Diagnosing 'Gapsangseon Cancer' with Artificial Intelligence

AI with Consistency and Interpretability
Clue to Cancer Diagnosis Platform

Meta-Analysis Based Machine Learning

[Asia Economy Reporter Junho Hwang] An artificial intelligence algorithm capable of distinguishing cancer using a meta-analysis-based machine learning algorithm has been developed. The research team stated that they utilized this algorithm to diagnose thyroid cancer with high performance and plan to apply it to other types of cancer as well.

The National Research Foundation of Korea announced on the 3rd that Professor Sungyoung Kim's research team at Konkuk University developed this highly reliable algorithm. The related research results are scheduled to be published in the latest issue of the international journal Briefings in Bioinformatics.

The research team integrated gene expression and related biological pathways of cancer tissues using a meta-analysis-based algorithm and used this as training data for artificial intelligence. Meta-analysis is a technique that integrates statistical data from independent but similar related studies to evaluate the consistency of results and improve statistical accuracy and test power. They combined the machine learning algorithm with meta-analysis to enhance reliability. Additionally, they improved accuracy by adding nonlinear principal component analysis and a dimension reduction model using biological pathways.

The research team reported that when this algorithm was validated on actual thyroid cancer samples, it demonstrated nearly perfect classification performance. Testing various subtypes of thyroid cancer, the algorithm accurately distinguished these subtypes with high precision. Furthermore, considering that thyroid cancer is an aging-related disease, the team confirmed the model's performance in a high-risk group adjusted for aging factors.

The research team stated, "The developed algorithm can be easily applied to cancers other than thyroid cancer," and predicted, "As a regression and pathway analysis-based algorithm, it will be preferred in clinical settings that emphasize causal analysis and consistency of results."

This content was produced with the assistance of AI translation services.