AI Reveals a Hidden Effect in a Failed Alzheimer’s Trial

By Arkadi | July 18, 2025

The Disease Does Not Progress the Same Way for Everyone

In a groundbreaking development, scientists have employed an AI model to stratify Alzheimer’s patients into various subgroups based on their disease progression rates. Specifically, patients are categorized as either slow or rapid progressors. This research has revealed that a particular treatment shows efficacy for the slow progressor group but fails to benefit those who progress rapidly. Moreover, this treatment appears to delay the transition for patients in the slow group from moving to the more rapid progression category.

This innovation stems from a recent study conducted at the University of Cambridge, published in Nature Communications, where researchers utilized the Predictive Prognostic Model (PPM) trained on data from the Alzheimer’s Disease Neuroimaging Initiative (ADNI).

Stratify and Conquer

The challenge with clinical trials is the often-homogenous testing groups that do not account for variations in patient responses to treatments. The question arises: what if a medication deemed ineffective actually works for a subset of patients that trial designs overlook? Scientists postulate this may contribute to the dismal success rates of Alzheimer’s clinical trials, historically plagued by the variability in patient populations.

The researchers aimed to identify those subsets through their AI model, achieving an impressive classification accuracy of 91.1% from a relatively small sample size of 256 patients and three key parameters: β-amyloid, APOE4, and medial temporal lobe grey matter density.

The Hidden Effect

Applied to the real-world AMARANT trial, which originally failed to demonstrate the efficacy of the drug lanabecestat (a BACE1 inhibitor aimed at reducing β-amyloid plaques), the AI model analyzed patients’ baseline data to assign a prognostic score indicative of the disease's rate of progression.

“Our AI model gives us a score to show how quickly each patient will progress towards Alzheimer’s disease,”

- Professor Zoe Kourtzi, Senior Author

Upon reevaluation of the drug’s efficacy for the slow-progressing subgroup, it was found that the drug resulted in a remarkable 46% slowdown in disease progression in patients receiving the higher 50mg dose. This result surpassed the effects observed in previous successful trials when data were aggregated across all patient types.

Alzheimer’s Escape Velocity

Further, the researchers investigated whether the treatment could help prevent patients from transitioning from slow to rapid progressors. Results indicated that patients in the 50mg lanabecestat group experienced a lower transition rate: only 33.3% of them evolved to rapid progressors compared to 60% in the placebo group.

This outcome is especially pertinent given that newer treatments for Alzheimer’s are anticipated to gain approval soon and may provide increased effectiveness for patients demonstrating slower progression rates. By integrating enhanced initial diagnostics, it’s conceivable to reach an “Alzheimer’s escape velocity,” a point at which one therapy sufficiently slows the disease, allowing subsequent interventions to take effect.

“AI can guide us to the patients who will benefit from dementia medicines, by treating them at the stage when the drugs will make a difference, so we can finally start fighting back against these cruel diseases,” - Kourtzi

Cheaper Trials

By utilizing the AI tool for stratification from the onset, Alzheimer's clinical trials may see a significant reduction in costs and an increase in success rates. The researchers posited that a future trial utilizing the slow-progressing AI-selected group would necessitate only 82 patients per group (treatment vs. placebo). In comparison, a mixed group would require 762 patients per group, indicative of a 90% reduction in required sample size, potentially saving hundreds of millions in research costs.

This approach does not suggest abandoning the population of rapid-progressing patients; rather, many within that cohort may have been former slow progressors. The insights contribute to the understanding of how promising treatments might fail when deployed too late in the disease trajectory.

Conclusion

In summary, the research highlights the importance of patient stratification in clinical trials, particularly for complex diseases such as Alzheimer’s. The potential for an AI-guided approach can transform the landscape of Alzheimer’s research, enhancing the precision of drug development, healthcare spending, and ultimately patient outcomes. As stated by Kourtzi, “with our AI model, we can finally identify patients precisely and match the right patients to the right drugs.”

Literature Cited

[1] Vaghari, D., Mohankumar, G., Tan, K., Lowe, A., Shering, C., Tino, P., & Kourtzi, Z. (2025). AI-guided patient stratification improves outcomes and efficiency in the AMARANTH Alzheimer’s Disease clinical trial. Nature Communications, 16(1), 1-12.