Talos: Scaling rare disease diagnosis with automated, iterative genomic reanalysis
| Source: Microsoft Research Blog
Tags: Microsoft Research, rare disease, genomics, Talos, healthcare AI, open source, Broad Institute, automated diagnosis
Microsoft Research and partners released Talos, an open-source AI system for automated rare disease genomic reanalysis that recovered 90% of diagnoses while flagging only 1.3 candidate variants per patient — delivering 241 new diagnoses across a 5,000-patient cohort with a median time-to-diagnosis of 32 days from new evidence publication.
Details
Talos, developed by Microsoft Research, the Broad Institute, Australian Genomics, and the Centre for Population Genomics, automates reanalysis of stored patient genome sequences against constantly evolving medical knowledge. Genomic testing historically leaves more than half of rare disease patients undiagnosed after initial testing — but as gene-disease associations accumulate, reanalysis of the same data often yields diagnoses that were impossible at first evaluation. The system's headline metrics: across a validation set of ~1,100 patients, Talos recovered 90% of in-scope diagnoses while flagging only 1.3 candidate variants per patient for expert review. This low false-positive rate is what makes automated reanalysis practically sustainable. Previous manual-review bottlenecks meant the vast majority of stored genomes were never revisited. In a prospective cohort of nearly 5,000 undiagnosed patients, Talos delivered 241 new diagnoses — a 5.1% additional yield. On monthly iterative cycles, analysts reviewed only one new variant per 200 patients on average. The median time between new evidence becoming public and a resulting diagnosis was 32 days — compared to the years-long waits common in manual reanalysis. Talos is open-source and peer-reviewed, making it immediately deployable by other genomic medicine programs.