Nearly 40% of AML patients are readmitted or die within a month of leaving the hospital. A new predictive model found that a history of solid tumors and cardiopulmonary symptoms documented in clinical notes are key risk factors. Researchers say capturing this data could help clinicians identify high-risk patients before discharge and enable more personalized follow-up care.
Patients with acute myeloid leukemia (AML) face a steep uphill battle even after leaving the hospital. A study of 305 AML patients found that nearly 40% were either readmitted or died within 30 days of discharge — a sobering reminder of how vulnerable this population remains post-treatment. Fever and infection were the most common culprits driving readmissions, largely because chemotherapy leaves patients severely immunosuppressed.
To get ahead of this, researchers at the University of Iowa built a three-tiered predictive model that layered in comorbidities and symptoms extracted from clinical notes using natural language processing. The enhanced model meaningfully outperformed the baseline, pinpointing two standout risk factors: a prior history of solid tumors and frequently documented cardiopulmonary symptoms. Notably, older patients on high-intensity chemotherapy faced compounding risk with each passing year.
Key Takeaways:
Why it matters: Clinicians document symptoms daily, but that data rarely feeds into risk prediction tools. This research makes the case for smarter, EHR-integrated decision support that can flag high-risk AML patients before discharge — enabling earlier follow-up and potentially saving lives.
