Hybrid AI models for predicting heat distribution in complex tissue structures with bioheat transfer simulation.

The accurate prediction of thermal behaviour in biological tissues is critical for various medical treatments, including hyperthermia, thermal ablation, and tissue engineering. This paper presents a novel deep learning-enhanced bioheat transfer model that integrates a Fractional Legendre wavelet approach to predict thermal effects in engineered tissue constructs precisely. The model incorporates a multi-phase analysis considering key properties such as blood perfusion, thermal conductivity, and metabolic heat generation. Experimental validation was conducted on a 5 cm tissue construct exposed to a 15W heat source over 120 min, with temperature distributions monitored across various regions. Results demonstrated temperature gradients ranging from 37 °C in cooler areas to 48 °C near the heat source. The model achieved a mean absolute error of 2.5 °C and delivered thermal predictions 15 % faster than conventional methods. The proposed integrated deep learning approach enables real-time prediction capabilities that are crucial for precise thermal therapy and tumour ablation applications. The model's versatility was demonstrated across different tissue types, including skin, muscle, fat, and bone, with prediction errors consistently below 0.4 °C across various power inputs (10W-30W). This enhanced predictive capability significantly improves thermal therapy planning and tissue engineering applications requiring precise temperature control.