Abstract

Mendelian disorders are prevalent in neonatal and pediatric intensive care units and are a major cause of morbidity and mortality in these facilities. Current diagnostic pipelines that integrate phenotypic and genotypic data are expertdependent and time-consuming. Artificial intelligence (AI) tools can help solve these challenges. Analyze the patient’s phenotype and genotype to establish an orderly differential diagnosis. We used Dx29 to retrospectively analyze 25 acutely ill infants diagnosed with Mendelian disorders using a targeted panel of approximately 5000 genes. For each case, trio files (subject and parents) were analyzed using information on genetic mutations and patient phenotypes provided to Dx29 through three approaches. AI extraction with manual review/editing, and manual entry. Next, we determined the rank of the positive diagnosis in the differential diagnosis of Dx29. Using these three approaches;Dx29 placed the correct diagnosis in the top 10 with 92-96% probability. These results are due to the use of automated phenotyping of her Dx29 by a layman followed by data analysis compared to the standard workflow developed by Bioinformatics by the expert used for the analysis. Suggests that Genomic data and diagnosis of Mendelian disease may be informative.