Applications of Artificial Intelligence-Based Models and Multi-Linear Regression for the Prediction of Thyroid Stimulating Hormone Level in the Human Body
Abstract
In this work, three different models were employed, which includes two artificial intelligence (AI) based modes (Adaptive neuro-fuzzy inference system (ANFIS) and (Artificial neural network (ANN)) and a linear model (Multilinear regression analysis (MLR)) for the prediction of thyroid hormone (TSH) using different macro-elements and vitamins as the input parameters. The predicted and experimental results were checked using four different performance indices namely; mean square error (MSE), correlation coefficient (R), root mean square error (RMSE) and determination coefficient (R2). The result obtained indicated the ability of the AI-based models (ANFIS and ANN) over the linear model (MLR) having an R2-values higher than 0.8 in both the testing and training stages. The result equally demonstrated that, based on the R2-values ANFIS outperformed MLR and ANN and enhance their performance efficiency up to 30% and 9% respectively in the testing stages. The overall results depicted the satisfaction and reliability of the non-linear models (ANFIS and ANN) for the simulation of TSH.