With the advent of smart healthcare services, rapid and automated diagnosis from images of skin lesions is critical to combat fast-spreading viruses such as monkeypox (Mpox or MPX) and significantly improve public health. The recent cases in Thailand reporting a suspected first case on August 21, 2024, and Sweden on August 14, 2024, among others, highlight the pandemic threat of Mpox. This study presents the Deep Digital Twin Services for Personalized Treatment (D2T-PT) model, which combines transfer learning and Digital Twin (DT) technology to improve the accuracy of Mpox detection and real-time monitoring, supported by the Squeeze-and-Excitation Block (SEB) attention mechanism, which opens up new horizons for personalized healthcare. Convolutional Neural Network (CNN) models were tested on the Monkeypox Skin Lesion Dataset (MSLD), with the advanced adaptive NasNetMobile model achieving excellent results: 100% recall, 98% ROC score, 97.78% accuracy with precision of 95%. This robust model enables physicians to make early and accurate Mpox diagnoses and monitor patient response to treatment in real-time, ultimately helping to contain the spread of the virus.