Identification of important symptoms and diagnostic hypothyroidism patients using machine learning algorithms
Salahuddin Rakhshani Rad1, Zahra H Mohammadi2, Mahdieh J Zadeh3
1Department of Biostatistics and Epidemiology, School of Public Health.
Annals of Medicine and Surgery (2012)
|June 7, 2024
Summary
Machine learning accurately identifies hypothyroidism using common symptoms, potentially reducing the need for blood tests. This approach aids physicians in diagnosing this common endocrine disease more efficiently.
Area of Science:
- Endocrinology
- Medical Informatics
- Machine Learning
Background:
- Hypothyroidism is a prevalent endocrine disorder often diagnosed late due to vague symptoms.
- Traditional diagnosis relies on blood tests, which can be costly and time-consuming.
- Machine learning (ML) offers advanced tools for improving diagnostic accuracy in medicine.
Purpose of the Study:
- To predict and identify key symptoms of hypothyroidism using ML algorithms.
- To evaluate the efficacy of ML in diagnosing hypothyroidism based on symptom presentation.
Main Methods:
- A cross-sectional study analyzed 1296 individuals presenting with hypothyroidism symptoms.
- Machine learning models including random forest, decision tree, and logistic regression were employed.
- Diagnosis was confirmed via thyroid-stimulating hormone testing, with outcomes binary (hypothyroid/euthyroid).
Main Results:
- Key symptoms for identifying hypothyroidism included tiredness, cold intolerance, jaundice, and weight gain.
- The random forest algorithm demonstrated superior performance with an accuracy of 0.83.
- Sensitivity and specificity for random forest were both 0.88, indicating high diagnostic capability.
Conclusions:
- ML algorithms can effectively identify hypothyroidism patients based on common symptoms, offering high accuracy.
- This approach may reduce reliance on blood tests, lowering costs and patient stress.
- Increased physician adoption of ML tools could streamline hypothyroidism diagnosis.
