In the last decade, artificial intelligence (AI) has advanced to a point where machines are now able to surpass human performance in tasks such as reading comprehension and image recognition. One of the technology’s strengths lie in processing enormous volumes of complex data, uncovering patterns and making predictions with minimal human intervention.
The intersection of AI and genetics provides an opportunity to derive new insights into human health and disease at a scale and speed never thought possible. AI algorithms, powered by machine learning models and deep learning networks, have the power to tackle the wealth of genomic data generated by next-generation sequencing and more recently, long-read sequencing. The possibilities are endless, from discovering genetic underpinnings of disease to identification of novel therapeutic targets.
Researchers at Mount Sinai’s Department of Genetics and Genomic Sciences are at the forefront of this pivotal shift, incorporating cutting-edge AI methods to advance precision medicine and patient care. The Department is ranked second in the country in National Institutes of Health (NIH) funding in genetics research, with several specialty research programs including AI approaches to drug discovery, functional genomics, single-cell sequencing, and genomics of infectious diseases.
As an example, the lab of Kuan-lin Huang leverages AI methods in developing new precision medicine approaches for better detection, treatment, and prevention of disease. To predict disease risk or treatment, the current precision medicine paradigm often relies on single genetic mutations or markers, disregarding much of the useful genomic information. Using biobank-scale genomic cohorts, like Mount Sinai’s BioMe, Huang and his colleagues are actively developing machine learning models that can more accurately predict disease risks and treatments using the full genome profile.
“BioMe is unique in that, compared to many other sequencing cohorts, it has a diverse population of not just Europeans, but also African Americans and Latino Americans,” says Huang, PhD, Assistant Professor of Genetics and Genomic Sciences at Icahn Mount Sinai. “We want to ensure that genetic tests of disease risk not only work well for Caucasians, but they also work for people from other ancestral backgrounds.”
For example, studies using large-scale DNA sequencing have detected rare pathogenic variants in 5% to 10% of adult cancer cases. On the other hand, genome-wide association studies have revealed hundreds of common variants that together may contribute to an elevated risk. However, rare and common genetic variants are often assessed separately and prohibit accurate risk assessments.
By analyzing a BioMe cohort of 49,854 individuals — including 22% African American, 1.9% Asian, 34% Latinx/Hispanic, 32% European, and 11% of other self-reported ancestries — Huang and his colleagues recently developed machine learning models for improved cancer risk predictions that combine both common and rare genetic variants. The models show higher predictive accuracy than using common or rare variants alone, and they also detected significant interactions of common and rare genetic variants contributing to risks of selected cancer types.
“Our study underscores the importance of jointly considering common and rare genetic factors to provide accurate cancer risk assessment for diverse individuals,” Huang says.
In a 2024 study, Huang and his colleagues harnessed the power of machine learning to pinpoint key predictors of mortality in dementia patients. The study used data from more than 40,000 patients from the National Alzheimer’s Coordinating Center, a database spanning about 40 Alzheimer’s disease centers across the United States. The researchers created machine learning models based on clinical and neurocognitive features that could predict mortality at one, three, five, and 10 years for eight types of dementia.
The results suggest that performance on neuropsychological tests may be a better indicator of mortality risk than age-related factors such as cancer and heart disease. Such findings could aid clinical practices, potentially allowing for earlier interventions and tailored treatment strategies to improve patient outcomes.
"The implications of our research extend beyond clinical practice, as it underscores the value of machine learning in unraveling the complexities of diseases like dementia," Huang says. "If cross-validated and carefully implemented at the primary care level, such predictive models have the potential to improve personalized care and precision medicine."
Born and raised in Taiwan, Huang earned Bachelor’s degrees in both studio art and molecular biology from Wesleyan University before obtaining his Ph.D. in genetics and genomics from Washington University in St. Louis. He joined Mount Sinai as a faculty member in October 2018. In addition to his many scientific publications, Huang has also written a best-selling book, “Solve It Yourself: Fix the World’s Problem with Science,” which provides science-backed ways for readers to take immediate action towards solving issues like climate change, viral pandemics, and social media addiction.
Ultimately, the mission of his life’s work is to empower people to live longer, healthier lives — and he believes AI can help to reach that goal.
“I think it's unquestionable that, in many fields of research, and eventually clinical applications as well, AI will be a powerful tool for discovery,” Huang says.