Evaluation Of A Heatmap-Based Mobile Application For Assessing Rosacea Severity Compared To Clinical Grading: A Blinded Study

This study aimed to evaluate the effectiveness of a heatmap-based mobile device application in assessing rosacea severity, comparing its accuracy to standard clinical grading by dermatologists

Private practice in Baltimore, MD ,US

Forty-eight rosacea patients’ facial images were analyzed using a mobile app generating heatmaps to highlight redness and inflammation. Standardized images were taken with an iPad, with light and distance controlled. Five blinded researchers evaluated the heatmaps, unaware of clinical histories, and compared these to severity grades assigned by dermatologists using the National Rosacea Society’s grading system. Statistical analyses, including intraclass correlation coefficients (ICC) were assessed agreement between app-based assessments and clinical evaluations. Heatmap amplitudes (0.4 to 1.10) were adjusted to identify the optimal sensitivity for accurate rosacea detection

Each of the three readers disagreed on the subjective clinical grading for many of the patients: among 47 patients initially recruited into the study, the readers agreed on clinical gradings for only 25 of them, thus excluding 22 patients from the study. Among these 25 patients (17 with mild rosacea, 5 with moderate, and 3 with severe), a Kruskal-Wallis test has not yet showed a significant difference (p = 0.156) between the heatmap amplitudes required for detection of the three clinical levels of rosacea. However, given the very limited sample size, recruitment of more moderate to severe rosacea patients into the study could likely reveal significant differences between the three groups.

Because the subjective clinical grading sometimes yielded inconsistent results among different readers, an objective tool would prove useful to provide a consistent and reliable grading. The SPARCA rosacea app has the potential to objectively distinguish between different types of rosacea, ultimately standardizing the diagnosis of the condition.