A team of researchers from Johns Hopkins Bloomberg School of Public Health and the National Cancer Institute has developed a new algorithm for genetic risk-scoring that can help reduce health care disparities in non-white populations. Genetic risk-scoring algorithms are used to identify individuals at high risk for diseases and conditions. However, these algorithms have mainly been developed using data from people of European ancestry, leading to performance gaps in non-European populations. The new method, called CT-SLEB, combines AI techniques and has been tested on data from diverse populations, showing promising results.
New method can improve assessing genetic risks for non-white populations
The team of researchers from Johns Hopkins Bloomberg School of Public Health and the National Cancer Institute has developed a new algorithm, called CT-SLEB, for genetic risk-scoring that holds promise for reducing health care disparities in non-white populations. Genetic risk-scoring algorithms are used to identify individuals who are at a higher risk for diseases and conditions, such as cancers and heart diseases. However, these algorithms have primarily been developed using data from people of European ancestry, which has led to performance gaps in non-European populations. The new method combines AI techniques, including machine learning and Bayesian statistical modeling, and has been tested on data from diverse populations, including European, African, Latino, East Asian, and South Asian populations.
Addressing the performance gap in risk-scoring algorithms
One of the main challenges in developing risk-scoring algorithms for non-European populations is the lack of large-scale genetic studies in these populations. Many existing risk-scoring models are based on relatively small-scale studies, leading to a performance gap between European and non-European populations. The researchers behind the CT-SLEB method have shown that their algorithm can help close this performance gap to some extent. However, they also emphasize the need for larger datasets on non-European populations to fully address the gap.
Combining AI techniques for improved risk-scoring
The CT-SLEB method combines various AI techniques, including machine learning and Bayesian statistical modeling, to improve genetic risk-scoring in non-white populations. By training the algorithm on data from diverse populations, including those from the 23andMe database, the Global Lipids Genetics Consortium, the National Institutes of Health’s All of Us research program, and UK Biobank, the researchers have been able to generate genetic scores for 13 traits across five different ancestry categories. The results of their study show promising improvements in risk-scoring performance for non-European populations, highlighting the potential of AI in addressing health care disparities.
