flâneur

Using Interpretability to Identify a Novel Class of Alzheimer's Biomarkers

goodfire.ai · 4,218 words · saved by 8 readers

An AI model was trained to detect Alzheimer's from blood samples. We opened it up to understand how—and found that DNA fragment length patterns dominate its decision-making. We distilled this insight into a human-interpretable classifier that generalizes better than the biomarker classes previously reported in the literature when tested on an independent cohort. Nicholas Wang†* Christoforos Nalmpantis‡ Ching Fang†* Pouya Niki‡ Mark Bissell†* Pooja Kathail‡ Dron Hazra† Andrey Karailiev‡ Michael Pearce† Javkhlan-Ochir Ganbat‡ Archa Jain† Luca Giacomoni‡ Daniel Balsam† Jonathan Wan‡ Ravi Solanki‡§ January 28, 2026 In this work, we showcase an early example of how we can translate model mechanisms into testable hypotheses. We apply our interpretability methods to Pleiades [5], Prima Mente's epigenetic foundation model, to understand how it detects Alzheimer's disease (AD) from cell-free DNA (cfDNA) in blood. In this post, we detail how we studied Pleiades to identify fragmentomics as a nov

Using Interpretability to Identify a Novel Class of Alzheimer's Biomarkers Research Using Interpretability to Identify a Novel Class of Alzheimer's Biomarkers An AI model was trained to detect Alzheimer's from blood samples. We opened it up to understand how—and found that DNA fragment length patterns dominate its decision-making. We distilled this insight into a human-interpretable classifier that generalizes better than the biomarker classes previously reported in the literature when tested on an independent cohort. Authors Nicholas Wang † * Christoforos Nalmpantis ‡ Ching Fang † * Pouya Nik

saved by

related reading