OPTIMIZING MEDICAL DECISION-MAKING FOR HYPOTHYROIDISM WITH EXPLAINABLE ARTIFICIAL INTELLIGENCE

Precise diagnoses are essential for the effective management of hypothyroidism. However, the subjective nature inherent in interpreting diagnostic tests poses a significant challenge. In this paper, we examined and evaluated the utility of the machine learning algorithms SHAPASH and LIME, with explainable AI, in addressing that challenge. The objective was to forecast hypothyroidism, a thyroid gland dysfunction, utilizing attributes derived from blood biomarkers and explain the model's resulting predictions. These attributes include thyroxine, thyroid-stimulating hormone, free thyroxine index, total thyroxine, and triiodothyronine. The results demonstrate the efficacy of employing these algorithms in accurately classifying hypothyroidism and providing diagnostic support, achieving an accuracy rate of 99.16%.