Diagnosing rare Mendelian disorders can be quite a challenge, even for geneticists with years of experience. To streamline this process, a team at Baylor College of Medicine has turned to artificial intelligence. They have created AI-MARRVEL (AIM), a machine learning system aimed at helping prioritize potentially causative variants for Mendelian disorders.The goal is to assist in prioritizing potential causative variants for Mendelian disorders. The research is released today in NEJM AI. Researchers from the Baylor Genetics clinical diagnostic laboratory found that AIM’s module can help make predictions without needing clinical knowledge of the gene being studied, which can aid in uncovering new disease mechanisms. “The diagnostic rate for rare genetic disorders is only about 30%, and on average, it takes six years from the time symptoms start to a diagnosis. There is an urgent need for new methods to speed up and improve accuracy of diagnosis,” said co-corDr. Pengfei Liu, who is an associate professor of molecular and human genetics and also serves as the associate clinical director at Baylor Genetics, stated that AIM is trained using the MARRVEL database, which is a public database of known variants and genetic analysis. The MARRVEL database, previously developed by the Baylor team, contains over 3.5 million variants from thousands of diagnosed cases. Researchers input patients’ exome sequence data and symptoms into AIM, and AIM then provides a ranking of the most likely gene candidates causing the rare disease. The recent benchmark papers use other algorithms for testing models with data from three cohorts: Baylor Genetics, the National Institutes of Health-funded Undiagnosed Diseases Network (UDN), and the Deciphering Developmental Disorders (DDD) project. AIM consistently ranked diagnosed genes as the top candidate in twice as many cases as all other benchmark methods using these real-world data sets.
“We trained AIM to mimic human decision-making, and the machine can do it faster, more efficiently, and at a lower cost. This method has effectively doubled the rate of accurate diagnosis,” said co.The lead author, Dr. Zhandong Liu, who is an associate professor of pediatrics and neurology at Baylor and a researcher at the Jan and Dan Duncan Neurological Research Institute (NRI) at Texas Children’s Hospital.
The AIM also provides new hope for rare disease cases that have remained unsolved for years. Every year, hundreds of new disease-causing variations that could be crucial in solving these unresolved cases are reported. However, it is challenging to determine which cases should be reanalyzed due to the high volume of cases. The researchers tested AIM’s clinical exome reanalysis on a dataset of UDN and DDD cases and discovered that it was able to correctly identify.AIM is able to identify 57% of diagnosable cases. According to Zhandong Liu, the reanalysis process can be made more efficient by using AIM to pinpoint potentially solvable cases with high confidence and then pushing those cases for manual review. It is believed that this tool can uncover a significant number of cases that were previously thought to be undiagnosable.
Furthermore, researchers tested AIM’s ability to discover new gene candidates that have not been associated with a disease. In two UDN cases, AIM accurately predicted two newly reported disease genes as top candidates. This marks a significant advancement in the field.Baylor researchers are using AI to help diagnose rare diseases by narrowing down genetic diagnoses to a few genes. According to Dr. Hugo Bellen, this approach has the potential to uncover previously unknown disorders. The combination of AI technology, certified clinical lab directors’ expertise, highly curated datasets, and scalable automation is making a significant impact by providing comprehensive genetic insights for diverse patient populations, including the most vulnerable.according to Dr. Fan Xia, who is the senior author of the article and an associate professor of molecular and human genetics at Baylor as well as the vice president of clinical genomics at Baylor Genetics. Dr. Xia stated that the new diagnostic intelligence, known as AIM, has shown superior accuracy in identifying genetic disorders in real-world training data from a Baylor Genetics cohort. The goal of Baylor Genetics is to continue developing this technology to improve clinical practice. The other authors of the article are Dongxue Mao, Chaozhong Liu, Linhua Wang, Rami AI-Ouran, Cole Deisseroth, Sasidhar Pasupuleti, Seon Young Kim, Lucian Li, Jill A.Rosenfeld, Linyan Meng, Lindsay C. Burrage, and M.Michael Wangler, Shinya Yamamoto, Michael Santana, Victor Perez, Priyank Shukla, Christine Eng, Brendan Lee and Bo Yuan are associated with Baylor College of Medicine, Jan and Dan Duncan Neurological Research Institute at Texas Children’s Hospital, Al Hussein Technical University, Baylor Genetics, and the Human Genome Sequencing Center at Baylor. The Chang Zuckerberg Initiative and the National Institute of Neurological Disorders and Stroke (3U2CNS132415) supported this work.
