A Case Study on Risk Prediction in Heart Failure Patients using Random Survival Forest

In order to provide prompt care to heart failure patients, a quick diagnosis and risk assessment is essential. It is possible for clinicians to make more informed decisions about the treatment intensity by anticipating future risks of the patient. Identifying the patients who are at low risk might also help to reduce patient stress and treatment costs. In this regard, the Random Survival Forest model is utilized to create a risk prediction system for heart failure patients using a real-world dataset. The most important indicators for predicting the survival of heart failure patients were determined to help clinicians make decisions. The model is compared with other popular survival analysis tools like the Cox Proportional Hazard model. The experimental results show that our developed risk model outperforms the other methods as well as previous studies with a Harrell’s Concordance Index of 0.81. Thus, our proposed approach can prove very useful for physicians in making well-informed decisions about heart failure mortality prediction based on an accurate interpretation of patient records.