Robust Analysis of Hypothyroidism Detection Using Ensemble Modeling Techniques

Abstract

Hypothyroidism is a prevalent medical condition that demands timely detection and cure because early detection can contribute to decreasing death rates. To check the thyroid level in the human body current diagnostic methods, for example, TSH, T3, FTI, etc are performed, which places a heavy workload on paramedical and healthcare professionals. This article dives deep into the application of Machine Learning algorithms for the detection and classification of hypothyroidism automatically. To develop such a system it is essential to increase the precision rate and decrease the error rates of the ML model. This study was done on six ML classifiers J48, Random Forest, Random Tree, Naïve Bays Multiclass and MLP, which produced significant results for their comparative analysis.  After evaluation, it was revealed that models built up to predict the issue using a MLP and random Tree generated the highest level of accuracy, 99.9565% and 99.8692% respectively. On the other hand, it was also seen that the time MLP took was way more than any other model. So far, our research shows the best results for MLP Accuracy and the lowest error rate. This effort provides vital insights into the possible Machine learning techniques for the detection and classification of Hypothyroidism making a pathway for more accurate and realistic diagnostic tools in clinical practices.

Keywords: Hypothyroidism, Ensemble Techniques, MLP, Naive Bayes, Random Forest, Multiclass, J48