Genetics in Medicine Open · Published 2026-07-01 · DOI 10.1016/j.gimo.2026.104488
Yinghong Pan, Patrick Danley, Tamar Kramer, Mehrdad Noruzinia, Karen Buser, Alexander N. Yatsenko, Daniel Bellissimo, Fen Guo
Purpose: This retrospective study examined the clinical and genetic characteristics of pediatric patients undergoing clinical exome sequencing (ES) and evaluated the performance of a commercially available artificial intelligence (AI) platform that was integrated into our analysis pipeline. Methods: ES was performed in 822 consecutive patients at a single clinical laboratory. AI-based tools were used to jointly assess genetic information and the proband’s Human Phenotype Ontology terms to support variant prioritization during the initial case review. Results: A definitive molecular diagnosis was established in 22% (181 of 822) of index cases, while 40% (325 of 822) had variants of uncertain significance. Among those with a definitive diagnosis, 93% (168 of 181) had a single finding and 7% (13 of 181) had multiple findings. Of the 152 reported pathogenic/likely pathogenic variants in the fully resolved cases, 98.7% were successfully flagged by AI, and 75.0% ranked among the top 10 “most likely” variants. Conclusion: Clinical ES provides a substantial diagnostic yield in complex pediatric disorders. Integration of AI-powered platforms can accelerate phenotype-driven variant prioritization and facilitate rare disease diagnostics, but underscores the need for careful validation and optimization in clinical workflows.
Abstract from DOAJ. Public domain (CC0 1.0).
Read the article at the publisher →
Pan, Y., Danley, P., Kramer, T., et al. (2026). AI-Assisted Clinical Exome Sequencing: Insights and Outcomes from 822 Pediatric Diagnoses. Genetics in Medicine Open. https://doi.org/10.1016/j.gimo.2026.104488