Machine Learning Strategies for Improved Phenotype Prediction in Underrepresented Populations

Area of Science:

• Genomic Medicine
• Bioinformatics
• Population Genetics

Background:

• Precision medicine models exhibit performance disparities due to over-representation of European ancestry in genomic datasets.
• This imbalance leads to less accurate predictions and treatment recommendations for underrepresented populations, exacerbating health disparities.
• Existing genomic prediction models require enhancement to ensure equitable application across diverse ancestries.

Purpose of the Study:

• To introduce an adaptable machine learning toolkit designed to improve prediction accuracy for underrepresented populations in genomic datasets.
• To address health disparities by enhancing the performance of precision medicine models for diverse ancestral groups.
• To develop novel techniques that integrate with existing methods for more inclusive genomic data analysis.

Main Methods:

• Utilized machine learning techniques, including gradient boosting and automated methods.
• Developed novel population-conditional re-sampling techniques to address data imbalances.
• Integrated existing methodologies with new techniques into an adaptable toolkit for genomic prediction.

Main Results:

• Demonstrated significant improvements in phenotypic prediction from single nucleotide polymorphism (SNP) data for diverse populations.
• Achieved prediction accuracy for underrepresented groups comparable to that of the majority European ancestry group.
• Validated the approach using the UK Biobank, showing substantial gains for Asian and African ancestry groups.

Conclusions:

• The developed machine learning toolkit significantly enhances phenotype prediction accuracy for underrepresented populations.
• This approach represents a crucial advancement in mitigating health disparities stemming from biased genomic datasets.
• The toolkit fosters more equitable validity and utility of statistical genetics methods, promoting inclusive precision medicine.