MIT-led team unveils FINGERS-7B foundation model to revolutionize early Alzheimer’s prediction and prevention

FINGERS-7B is the first AI foundation model built to make Alzheimer’s preventable, offering 4x better accuracy in detecting early risks before symptoms appear.

By: AXL Media

Published: Apr 28, 2026, 12:25 PM EDT

Source: Information for this report was sourced from The Picower Institute

A New Frontier in Preclinical Alzheimer’s Detection

The release of FINGERS-7B at the 2026 ICLR conference in Rio de Janeiro marks a fundamental shift in how neurodegenerative risks are assessed. Developed by researchers at MIT, the Broad Institute, and the Picower Institute, the model is designed to identify Alzheimer’s during its preclinical stage—a phase that can precede visible memory symptoms by over a decade. Unlike previous diagnostic tools that analyze biological data in isolation, FINGERS-7B utilizes a 7-billion-parameter architecture to synthesize lifestyle, clinical, genomic, and proteomic data into a unified predictive framework. This "multi-omic" approach allows the model to detect subtle correlations across different biological domains that traditional standalone biomarker tests often miss.

The Multi-Omic Breakthrough and "Biological Fingerprints"

At the heart of the model’s success is the concept of the "biological fingerprint," a unique combination of signals that reveal an individual's specific disease risk. By reading diverse data types together, FINGERS-7B has identified a novel set of diagnostic biomarkers that significantly outperform prior art. On the World-Wide FINGERS (WW-FINGERS) network datasets, the model demonstrated a 400% increase in the accuracy of preclinical diagnoses. Furthermore, it offers a 130% improvement in responder stratification, a critical metric that helps clinicians determine which patients are most likely to benefit from particular preventive measures, such as specific dietary changes or early-stage pharmacological treatments.

FINGERPRINT: An AI-Driven Discovery Acceleration Engine

FINGERS-7B serves as the foundational core of FINGERPRINT, a broader discovery platform that employs specialized AI agents to automate complex multi-omic analyses. According to Adrian Noriega, a postdoctoral fellow at MIT-Novo Nordisk AI, FINGERPRINT acts as an acceleration engine that interprets complex biological signals to find novel interventions and therapeutics. The model does not just offer a binary diagnosis; it provides personalized analyses that predict the likely time course of an individual’s cognitive decline. This level of temporal forecasting allows for the precision-timing of interventions, potentially delaying or even halting the onset of clinical dementia symptoms.