Jun Ding, Simon Rousseau, Gregory Fonseca and collaborators have developed a powerful new tool called RAMEN that could transform how we understand and treat complex diseases like COVID-19, sepsis, and COPD. RAMEN uses a unique combination of two AI techniques—absorbing random walks and genetic algorithms—to analyze clinical data and uncover relationships between symptoms, lab results, and disease outcomes that traditional methods miss.
What makes RAMEN innovative is its ability to build smart, disease-focused maps of clinical variables (called Bayesian networks) that not only find direct links but also reveal subtle, indirect connections that may be critical for diagnosis and treatment. The researchers tested RAMEN on thousands of patient records and found that it consistently outperformed existing methods in identifying key indicators of disease severity, some of which were later confirmed using gene and protein data.
Because RAMEN works across different diseases and doesn’t rely on prior assumptions, it holds promise for improving personalized medicine—helping doctors tailor care based on patterns hidden in patient data. Its speed and scalability make it especially valuable for analyzing large healthcare datasets, paving the way for faster insights into disease mechanisms and better patient outcomes.
Read More
Efficient and scalable construction of clinical variable networks for complex diseases with RAMEN. Xiong Y, Wang J, Shang X, Chen T, Fraser DD, Fonseca GJ, Rousseau S, Ding J. Cell Rep Methods. 2025 Apr 21;5(4):101022. doi: 10.1016/j.crmeth.2025.101022. PMID: 40215965