0
0
A groundbreaking artificial intelligence (AI) tool can now analyze blood samples to diagnose infections, autoimmune diseases, and vaccine responses simultaneously, according to a new study published in Science on February 21.
The AI algorithm examined immune cell receptor genes to identify patients with COVID-19, HIV, type 1 diabetes, lupus, those who had recently received a flu vaccination, and healthy individuals.
“These receptor sequences are unique sources of potential diagnostic information,” said Scott Boyd, MD, PhD, a professor of pathology at Stanford University and one of the senior authors of the study. “They haven’t really been used to diagnose anything apart from cancers that originate from B cells and T cells at present.”
While this research serves as a proof of concept, the AI tool has the potential to distinguish between diseases with similar symptoms, according to study co-author Maxim Zaslavsky, PhD, a computer science postdoc at Stanford University.
Decoding the Immune System’s Language
B-cell and T-cell receptors record immune activity, providing a unique diagnostic opportunity. If researchers can develop a method to effectively interpret these immune responses, “we could have an amazing diagnostic tool” capable of identifying multiple health conditions from a single test, said Ramy Arnaout, MD, DPhil, an associate professor of pathology at Harvard Medical School and associate director of the Clinical Microbiology Laboratories at Beth Israel Deaconess Medical Center in Boston. Arnaout was not involved in the study.
However, individual immune responses vary significantly. Even if two people have the same disease, their immune cell receptors may differ.
“Traditional sequencing approaches struggle to group immune receptors that may look slightly different but recognize the same target,” Zaslavsky explained. Large language models (LLMs) — the same AI technology that powers ChatGPT — “excel at this type of pattern recognition.”
“The idea here is that everyone’s repertoire of B-cell and T-cell receptors is like dialects of the same language,” Arnaout said. By analyzing immune sequences from multiple patients, LLMs can detect commonalities and interpret the “language” of the immune system.
Combining B-Cell and T-Cell Data
Boyd, Zaslavsky, and their team combined multiple AI models to analyze 23.5 million T-cell receptor sequences and 16.2 million B-cell receptor sequences from 593 individuals. The study group included 63 COVID-19 patients, 95 HIV patients, 86 lupus patients, 92 type 1 diabetes patients, 37 individuals who recently received a flu vaccine, and 220 healthy controls.
The AI algorithm, called Machine Learning for Immunological Diagnosis, showed the highest accuracy when analyzing both B-cell and T-cell receptor data together. Among the 542 samples with paired data, the AI tool achieved an area under the receiver operating characteristic curve (AUROC) of 0.986 and 85.3% accuracy.
When using only B-cell receptors, the AI tool reached an AUROC of 0.959 and 74.0% accuracy. With only T-cell receptors, it achieved an AUROC of 0.952 and 75.1% accuracy.
While past research has focused on either B cells or T cells, Zaslavsky emphasized that combining both provides a more comprehensive view of immune activity. B-cell receptor sequences were particularly useful in identifying COVID-19, HIV, and flu vaccination responses, while T-cell receptor sequences were more informative for diagnosing lupus and type 1 diabetes.
Clinical Applications and Future Prospects
While this AI tool is still far from clinical use, it represents an important step toward more efficient diagnostics, said Arnaout. Multiplex diagnostic tests, which detect multiple biomarkers from a single sample, could streamline and shorten the diagnostic process. As sequencing costs continue to decline, such AI-driven diagnostics could become more cost-effective by reducing the need for multiple lab tests.
Beyond diagnosis, this AI tool could have broader applications. Just as genomic sequencing has revolutionized cancer treatment, immune cell receptor sequencing may pave the way for more targeted treatments for autoimmune diseases, Zaslavsky suggested. By mapping immune system activity, the tool could help identify the most effective treatments for individual patients, potentially reducing trial-and-error approaches.
“This is all still in the research phase, but we hope this approach could follow the path of genomic sequencing in transforming medicine,” Zaslavsky said.
The study received funding from the National Institutes of Health and other institutional grants. Several authors, including Boyd and Zaslavsky, are co-inventors of a related patent. Boyd has also served as a consultant for pharmaceutical companies such as Regeneron, Sanofi, Novartis, Genentech, Visterra, and Janssen Pharmaceuticals and owns stock in AbCellera Biologics. Some other authors have financial relationships with biotechnology firms. Arnaout disclosed no conflicts of interest.
Disclaimer: This article summarizes research findings and does not constitute medical advice. The AI tool discussed is in the experimental phase and is not yet approved for clinical use. Always consult a healthcare professional for medical diagnosis and treatment.
