AI Model Detects 13 Types of Cancer with 98.2% Accuracy

AI Model Detects 13 Types of Cancer with 98.2% Accuracy

A new artificial intelligence (AI) model, EMethylNET, has demonstrated the ability to detect 13 different types of cancer with an accuracy of 98.2% using DNA data from tissue samples, according to a recent study by Decrypt, researchers at the University of Cambridge in the U.K. Published in Biology Methods and Protocols, the findings suggest that this technology could significantly accelerate early cancer detection, diagnosis, and treatment.

The study focused on DNA methylation, a chemical process occurring early in cell growth, including in cancer cells. Researchers trained the machine learning model to identify early cancer structures and pathways using data from over 6,000 tissue samples representing 13 cancer types, including breast, lung, and colorectal cancers, sourced from The Cancer Genome Atlas. The model was then tested on more than 900 samples from independent datasets, achieving over 98% accuracy in classifying cancer types and non-cancer samples.

The AI model integrates XGBoost for feature selection and a deep neural network for classification, allowing it not only to detect cancer accurately but also to provide insights into the body’s regulation of non-genetic factors that mutate normal cells into cancer cells. The study identified 3,388 methylation sites linked to cancer-related genes and pathways.

While the research is promising, the authors caution that further study and testing are required before clinical use. The team is now working to adapt the model for liquid-tissue samples, potentially enabling non-invasive early cancer screening. report from News Medical.

As AI continues to advance in healthcare, EMethylNET represents a significant step towards earlier, more accurate cancer diagnosis, which could have profound implications for public health. According to the International Agency for Research on Cancer, over 19 million new cancer cases are diagnosed annually, with 10 million cancer deaths.