AI assistant prototype promises faster rare disease diagnosis by flagging cases for reanalysis and synthesising gene and variant data.<\/p>\n
Microsoft, Drexel University, and the Broad Institute have developed a generative AI assistant<\/a> to support genome sequencing. The study in ACM Transactions on Interactive Intelligent Systems demonstrates how AI can accelerate searching, filtering, and synthesising data in rare disease diagnosis.<\/p>\n Whole genome sequencing often takes weeks and yields a diagnosis in fewer than half of cases. Analysts must decide which unsolved cases to revisit as new research appears. The AI assistant flags cases for reanalysis and compiles new gene<\/a> and variant data into a clear, usable format.<\/p>\n The team interviewed 17 genetics professionals to map workflows and challenges before co-designing the prototype. Sessions focused on problems such as data overload, slow collaboration, and difficulty prioritising unsolved cases, helping ensure the tool addressed real-world pain points.<\/p>\n The prototype enables collaborative sensemaking, allowing users to edit and verify AI-generated content. It offers flexible filtering to surface the most relevant evidence while keeping a comprehensive view, saving time and improving decision-making.<\/p>\n Microsoft-led researchers plan to test the assistant in real-world environments to measure its effect on diagnostic yield and workflow efficiency. They emphasise that success will depend on collaboration among developers<\/a>, genetic experts, and system designers to build trustworthy and explainable tools.<\/p>\n