Machine Learning-Based Prediction of Unplanned Readmission Due to Major Adverse Cardiac Events Among Hospitalized Patients with Blood Cancers.

Summary: Patients with blood cancers, like leukemia or lymphoma, often receive treatments that can be tough on their hearts. Sometimes, this leads to unexpected trips back to the hospital for serious heart problems. Researchers used a large database of almost 77,000 patients to train smart computer programs (machine learning) to predict who might need to go back to the hospital within 90 days. The best computer model, called CatBoost, was able to spot the highest-risk patients. Older age and previous heart issues were the biggest warning signs. This tool could help doctors plan better care before patients go home, keeping them healthier and out of the hospital.