Evaluating the Clinical Utility of Face-Recognition Technology in Syndrome Diagnosis

Clinicians see over 2000 patients per year who have rare syndromes in Manchester NHS clinics. In the majority of these individuals the cause is unknown. Many will have a genetic cause, but knowing which genes to test and being able to access these tests is difficult. When a patient comes to the clinic, details of their medical and developmental history are collected , and they are examined in detail to look at their investigation results. New tests may also be ordered for patients. In many cases, subtle differences in physical features, especially facial features may provide an important clue to the underlying diagnosis. However, because many of the conditions seen are so rare and doctors may not have seen that particular condition before, the diagnosis may not be made immediately at the appointment. In those cases, permission will be sought to take photographs so that further opinions can be sought within the department or by sharing with national and sometimes international colleagues. This is routine practice. Where consent has been obtained for photos, these are then first reviewed in a departmental case-review meeting. They may then also be presented at regional, national or even international meetings aimed at syndrome diagnosis with patient consent.

The study aims to recruit patients who are attending clinics for syndrome diagnosis, and who have differences in their facial features. Such patients will undergo a full routine diagnostic work-up as outlined above. Following that, if patients have consented to having photographs taken as part of standard care, they will be asked if they would consent to upload of the facial photographs to a digital face recognition system, along with upload of a list of key clinical features, to see which diagnoses are suggested by this software.

A group of Inherited Metabolic Disease patients with known diagnoses will be included under the Faces sub-study to establish whether the technology may help to define their phenotype. This group of patients would be asked to send photos to the research team either by email via a secure email address or by post. The study would request one facial photo of each of the biological parents (where applicable and available).

There will first of all be routine discussion of patients and photographs in a case review meeting as per standard practice. Differential diagnoses will be formulated and recorded based on this. Following this, the facial photo will be uploaded to a face recognition system and suggested diagnoses from this recorded. Any diagnostic suggestions considered worthy of investigation will be followed up in line with standard practice.

The investigators will then determine whether this was made a) in the standard way b) only suggested by the face recognition software or c) utilising the two methods together.