using predictive algorithms, unblinded data must be handled within a secure, isolated computing architecture. This 'enclave' prevents any human—including statisticians, doctors, or the analysts themselves—from accessing the data, thus eliminating the risk of operational bias.
For high-stakes decisions like halting a clinical trial, current AI models lack the reproducibility and explainability demanded by regulators. The 'Brakes' platform deliberately avoids AI in its core decision engine, applying it instead to adjacent problems like patient subgroup analysis where the stakes for error are different.
AI can identify patient subgroups that respond best to a drug, but this creates a paradox. The more narrowly the group is defined, the smaller it becomes, which weakens the trial's statistical power to detect an effect. The core challenge is optimizing the trade-off between signal clarity and statistical viability.
Instead of stopping a trial early for success—which regulators may restrict—a high-confidence early signal provides immense value as business intelligence. This allows sponsors to de-risk and accelerate planning for subsequent phases months earlier, creating millions in value by reducing the gap between Phase 2 and 3.
Contrary to the belief that trial costs are entirely front-loaded, a significant portion—potentially 80%—are marginal and scale with patient progression. Therefore, making a futility call even late in a trial can still generate substantial cost savings, challenging conventional financial assumptions about trial budgeting.