Thyroid Disease Classification Based on Blood Tests Using Machine Learning Techniques

EasyChair Preprint 13616

Abstract

The thyroid gland is an important organ in the human body responsible for secreting hormones that support and stabilize the metabolic cycle. The dynamic development of technology and progress in monitoring thyroid diseases constitute a rapidly evolving area of research. The aim of the study was to classify thyroid diseases based on blood test results using machine learning techniques. Literature analysis, anatomical-physiological aspects of the thyroid, and advanced technologies such as machine learning were presented as powerful tools in monitoring and diagnosing thyroid disorders. A literature analysis was conducted based on six articles from the machine learning and thyroid disease classification domain. Laboratory diagnostics based on blood tests, as well as statistics and epidemiological data shaping the understanding of existing thyroid-related issues, were presented. The classification stage described the database, its content, and size. The tools, technology, and learning models used were also presented and discussed. LGBM, XGBoost, and Random Forest models were highlighted. Subsequently, optimization of these three models was performed. The Random Forest model provided the highest accuracy, F1-score and BACC. Classifying thyroid diseases based on blood test results using machine learning can have a significant impact on public health. This modern approach enables early disease detection, leading to shorter diagnostic times and faster treatment implementation.

Keyphrases: Artificial Intelligence, Blood tests, Classification, machine learning, thyroid diseases

@booklet{EasyChair:13616,
  author    = {Michalina Sierko and Sandra Śmigiel and Damian Ledziński},
  title     = {Thyroid Disease Classification Based on Blood Tests Using Machine Learning Techniques},
  howpublished = {EasyChair Preprint 13616},
  year      = {EasyChair, 2024}}