A Comprehensive Approach to Predict Early Diagnosis of Hyperthyroidism
摘要
In this digital era, machine learning has become one of the enablers across multiple domains including healthcare. This work investigates the utility of ML algorithms in prediction of hyperthyroidism, a prevalent but under diagnosed ailment. Using different machine learning techniques and conducting a performance comparison to identify which model gave the best output using a set of hyper parameters that were manually chosen. Among the algorithms tested, Gradient Boosting had best performance in predicting hyperthyroidism (accuracy 99.8%). This model continuously demonstrated 99.6% accuracy of 0.998, recall value of 0.997, F1 score of 0.997, and Receiver operating curve (ROC AUC) of 0.995 with cross-validation. These measurements underscore the model's remarkable capacity to identify hyperthyroidism, hence bolstering its promise as a dependable instrument for prompt diagnosis. Moreover, the study represents that the dataset imbalance limits the performance disparities between classes. Despite this challenge, the excellent performance of Gradient Boosting highlights its applicability in developing robust predictive models in the healthcare industry. This study highlights the prospective of machine learning to improve the early detection and accurate prediction of hyperthyroidism, paving the way for better patients and better healthcare.