In the study, scientists used a deep learning algorithm to analyze patient photos for observable, phenotypic data. When this data was combined with genetic data into existing workflows, AI systems were able to identify disease-causing genes with greater accuracy than systems fed only genotypic data.
Facial dysmorphism and other phenotypic data are regularly used to detect genetic diseases, but must be entered manually into AI-powered diagnostic tools. The system outlined in the report therefore not only streamlines the process of analyzing phenotypic data, but also improves the abilities of AI tools for diagnosing rare diseases.
More articles about AI:
Microsoft building AI hub in Louisville to boost local healthcare industry
Deep learning tool helps radiologists spot brain aneurysms
AMIA encourages stricter FDA oversight of machine learning algorithms
