THE DEVELOPMENT OF AN ACCURATE MobileNetV2 COMPUTER VISION MODEL FOR THE DIAGNOSIS OF ORTHOPOXVIRUS

As of November 2024, the World Health Organization reported over 95,000 global cases of Mpox, also known as Monkeypox, with 185 deaths since the outbreak began in 2022 (WHO, 24). The virus, which causes flu-like symptoms and rashes, has had widespread impacts on global health, with survivors often experiencing long-term complications such as chronic pain, scarring, and neurological issues (NIH, 23). Economically, Mpox has disrupted industries like travel and tourism, strained healthcare systems, and increased unemployment due to quarantine measures and reduced business operations. Proper diagnosis of Mpox is crucial for managing outbreaks and providing timely medical assistance. Traditional diagnostic methods like RT-PCR and serological assays remain the gold standard but face limitations such as high costs, resource dependency, and inefficiencies in low-resource settings. To address these challenges, we developed a computer vision model using MobileNetV2 that analyzes lesion images to diagnose Mpox. While this model requires lesion images for input, it offers a rapid and cost-effective diagnostic alternative, particularly in high-risk areas where access to traditional methods is limited. By leveraging transfer learning, convolutional neural networks, and key dependencies such as TensorFlow and Keras, the model demonstrated high accuracy with minimal resource demands. Our solution represents an innovative approach to Mpox diagnosis, overcoming the challenges of traditional methods and improving outbreak management.