In recent years, people's dietary structure and living habits have changed a lot. Unhealthy eating habits and living habits have led to more and more people suffering from hypertension. It will cause damage to target organs, and then leading to hypertension complications. There are some serious complications in hypertension complications, which poses a huge threat to the life safety of patients with hypertension. And at present, complications of hypertension have become a research hotspot globally. This paper proposes the XGBLR model which mainly combines the XGBoost algorithm and the Logistic regression method, it uses the XGBoost algorithm to construct a tree structure, then uses the tree structure to reconstruct features, and finally uses the Logistic regression method to classify. The model can predict whether patients will experience severe hypertension complications within three years. The final experimental results prove that the model can be used to predict whether severe hypertension will occur in patients with hypertension within three years. Based on the results, the patient can conduct a more comprehensive examination to avoid blind treatment and avoid the huge threat on life safety. Doctors can also use the results as a reference for threatening diseases. This article applies machine learning methods to the medical field, which is of practical significance.