Mpox-AISM: AI-Mediated Super Monitoring for Mpox and Like-Mpox

TL;DR

Mpox-AISM tackles the challenge of early mpox diagnosis by combining AI with cloud-based, internet-enabled workflows. It leverages self-supervised learning (SimCLR) and data augmentation to train robust classifiers (notably ResNeXt-101_64x4D) that distinguish mpox from similar dermatoses and normal skin, achieving an overall accuracy around $94.51%$ and mpox-specific metrics such as $99.3%$ precision and $99.9%$ specificity. The system includes rash-stage grading, Grad-CAM-based interpretability, and smartphone/PC apps, enabling real-time, accessible screening with potential to curb outbreaks in resource-limited settings. The work highlights SSL as a key driver for performance with limited labeled data and proposes future multimodal extensions (rash images plus epidemiological data) and broader disease coverage for practical deployment.

Abstract

Swift and accurate diagnosis for earlier-stage monkeypox (mpox) patients is crucial to avoiding its spread. However, the similarities between common skin disorders and mpox and the need for professional diagnosis unavoidably impaired the diagnosis of earlier-stage mpox patients and contributed to mpox outbreak. To address the challenge, we proposed "Super Monitoring", a real-time visualization technique employing artificial intelligence (AI) and Internet technology to diagnose earlier-stage mpox cheaply, conveniently, and quickly. Concretely, AI-mediated "Super Monitoring" (mpox-AISM) integrates deep learning models, data augmentation, self-supervised learning, and cloud services. According to publicly accessible datasets, mpox-AISM's Precision, Recall, Specificity, and F1-score in diagnosing mpox reach 99.3%, 94.1%, 99.9%, and 96.6%, respectively, and it achieves 94.51% accuracy in diagnosing mpox, six like-mpox skin disorders, and normal skin. With the Internet and communication terminal, mpox-AISM has the potential to perform real-time and accurate diagnosis for earlier-stage mpox in real-world scenarios, thereby preventing mpox outbreak.

Figures (7)

• Figure 1: Application process of Mpox-AISM. b-e: Application settings in different situations.
• Figure 2: The workflow of this study.
• Figure 3: a,b: Test Accuracy and Train loss trends for the ten models, respectively. c: F1-score for test set of Data_C. d: Precision for test set of Data_C. e: Recall for test set of Data_C. f: Specificity for test set of Data_C. g - j: Confusion matrix of Renext101_64x4D with SimCLR, Renext101_64x4D, Efficientnet_B0 with SimCLR and Efficientnet_B0 for test set of Data_C.
• Figure 4: a: Diagrams of mpox rash (Grade-I, Grade-II, Grade-III and Others). b: Diagrams of human body parts. c: Diagrams of mpox rash at earlier-stage and later-stage.
• Figure 5: Heat maps for eight categories of skin diseases generated by the Grad-CAM method.