SOTA genome interpretation with agentic AI: Interstitial lung disease case study
9 hours ago
- #AI Diagnosis
- #Rare Diseases
- #Whole Genome Sequencing
- Whole genome sequencing improves outcomes in NICUs but faces challenges in complexity, cost, and time.
- Gamow Labs' AI agent, George-0.1, matched expert performance in diagnosing genetic diseases from FASTQ files, solving 19/26 cases and adding 2 more.
- In a study with 46 individuals, George-0.1 outperformed clinical labs and matched a human expert, while ChatGPT 5.5 Pro also beat labs but had more errors.
- Case studies highlight issues like false negatives from lab filtering errors and false positives from AI jumping to conclusions or ignoring data.
- The research suggests AI agents can enhance rare disease diagnosis by avoiding pre-filtering pitfalls and scaling efficiently.