Commentary

Video

Tailoring Precision Medicine Treatment of Hemorrhagic Stroke Through Predictive Models: Kara R. Melmed, MD

Fact checked by:

The clinical associate professor of neurology and neurosurgery at NYU Langone talked about the promise of machine learning in identifying patients at high risk for hematoma expansion to guide care. [WATCH TIME: 2 minutes]

WATCH TIME: 2 minutes

"If we can find the patients who really are expanding early, those are the patients where we can start to tailor our care and start to give more precise and focused management of blood pressure, anticoagulation reversal, etc.”

High blood pressure (BP) is recognized as a major modifiable risk factor for intracerebral hemorrhage (ICH) and has been associated with a greater likelihood of initial hemorrhage, poorer outcomes, and increased risk of recurrent stroke, cardiovascular events, as well as cognitive decline. Research suggests that early BP reduction, aiming for a systolic level below 140 mm Hg in the first few hours of presentation, may be appropriate for many patients after ICH. Over the long term, guidelines recommend to titrate antihypertensive therapy to achieve BP levels below 130/80 mm Hg in individuals with a history of ICH, regardless of demographic or clinical factors.1

Building on efforts to improve outcomes after ICH, recent research has also explored the use of machine learning (ML) to support clinical decision-making. One recently published study assessed ML models for predicting in-hospital mortality among patients with acute ICH, using both brain imaging and clinical data.2 At the conclusion of the analysis, results showed that the models were comparable or better predictive than traditional scoring systems, particularly when clinical and imaging inputs were reviewed by specialists. Overall, these findings suggest that ML may offer a useful tool for risk stratification in the acute setting, with potential to inform individualized treatment strategies alongside established interventions such as BP management.

Kara R. Melmed, MD, a clinical associate professor of neurology and neurosurgery at NYU Langone, recently argued against lowering the target BP too much for ICH in a debate session at the 2025 American Academy of Neurology (AAN) Annual Meeting, held April 5-9, in San Diego, California.3 After the session, Melmed had a conversation with NeurologyLive® to outline the growing role of precision medicine in managing ICH. In the interview, she emphasized the limitations of current clinical prediction methods and underscored the potential of ML to improve prognostication. Furthermore, Melmed noted that these tools could potentially enable earlier and more aggressive interventions and minimize risks associated with overtreatment in patients.

Click here for more coverage of AAN 2025.

REFERENCES
1. Mullen MT, Anderson CS. Review of Long-Term Blood Pressure Control After Intracerebral Hemorrhage: Challenges and Opportunities. Stroke. 2022;53(7):2142-2151. doi:10.1161/STROKEAHA.121.036885
2. Matsumoto K, Ishihara K, Matsuda K, et al. Machine Learning-Based Prediction for In-Hospital Mortality After Acute Intracerebral Hemorrhage Using Real-World Clinical and Image Data. J Am Heart Assoc. 2024;13(24):e036447. doi:10.1161/JAHA.124.036447
3. Melmed KR. No: Blood Pressure Targets in ICH: Should We Go Low?. Presented at: 2025 AAN Annual Meeting; April 5-9; San Diego, CA. Controversies in Neurology Plenary Session.
Related Videos
Roy H. Hamilton, MD, MS, FAAN
 Chitra Venkatasubramanian, MBBS, MD, MSc, FNCS
5 experts in this video
Christine Greiss, DO
5 experts in this video
Nicholas J. Silvestri, MD, FAAN
© 2025 MJH Life Sciences

All rights reserved.