Improved Machine Learning for Thyroid Cancer Diagnosis
Author Information
Author(s): Książek Wojciech
Primary Institution: Cracow University of Technology
Hypothesis
Can the naked mole-rat algorithm enhance machine learning models for thyroid cancer diagnosis?
Conclusion
The study developed a machine learning model that achieved a classification accuracy of 81.82% for thyroid cancer diagnosis.
Supporting Evidence
- The LightGBM classifier achieved a classification accuracy of 81.82%.
- The study utilized a dataset from Shengjing Hospital of China Medical University.
- SHAP values were used for model explainability.
- The naked mole-rat algorithm was applied for parameter optimization and feature selection.
- Ten well-known classifiers were tested in the study.
Takeaway
This study created a smart computer program to help doctors tell if thyroid tumors are cancerous or not, using a special method inspired by nature.
Methodology
The study used a dataset of 1232 records and applied ten machine learning classifiers, optimizing parameters and selecting features using the naked mole-rat algorithm.
Limitations
The research was based on a single dataset from China, which may limit the generalizability of the findings.
Participant Demographics
83.77% of participants were women, with an average age of 46.61 years.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website