AI-augmented Biophysical modeling in thermoplasmonics for real-time monitoring and diagnosis of human tissue infections.

Identifying tissue infections from the body still poses an unprecedented challenge in society. Conventional diagnostic procedures are time-consuming and lack a real-time monitoring mode. This study proposes a system with an Artificial Intelligence (AI)-assisted Thermoplasmonics scheme that has a 57.7% shorter detection time than traditional techniques. The proposed system combines AI with Localised Surface Plasmon Resonance (LSPR) technology. Employing 2,333,481 single-cell transcriptomic profiles from 486 people (107 non-affected, 379 affected), an effective circuitry deep learning setup was designed and validated to analyse Thermoplasmonics sensor data in real-time. The system achieved an overall accuracy of 92.3% It achieved a 42.3% reduction in false positives and a 35.6% decrease in cost per healthcare diagnosis. It also achieved a classification accuracy of 1-94.5%, significantly higher than traditional culture methods' accuracies. The mean detection time was brought down to 42.3 min (SD = 12.8), and 99.7% of the time, all the analyses were done in less than 1 s. Clinical implementation in three major medical centres (n = 1655 cases) demonstrated significant improvements: a 31.3% decrease in the proportion of antibiotic cases misuse and a 23% decrease in hospital stays. Cost-benefit studies showed the system's feasibility in saving $2.8 million per hospital annually.