Biologists have developed an artificial intelligence (AI) system to analyze data collected from individuals with Chronic Fatigue Syndrome (CFS). The system has made it possible to identify previously unknown biomarkers related to immunity, gut microbiota, and metabolism. The combined analysis of these biomarkers enables the identification of CFS patients with 80–90% accuracy.
According to “Tehsil365”, this was reported by the press service of the Jackson Laboratory (JAX) in the United States.
“As part of our study, we were able to identify individuals with Chronic Fatigue Syndrome with 90 percent accuracy. This is highly important for medical practice, as doctors still lack reliable tools to diagnose the disease. Some even question the existence of this syndrome due to the previous absence of a specific set of biomarkers,” said Professor Derya Unutmaz.
Over the course of four years, Unutmaz and his colleagues studied the health conditions of fifteen individuals diagnosed with CFS and a control group of healthy volunteers matched by age and gender. During this period, they regularly collected samples of microbiota, blood, and other biological fluids while also monitoring symptom progression, lifestyle factors, and other potential influences.
To analyze this medical and social data, the researchers developed a specialized AI system based on deep neural networks. This system not only identifies individuals with CFS but also explains the reasoning behind its decisions. Thanks to this machine learning approach, the researchers discovered dozens of previously unknown biomarkers related to the development of CFS and its various aspects.
For example, they found that dysfunction in specific subtypes of B and T cells is associated with the onset of the syndrome. Additionally, Dysosmobacter welbionis—a bacterium previously linked to diabetes and obesity—was found to cause sleep disturbances in individuals with CFS. Meanwhile, Faecalibacterium prausnitzii and certain dendritic cells of the immune system were associated with increased pain sensitivity.
Based on these biomarkers, the scientists developed an algorithm to detect CFS and tested it on data from other patients. Despite significant differences in biomarker profiles, the system successfully identified approximately 80% of CFS patients. According to the researchers, this approach could enable earlier diagnosis of CFS and support the development of effective treatment strategies in the future.