In the WHAT HEALTH !? newsletter, find the latest news that has made the buzz, cutting-edge trends and groundbreaking innovations in the digital health field.
This week, researchers at Charité Hospital in Berlin have unveiled a promising breakthrough in oncology: an artificial intelligence model called CrossNN, capable of classifying 170 types of tumors with up to 97.8% accuracy! By analysing DNA methylation profiles, even from incomplete samples, the system outperforms conventional diagnostic methods, including histological analysis in certain cases.
Initially applied to brain tumors, where it achieved 99.1% precision, the model was tested on more than 5,000 tumors and is based on a reference database of over 8,000 tumor profiles. Despite its relatively simple and interpretable architecture, CrossNN offers enhanced diagnostic certainty and can even work with cerebrospinal fluid samples, without requiring surgical tissue extraction.
Clinical trials are now being launched across eight German hospitals to assess its integration into routine care.
A development to watch closely, at the crossroads of AI, epigenetics, and precision medicine.
➡️ Find all the latest news in the columns of your weekly.
Happy reading,
