🦠 Scientists Develop a Universal AI Blood Test for HIV, COVID, and Diabetes
Researchers at Stanford University have created an AI model capable of detecting human diseases (whether in active phases or previously encountered) and identifying vaccines received—all through a single blood test. According to the developers, this system could replace the numerous tests typically prescribed by doctors.
🧬 How the Model Works
Diseases and vaccines activate immune cells known as B and T lymphocytes. To combat threats like viruses effectively, these cells must "remember" them, which is accomplished through specialized receptors. Initially, these receptors are generated randomly, but after encountering a threat, they adapt to recognize it more efficiently. Using AI algorithms, scientists have now learned to decode these "imprints" of the immune response (a set of biomarkers).
The new "machine learning system for immunological diagnostics" (abbreviated as Mal-ID) combines two approaches. First, it identifies identical biomarker sequences in people with the same disease. Second, it uses AI to analyze proteins, such as how language models process texts, finding patterns and meaning within them.
"The immune system is a natural diagnostic tool. If we can understand how it works, we can replicate its processes," says Victor Greiff, an immunologist from the University of Oslo.
🦠 Diseases Identified Using Mal-ID
The testing involved 542 participants, including individuals with HIV, lupus, type 1 diabetes, COVID-19, those recently vaccinated against the flu, and 220 healthy individuals in a control group. Based on blood samples, Mal-ID accurately determined patients' conditions with 98.6% precision.
According to study leader Maxim Zaslavsky, this method not only speeds up and costs down clinical testing but also enables personalized treatment for various diseases, sparing patients from suffering caused by unsuitable medications.
More on the topic:
🔬 How AlphaFold by Google DeepMind is Transforming Biology and Pharmaceuticals