Lessons Learned from the Point-of-Care Use of a Facial Analysis Technology

Abstract

Purpose We aimed to evaluate the utility of facial analysis technology for genetic diagnoses in a typical pediatric genetic clinic. Methods A retrospective review identified children (aged <18 years) who had not previously received a definitive genetic diagnosis and underwent a comprehensive genetic evaluation. Their photographs and relevant clinical non-facial features were uploaded to the CLINIC application of the Face2Gene web interface, and the resulting analysis was accessed and correlated to the molecular diagnosis. Results Of the 23 children included, the overall diagnostic yield in this study was 60.9% (14/23). In total, 64.3% of patients had the correct condition suggested in the top 10 differential diagnoses. The gestalt similarity was only 55.6%, but the phenotypic features added by the clinician showed a similarity of more than the medium level in all patients. Conclusion Our data underscore the usefulness of facial analysis technology as an auxiliary point-of-care tool in pediatric genetic clinics, and we also present some considerations to increase accuracy.

Keywords

• face
• genetic techniques
• machine learning