German scientists have successfully developed the first artificial intelligence (AI) model capable of personalized antibiotic optimization selection for sepsis patients. The relevant results were published in the new issue of the medical journal "NPJ · Digital Medicine". Sepsis is a life-threatening disease that occurs when the body's immune system reacts extremely to infection, and is one of the most common causes of death in Germany, with an average of about 160 deaths per day. Due to the fact that the pathogen is usually unknown at the beginning of treatment, doctors often seek broad-spectrum antibiotics, but this can lead to serious side effects such as kidney or liver damage. Now, German scientists are trying to use AI to assist in selecting more suitable antibiotics to address this challenge. With funding from the health department of the canton of Leifa, a team of scientists from Koblenz University, Koblenz University of Applied Sciences, and Bonn University School of Medicine has developed the first online antibiotic selection model (OptAB). This model is entirely based on AI data-driven and is suitable for sepsis patients considering side effects. Its goal is to maximize the success rate of treatment (measured by SOFA score) while minimizing serious side effects such as kidney and liver damage. The OptAB model is based on an innovative algorithm that combines neural networks with differential equations, and can handle complex patient data characterized by irregular measurements, missing values, and time-dependent confounding factors. OptAB can provide disease progression prediction for vancomycin, ceftriaxone, and piperacillin/tazobactam (combination), and provide accurate predictions of disease course and possible side effects based on experimental values such as creatinine, total bilirubin, and alanine aminotransferase. The optimal antibiotic combination determined by OptAB demonstrates efficacy faster than commonly used broad-spectrum antibiotics in clinical practice, demonstrating the breakthrough innovation brought by the combination of medicine and AI. The researchers also plan to further test the results to test the robustness of the model, so as to apply this model to actual diagnosis and treatment as soon as possible. (New Society)