Improvements in the management and treatment of hemorrhagic stroke, one of the leading causes of disability and death worldwide, are of great public health importance. In this study, we investigated the process of hematoma expansion and the development of peripheral edema by analyzing clinical case data to improve accurate prediction of patient prognosis. Hematoma expansion within 48 hours of onset was identified by careful analysis of the first imaging data, and risk analysis and prediction were performed using an integrated tree model, such as the Random Forest algorithm. The model optimization significantly improved the prediction accuracy and provided strong support for clinical decision-making. In exploring the evolution of perihematoma edema, polynomial function fitting K-means clustering and particle swarm algorithms were used to reveal the individual differences in edema progression, which provides a new scientific basis for clinical treatment. This study provides an in-depth analysis of the pivotal clinical features of hemorrhagic stroke and offers new insights into hematoma expansion and the development of peripheral edema through advanced data processing and modeling, which in turn provides a scientific basis for clinical treatment decisions, enhances the understanding of this complex brain disease, and contributes to improved patient outcomes.

Wenteng Huo, Zunshu Li, Xueying Liu. Multivariate Data Analysis-Based Study on Clinical Decision Support for Hemorrhagic Stroke. International Journal of Frontiers in Medicine (2024), Vol. 6, Issue 6: 45-52. https://doi.org/10.25236/IJFM.2024.060607.

[1] SMITH E E, ROSAND J, GREENBERG S M. Imaging of hemorrhagic stroke[J]. Medical Physics, 2006, 14(2): 127–140.

[2] BANDI V, BHATTACHARYYA D, MIDHUNCHAKKRAVARTHY D. Prediction of brain stroke severity using machine learning[J]. Revue d’Intelligence Artificielle, 2020, 34(6): 753–761. 

[3] MUSCHELLI J, SWEENEY E M, ULLMAN N L, et al. PItcHPERFeCT: primary intracranial hemorrhage probability estimation using random forests on CT [J]. NeuroImage: Clinical, 2017, 14: 379–390. 

[4] LIN C-H, HSU K-C, JOHNSON K R, et al. Evaluation of machine learning methods to stroke outcome prediction using a nationwide disease registry[J]. Computer Methods and Programs in Biomedicine, 2020, 190: 105381. 

[5] HAKIMI R, GARG A. Imaging of hemorrhagic stroke[J]. CONTINUUM: Lifelong Learning in Neurology, 2016, 22(5): 1424–1450.

[6] HUANG L, LI T, JIANG H. Technical note: thermoacoustic imaging of hemorrhagic stroke: a feasibility study with a human skull [J]. Medical Physics, 2017, 44(4): 1494–1499. 

[7] ZAHEER A, OZSUNAR Y, SCHAEFER P W, et al. Magnetic resonance imaging of cerebral hemorrhagic stroke[J]. CONTINUUM: Lifelong Learning in Neurology, 2016, 11(5): 288–299.

[8] BOUTS M J, TIEBOSCH I A, RUDRAPATNA U S, et al. Prediction of hemorrhagic transformation after experimental ischemic stroke using mri-based algorithms[J]. Journal of Cerebral Blood Flow & Metabolism, 2017, 37(8): 3065–3076. 

[9] CHOI J-M, SEO S-Y, KIM P-J, et al. Prediction of hemorrhagic transformation after ischemic stroke using machine learning[J]. Journal of Personalized Medicine, 2021, 11(9): 863.

[10] Rong Fang, Gao Z ,Liu. An Improved AdaBoost Algorithm for Hyperparameter Optimization[J]. Journal of Physics: Conference Series, 2020.DOI:10.1088/1742-6596/1631/1/012048.

[11] TAO Y, WANG Y. Improved k means algorithm based on the selection of initial clustering centers [J]. Foreign Electronic Measurement Technology, 2022, 41(9): 54–59.

[12] ZHENG Z, GONG Q, ZHANG J. Improved association rule mining algorithm: mifp-apriori algorithm [J]. Science Technology and Engineering, 2019, 19(16): 216–220.

[13] LIU W, XU Y. Association rule mining of metro failures based on improved apriori algorithm[J]. Journal of Ordnance Equipment Engineering, 2021, 42(12): 210–215.