This paper uses artificial intelligence to analyze standard tissue slides from breast cancer patients, not just to predict recurrence, but to pinpoint the exact microscopic regions associated with high risk. By then physically studying the proteins in these AI-identified "hotspots," the researchers uncovered the specific molecular changes driving recurrence.
Spatial proteomics guided by H&E-based AI reveals recurrence-risk niches in triple-negative breast cancer · Yesung Cho, Ji Hwan Park, Chanil Kim et al. · arXiv:2608.03145
A sophisticated computer model that precisely maps how ultrasound waves travel through the skull and brain, predicting which specific neurons will activate based on the physical forces exerted on them.
An open-source framework for predicting ultrasound neuromodulation: bridging tissue elastomechanics and neuron firing dynamics · Gianmarco Pinton · arXiv:2608.06321
MWF-MIMOSA, a new MRI technique that uses advanced AI to quickly and simultaneously map several important brain properties, particularly focusing on myelin.
MWF-MIMOSA for efficient simultaneous relaxometry and myelin water fraction mapping · Yuting Chen, Yohan Jun, Hyeong-Geol Shin et al. · arXiv:2607.28984
A smarter way to train AI models that learn from biological sequences, like proteins or RNA. Instead of treating all sequences equally, they teach the AI by progressively introducing sequences from close evolutionary relatives to more distant ones, much like a student mastering a subject.
Large AI models are transforming how we analyze DNA, but adapting them for every new biological question is incredibly expensive. This paper investigates when these "frozen" DNA models can be used as-is, finding they excel at some tasks like identifying gene promoters, but struggle with others like predicting how genes are cut and pasted.
Frozen but Not Always Accessible: A Representation Analysis of Genomic Language Models · Nirjhor Datta, Swakkhar Shatabda, M. Sohel Rahman · arXiv:2608.05329
A "digital twin" system that predicts how a cancer patient's head and neck anatomy will change throughout their proton therapy course.
Anticipatory Digital Twins for Online Head-and-Neck Adaptive Proton Therapy via Foundation-Model Registration · Yizhou Wu, Yuheng Li, Xiaofeng Yang et al. · arXiv:2608.00831