Supplementary Material for: Using smartphone data to frequently measure atopic dermatitis severity and understand triggers: A decentralized 12-week observational study

Abstract Introduction Atopic dermatitis (AD) severity is traditionally evaluated during in-office consultations; however, this does not provide continuous monitoring and any intermittent flare/remission cycles are usually not recorded. The aim was to apply smartphone technology to evaluate AD severity; to explore if severity based on highly frequent sampling of photographs are associated with patient reported outcomes, and to investigate disease fluctuations and trigger associations based on passively collected environmental data. Methods In this 12-week decentralized observational study adult patients with AD were recruited online and used a tailored remote clinical trial platform app to perform all study tasks including capturing photographs and completing the patient-oriented eczema measure (POEM) weekly. AD severity was assessed based on photographs using the Eczema Area and Severity Index (EASI) and SCORing Atopic Dermatitis (SCORAD). Geographical location collected in the app was used to retrieve data on ambient temperature and carbon monoxide (CO), a common air pollutant. Results A total of 42 patients (35 women) were recruited online. A total of 712 photographs were taken, with an average of 17 photographs per participant. Photographic SCORAD (r=0.450) and EASI (r=0.206) were significantly associated with subjective severity POEM. Patients experiencing AD fluctuation (n=10) based on SCORAD had significantly higher risk of also having a psychiatric disorder (60 vs. 17%, p=0.008). Anxiety was significantly associated with disease fluctuation based on EASI (40 vs. 7%, p=0.01) and a tendency was observed for depression (40 vs. 13%, p=0.06). Decreasing temperature was significantly associated with higher POEM (estimate -0.18, p=0.012) and EASI score (estimate -0.14, p=0.007), but not with SCORAD. High levels of CO were significantly associated with higher SCORAD (estimate 15.9, p<0.001). Conclusion In this small study with predominance of young adults, primarily women, we were able to recruit patients and monitor AD entirely remotely via smartphone-enabled photographic assessments, patient reported outcomes and passively collected environmental data without physical contact between patient and investigator.