IMPE Abstracts

Introduction: Face2Gene (FDNA Inc., Boston, MA) is a web application for artificial intelligence-driven facial phenotyping, providing a list of the top thirty candidate syndromes based on gestalts from two-dimensional facial images of patients. Although the effectiveness of Face2Gene for the diagnosis of congenital syndromic disorders has been reported mainly in Caucasian populations, the sensitivities for syndromic endocrine-related disorders and the performance of this system for East Asian patients remain unclear.

Objective: To reveal the sensitivities of Face2Gene for various syndromic endocrine-related disorders and the clinical usefulness of Face2Gene as a diagnostic tool in Japanese patients.

Methods: We recruited 21 Japanese patients with syndromic endocrine-related disorders for this study, consisting of eight cases of 22q.11.2 deletion syndrome (22q11.2DS), seven cases of Prader-Willi syndrome (PWS), four cases of Turner syndrome (TS), and two cases of Noonan syndrome (NS), and investigated the sensitivity of Face2Gene for each syndrome by analyzing their facial photographs.

Results: All the cases of 22q.11.2DS, PWS, and NS were successfully identified as the correct syndromes with high scores and the highest rank, although facial recognition failed for all the cases of TS.

Conclusions: This study indicates the usefulness of Face2Gene also in the Japanese population. Although early diagnosis and medical intervention are important for improving the quality of life in patients with syndromic endocrine-related disorders, clinical diagnosis of the patients is sometimes difficult because of their rarity and phenotypic complexity. The present studies imply that Face2Gene is a powerful tool for leading to early diagnosis of 22q11.2 DS, PWS, and NS.