National Human Genome Research Institute (NHGRI)’s Post

Using tools like artificial intelligence (AI) offers great promise for advancing genomics research – such as helping to interpret large genomic datasets or even help diagnose a patient with a genetic disorder. But researchers are finding that AI still needs significant improvements for routine use in genomics and healthcare.     In a new study published in the American Journal of Human Genetics, NHGRI researchers tested large language models like ChatGPT to answer patient questions and suggest a diagnosis. They found that while artificial intelligence (AI) tools can make accurate diagnoses using textbook-like descriptions of genetic diseases, the tools are significantly less accurate when analyzing health summaries written by the patients themselves.    Researchers hope that AI tools eventually expand access to genomic-based diagnoses. This study highlights the need for studies that use more diverse types of training data to improve the AI tools, making them more robust for clinical applications. I encourage you to learn more about these researchers’ findings! https://lnkd.in/emD7yyRc