GestaltMatcher facilitates rare disease matching using facial phenotype descriptors

Rare genetic disorders affect more than 6.2% of the global population1. Because genetic disorders are rare and diverse, accurate clinical diagnosis is a time-consuming and challenging process, often referred to as the 'diagnostic odyssey'2, and all informative clinical features have to be taken into consideration. [...]pediatricians and geneticists increasingly use such next-generation phenotyping tools for differential diagnostics in patients with facial dysmorphism. [...]most phenotypes have not been modeled because sufficient data are lacking. [...]the need to train on large numbers of photographs is a major limitation for the identification of ultra-rare syndromes. The distance between cases in the CFPS quantifies the similarities between the faces, thereby matching patients with known syndromes or identifying similarities between multiple patients with unknown disorders and thereby helping to define new syndromes. Because GestaltMatcher quantifies similarities between faces in this way, it addresses all three of the limitations described above: (1) it can identify 'closest matches' among patients with known or unknown disorders, regardless of prevalence; (2) it does not need new architecture or training to incorporate new syndromes; and (3) it creates a search space to explore similarity of facial gestalts based on mutation data, which can point to shared molecular pathways of phenotypically similar disorders.