There is a research gap in the field of risk feature extraction and prevention model construction for current novel algorithms in university training rooms. The research combines the XGBoost algorithm to construct the risk prevention and control model of the university training room, and quantitatively analyze the risk characteristics and weights of the training room. The experimental results show that the risk management prevention and control model based on the XGBoost tube has better performance and has significant practical application advantages. In the risk evaluation of the management of the biomedical training room of a university, it was found that the management weight of hazardous chemicals and controlled drugs was the highest, reaching 0.425, which had the greatest impact on the safety management of the biomedical training room. Relevant departments should strengthen the prevention and control. Overall, the research has important application value for the precise prevention and control of management risks in university training rooms.