Evaluating Legit.Health Plus Support for Improving Diagnosis of Generalized Pustular Psoriasis and Other Skin Conditions Among Primary Care Physicians and Dermatologists

A Multi-Reader Multi-Case (MRMC) Study for Assessing the Impact of Legit.Health Plus on the Clinical Assessment of Generalized Pustular Psoriasis and Other Skin Conditions by Healthcare Professionals.

This study aims to determine if an artificial intelligence (AI) medical device can help healthcare professionals more accurately diagnose rare and complex skin conditions. Dermatological issues are common in primary care, but there is often a gap in diagnostic accuracy between general practitioners and specialists, which can lead to treatment delays for serious conditions like Generalized Pustular Psoriasis (GPP) and Hidradenitis Suppurativa (HS).

The researchers hypothesized that the AI device would enhance the diagnostic accuracy of healthcare professionals for GPP and other dermatological conditions. To test this, the study followed a prospective observational design involving 15 practitioners, including both general practitioners and dermatologists.

During the study, participants were asked to evaluate 100 clinical images. For each case, they first provided a diagnosis based on the image and patient history alone. They were then shown the AI's analysis-which included the top five suggested diagnoses and confidence levels-and asked if they would like to adjust their initial assessment.

The primary question the study sought to answer was whether the information provided by the AI device could significantly increase the number of correct diagnoses made by these professionals, particularly for rare diseases that are often difficult to identify in a standard clinical setting

This investigation is structured as a multi-reader multi-case (MRMC) study. A cohort of 15 healthcare professionals, including 11 primary care physicians and 4 dermatologists, acted as the "readers". These readers evaluated a "case" set of 100 clinical images to assess diagnostic performance both with and without the assistance of the AI device.

Study Design and Technical Methodology The research was conducted as a prospective observational and cross-sectional study. It utilized a "physician-as-their-own-control" design to measure the impact of Artificial Intelligence (AI) on diagnostic performance.

• Intervention Workflow: Participants accessed a dedicated web platform where they were presented with 100 clinical cases.
• Evaluation Steps: For each case, practitioners first evaluated a clinical image alongside anamnesis data (e.g., allergies, systemic symptoms) to provide an initial diagnosis.
• AI Support: Subsequently, they were presented with the AI's top 5 suggested diagnoses and associated confidence levels before making a final assessment.
• Image Sourcing: Cases consisted of high-quality images of Generalized Pustular Psoriasis (GPP), Hidradenitis Suppurativa (HS), and various differential "look-alike" conditions such as subcorneal pustular dermatosis and palmoplantar pustulosis.
• Data Sources: These images were curated from public dermatology atlases and internal research databases from the sponsor.

Quality Assurance and Data Management

To ensure the scientific integrity and reliability of the findings, several quality control measures were implemented:

• Data Validation: A validation process of the clinical data was carried out by running computer filters based on validation rules.
• Error Detection: These filters automatically identify missing values or inconsistencies. This was supplemented by manual editing and exploratory statistical techniques to detect logical errors and inconsistencies.
• Monitoring Plan: The investigation was overseen by a designated clinical monitor appointed by the sponsor.
• Remote Supervision: Monitoring activities included remote meetings every three months to review study progress and ensure ongoing compliance.
• Bias Minimization: Random selection of healthcare professionals (HCPs) was used to ensure that outcomes were not influenced by pre-existing participant characteristics.
• Standardization: The use of standardized procedures ensured that all participants were evaluated in the same way.

Statistical Analysis Plan

The primary goal of the analysis was to quantify Top-1 accuracy, sensitivity, and specificity for both general practitioners and dermatologists.

• Comparative Metrics: The analysis focused on the absolute metric values compared against the state of the art, as well as the percentage of variation attributable specifically to the use of the device.
• Subgroup Analysis: A dedicated analysis was performed for rare diseases, including GPP, Acne Conglobata, and Pemphigus Vulgaris, to assess the device's utility in high-complexity, low-incidence conditions.

Ethical and Confidentiality Framework The study adhered to UNE-EN ISO 14155:2021, the Declaration of Helsinki, and the General Data Protection Regulation (GDPR).

• Anonymization: No data enabling the personal identification of participants or patients was collected.
• Encryption: All information was managed securely in an encrypted format.
• End of Study: All information stored in the device will be totally and permanently deleted at the end of the study.

Inclusion Criteria:

• Board-certified general practitioners and dermatologists, regardless of their professional experience.
• Good quality images of patients with GPP.
• Good quality images of patients with HS.
• Good quality images of patients with pathologies that can be confused with GPP or HS, leading to a wrong diagnosis.

Exclusion Criteria:

• Images of patients with pathologies different from GPP or HS that can be easily identified.