Methods of preoperative risk screening based on artificial intelligence using medical history and laboratory data

the article is devoted to the analysis of modern approaches to preoperative risk screening using artificial intelligence algorithms that integrate anamnestic and laboratory data. The relevance of the transition from traditional scoring scales to more accurate and adaptive AI models capable of taking into account a large number of variables and identifying complex patterns in clinical and biochemical information is substantiated. The research examines the principles of predictive model construction, including gradient boosting, XGBoost, and logistic regression with L1 regularization, as well as their interpretability using SHAP. Special attention is paid to substantiating the prognostic value of such factors as the level of C-reactive protein, the number of preoperative consultations, the volume of blood transfused, chronic diseases, gender, age, parameters of drug therapy and laboratory abnormalities that are not included in traditional scales. The work aims to systematize existing techniques, proving the advantage of AI models in a real clinical process. For its implementation, comparative analysis and structuring of scientific sources were used. In conclusion, it is shown that AI screening transforms the practice of risk assessment by offering personalized and reproducible solutions. The article will be useful for specialists in the field of surgery, anesthesiology and medical informatics.

preoperative screening, machine learning, laboratory data, prognosis of complications, SHAP analysis, anesthetic risk, electronic medical records, personalized medicine, interpretability of models, medical analytics