Artificial Intelligence in Medicine: Applications, Limitations and Future Directions

Abstract

In the area of genomics, Machine Learning (ML) is used to understand how genotype can influence phenotype, such as variations linked to genetic disease. This advanced use of ML was enabled by the recent explosion in automatic data collections through high-throughput sequencing technologies and the increased compute power through public cloud providers. However, the opportunity to process more data than ever before and train more sophisticated methods also means that the discipline of genomics increasingly becomes a digital domain, requiring new algorithms and different analysis strategies.

In this chapter, we discuss real world examples on how ML is used in clinical genomics to identify novel disease genes, prioritize pathogenic variants and as chatbots, supporting genetic counselling. We also discuss ML-applications in the biosecurity space where pathogen strains are grouped into evolutionary networks or the most effective CRISPR binding sites are recommended by ML. The chapter concludes by discussing the funding landscape for ML in the genomics space and provides an outlook on the ML-heavy future applications cases such as Gene Therapy, tracking Antimicrobial Resistance and personalized disease risk prediction for cardiovascular health.