Researchers at the Johns Hopkins Bloomberg School of Public Health and the National Cancer Unit have developed a new genetic risk-scoring algorithm for major diseases that include people of different ethnic backgrounds.
In some ethnic populations, certain diseases are more prevalent due to their common ancestry and the new scalable artificial intelligence (AI)-based method, CT-SLEB, has shown promise in reducing healthcare disparities.
People of African and Mediterranean descent are more susceptible to sick-cell disease, while cystic fibrosis and haemochromatosis are more common among European populations.
Genetic risk-scoring algorithms work to identify high-risk groups who could benefit from preventative interventions for various diseases, including cancer and heart disease.
Published in Nature Genetics and funded by the National Institutes of Health, CT-SLEB included over five million individuals across diverse populations to generate genetic scores for 13 traits, including coronary artery diseases and depression, in five different ancestry categories: European, African, Latino, East Asian and South Asian.
Researchers also tested the new method in large-scale stimulation studies.
Using AI techniques and Bayesian statistical modelling, researchers trained CT-SLEB on data from 23andMe, Global Lipids Genetics Consortium, the National Institutes of Health’s All of Us research programme and UK Biobank.
They found that CT-SLEB improved genetic risk scores in African ancestry populations where scoring accuracy is generally low.
Additionally, they found that CT-SLEB was computationally faster compared to its competitors and analysed larger numbers of DNA variants and more populations.
Nilanjan Chatterjee, senior author of the study and Professor at Johns Hopkins Bloomberg School’s Department of Biostatistics, said: “Our method can help close the risk-scoring performance gap for non-European-ancestry populations.
“At the same time, we also concluded that we can’t fully close the gap with new methods alone – we also need larger datasets on these populations.”
The team’s benchmarking analyses found that new ancestry-specific risk-scoring models for non-European populations outperformed standard polygenic risk score models based on European-ancestry datasets or smaller non-European ancestry datasets.
Researchers are now working with more advanced, better-performing methods to achieve polygenic risk score models that work well in both non-European-ancestry populations and non-European-ancestry populations, which will require additional genome-wide association studies in non-European-ancestry populations.