Artificial Neural Network, Experimental and Numerical Study on Air Cooling Rubber Mattresses for Elderly and Bedridden Patients

Abstract

An artificial neural network (ANN) technique assesses the mattress temperature (body surface) of elderly and bedridden patients. The experimentally recorded body surface temperature data is used in the ANN analysis. The input parameters for the ANN model are sex, body mass index (BMI), room temperature, and cooling air temperature, while the output parameter is the body surface temperature. Many hidden numbers of ANN are trained to anticipate the parameter's optimal output. In addition, a numerical process is performed to consider the influence of room temperature by solving the governing equations for body surface and air temperature. ANN prediction errors for training and testing all datasets fall under the ±2.5% range. The numerical findings are compared to the measured data and the published results with 5.08% - 9.32% errors. The ANN findings were more accurate than the numerical model results. The current research employs the ANN technique for body surface temperature monitoring, effectively minimizing the risk of pressure sores in elderly and bedridden patients.