Clinical Validation of Digitally Acquired Clinical Data and Machine Learning Models for Remote Measurement of Psoraisis and Psoriatic Arthritis: A Proof-of-Concept Study
- Description
This dataset contains information from a cross-sectional study of 104 adult patients with psoriasis (PsO) and psoriatic arthritis (PsA) and healthy controls. The goal of the study was to develop and validate the Psorcast app, a suite of novel, sensor-based smartphone assessments that can be self-administered to measure symptoms of psoriatic disease. Participants were recruited from the NYU Langone PsA Center and the Center for Skin and Related Musculoskeletal Diseases from the Brigham and Women’s Hospital (BWH) in Boston, Massachusetts. Demographic and clinical evaluation information were recorded on a web form; clinical assessments included percent body surface area (BSA) affected by PsO, target plaque physician global assessment (PGA; target PGA), location(s) of active PsO, nail exam, a tender and swollen joint count (TJC; SJC), enthesitis count, and dactylitis count.
Digital assessments completed on the Psorcast app included the following battery of assessments, the results of which were compared to in-clinic assessments:
- PsO Draw: patients reported BSA by drawing pixels on the body to represent affected areas, which was then compared to the in-clinic physician BSA estimate.
- PsO Area Photo: patients took a photo of an area of their body that they determined to contain a plaque representative of their current disease state, which was then rated by both on-site study physicians and a panel of remote physicians on a scale from 0 (clear) to 4 (severe).
- Finger Photos: patients took photos of their hands; physicians then rated each fingernail for the presence/absence of psoriatic changes in clinic.
- Painful Joint Count: patients selected at least one area of the body that contained painful joints and were asked to identify which joints in that area were painful or tender.
- Digital Jar Open: patients used sensors to measure full-arm rotational motion in order to detect upper extremity involvement.
- Thirty-Second Walk: patients walked for 30 seconds with their smartphone in their pocket or waistband; motion sensor data was used to extract gait features and determine lower extremity involvement.
- Timeframe
- 2019 - 2021
- Geographic Coverage
-
Massachusetts - BostonNew York (State) - New York City
Access
- Restrictions
-
Free to All
- Instructions
- Deidentified validation study datasets are available on the Synapse platform. Guidelines to reproduce the analysis results and figures, as well as app software, can be found on GitHub.
- Other Resources
-
Psorcast
App Software