Development of self-phenotyping tools to empower patients and improve diagnostics

Deep phenotyping is important for rare disease diagnostics, often using the Human Phenotype Ontology (HPO). Patients are an under-utilised source; to facilitate self-phenotyping we previously "translated" HPO into plain language. However, self-reported data has not been assessed to date for diagnostic efficacy nor patient opportunities for collaboration with clinical diagnosticians.

Approximately one-third of the HPO is translatable to layperson HPO terms. We used two instruments to evaluate patients' use of the layperson HPO - the GenomeConnect survey mapped to layperson HPO terms, and a natively layperson HPO-based application, Phenotypr. We created phenotype profiles for 7344 Mendelian diseases for each instrument, representing the theoretical maximum performance that patients might achieve. To explore real-world capabilities, we randomised participants with diagnosed genetic diseases to use GenomeConnect, Phenotypr, or both instruments. For each diagnosed disease, we compared the layperson HPO profile to the patient-completed profile for each instrument. We also performed qualitative interviews.

Both instruments performed well in retrieving the correct disease (area under the curve 0.991 and 0.954). Profiles resulting from participant responses to the GenomeConnect survey were more accurate than Phenotypr. Phenotypr had a tighter distribution of scores for respondents who did both instruments and was more precise. We conducted interviews and generally participants preferred the GenomeConnect multiple choice format over the autocomplete Phenotypr format and valued the opportunity to contribute to their diagnostic workup.

Our results demonstrate that individuals are capable of providing rich phenotype data and suggest that self-phenotyping could be used to supplement profiles created by clinicians, providing a more robust foundation for rare disease patient engagement and reducing burden on clinical diagnosticians. Further user interface innovation is needed to garner maximally useful phenotype data from patients.

This research was supported by PCORI #HSRP20181624. The Manton Foundation provided philanthropic support for patient studies. NIH P50HD105351 provided support for genotyping patients at BCH.