Ai system classifies genetic disease severity using medical guidelines
Large Language Model Agents for Evidence Based Genetic Disease Severity Classification
Artificial IntelligenceComputation and Language
Summary
Classifying how severe genetic diseases are can be hard and slow because it often depends on expert judgment. The authors created an AI program that reads medical research papers and uses official guidelines to decide how severe different genetic conditions are. This AI explains its reasoning and checks its own conclusions to make sure they are supported by evidence. It was tested and found to agree well with expert lists, making it useful for designing genetic screening panels.
What this means in practice
- •For clinical genetic testing teams: Automate severity classification of genetic conditions to standardize and improve gene panel designs for patient screening.
- •For healthcare data analysts: Use AI-supported evidence retrieval and interpretation to validate and curate gene severity annotations in genomic databases.
Authors
Tohid Ghasemnejad, Ahmadreza Argha, Mark Grosser, John Wang, Min Yang, Thantrira Porntaveetus, Tony Roscioli, Nigel H. Lovell, Mahmoud Aarabi, Hamid Alinejad-Rokny
Abstract
Disease severity classification for genetic conditions is subjective and labor-intensive, creating bottlenecks in genomic screening, where commercial panels vary widely in size and overlap. We developed an autonomous AI agent integrating Reasoning and Acting (ReAct) with Retrieval-Augmented Generation (RAG) to classify 10,211 Human Phenotype Ontology terms. It uses American College of Medical Genetics (ACMG)-endorsed severity guidelines and American College of Obstetricians and Gynecologists (ACOG) quality-of-life criteria to retrieve PubMed literature, generate interpretable reasoning chains, and independently verify claims. At the phenotype level, using expert-curated cohorts, the agent achieved 93.55% accuracy (MCC 0.9237) with 82.6% to 91.4% of claims supported by direct evidence or valid inferences. Gene-level severity was aggregated across 8,738 pairs, identifying 3,283 autosomal recessive pairs with severe or profound presentations. External validation showed 95.2% concordance with Mackenzie's Mission gene list. This system enables standardized panel design by providing reliable, automated classification supported by direct evidence.