AI Diagnoses Diseases by Mining Immune System Data; New AI Model Tracks Brain Aging Speed

Stanford Medicine researchers developed Mal-ID, a machine-learning tool that diagnoses diseases by analyzing B and T cell receptor sequences. Tested on nearly 600 individuals, Mal-ID successfully identified conditions like COVID-19, lupus, and Type 1 diabetes. The tool could also track responses to cancer immunotherapies and refine disease subcategories. The algorithm uses language models trained on proteins to identify threat-recognizing receptors on immune cells. Separately, USC researchers created a new AI model that measures brain aging speed using MRI scans. This tool tracks brain changes non-invasively and correlates faster brain aging with a higher risk of cognitive impairment. The model, a 3D convolutional neural network (3D-CNN), compares baseline and follow-up MRI scans to pinpoint neuroanatomic changes. When tested on adults, the model's calculations correlated with cognitive function changes. It could also distinguish different aging rates across brain regions and between sexes.