Clinicians have increasingly used artificial intelligence in various medical fields to enhance decision-making. Recent findings suggest that AI in genetic diagnosis can significantly accelerate the identification of causative factors for complex conditions. Specifically, researchers examined the utility of Med-PaLM 2 and Gemini 2.5 Pro across four levels of genetic complexity. First, Med-PaLM 2 successfully identified murine genes with verified genetic factors for several biomedical traits. Furthermore, the model discovered a novel causative factor for spontaneous hearing loss. Consequently, experimental validation using knock-in mice confirmed this finding, which proves the model's predictive power.

Optimizing Clinical Outcomes with AI in Genetic Diagnosis

In addition, the team developed a retrieval and grounding pipeline for human applications. This pipeline enabled Gemini 2.5 Pro to analyze genomic sequences of 20 human subjects suffering from hearing loss. The AI identified causative genetic factors for these individuals with high accuracy. Moreover, the researchers modified the pipeline to analyze subjects with rare genetic diseases. These subjects required dozens of terms to describe their multi-faceted symptom complexes. Surprisingly, the model succeeded without any task-specific fine-tuning. Therefore, these results demonstrate that an integrated AI pipeline can streamline genetic discovery in both mice and humans.

Frequently Asked Questions

How does AI in genetic diagnosis improve patient care?

AI helps clinicians by sifting through massive genomic datasets and linking complex phenotypes to causative genes, significantly shortening the diagnostic journey for rare diseases.

Can LLMs like Gemini 2.5 Pro identify genes without specialized training?

Yes, research shows that with a proper retrieval and grounding pipeline, Gemini 2.5 Pro can identify causative factors for rare diseases and hearing loss without task-specific fine-tuning.

Disclaimer: This content is for informational and educational purposes only and does not constitute medical advice or a professional relationship between the reader and the author. While we strive for accuracy, medicine is a constantly evolving field. The information provided may not reflect the most current research or clinical guidelines. Always seek the advice of a qualified healthcare provider with any questions regarding a medical condition. Never disregard professional medical advice or delay seeking it because of something you have read here. Refer to the latest local and national guidelines for clinical practice.

References

Tu T et al. Genetic Diagnosis and Discovery Enabled by Large Language Models. Adv Sci (Weinh). 2026 Feb 08. doi: 10.1002/advs.202518656. PMID: 41655254.

Saab K, Tu T, et al. Capabilities of Gemini Models in Medicine. arXiv:2404.18416. 2024.

Singhal K et al. Large Language Models Encode Clinical Knowledge. Nature. 2023;620(7972):172-180.