Ai-Powered Screening for Psoriatic Arthritis: a Comparative Study With Existing Tools

Abstract

Aim: Psoriatic arthritis (PsA) is common in psoriasis patients but is sometimes overlooked. Delayed diagnosis of PsA can lead to joint erosion, axial damage, and impaired physical function. Screening tools are essential for early diagnosis and selecting the right patients for rheumatological evaluation. We aimed to develop a practical and comprehensive screening tool using the ChatGPT and compare its performance with that of validated questionnaires. Material and Methods: A prospective study was conducted on adult psoriasis patients who had musculoskeletal complaints but were not diagnosed with PsA. Artificial intelligence (AI)-powered PsA screening (AIPS) was developed by selecting questions on peripheral arthritis, axial inflammation, and enthesitis from multivariate analyses conducted via Chat GPT 4.0. The Psoriasis Epidemiology Screening Tool (PEST), the Early Arthritis for Psoriatic Patients Questionnaire (EARP), and the AIPS questionnaires were completed concurrently by all psoriasis patients. All patients were evaluated for PSA diagnosis by three rheumatologists who were blinded to the questionnaire responses. Results: The study included 199 patients, 115 (57.8%) of whom were female. The mean age was 44.4 +/- 13.3 years. PSA was detected in 84 psoriasis patients (42.2%). The sensitivity of the EARP questionnaire, 98%, was greater than those of the AIPS and PEST questionnaires, which had 92% and 83% sensitivity values, respectively. However, the AIPS had a higher specificity at 96% than did the PEST and EARP, with specificities of 91% and 80%, respectively. Discussion: The AIPS questionnaire is an effective tool for screening for PsA, exhibiting high sensitivity and specificity. Artificial intelligence can help screen patients, saving time and money.