Measurement of medical haemodilatation based on statistical analysis of big data

Abstract

This study aims to explore the association between hematoma expansion, edema development, and patient prognosis in hemorrhagic stroke patients. Utilizing clinical information, imaging characteristics, and prognostic data from 160 hemorrhagic stroke patients, several predictive models were constructed to examine the pathological progression and outcomes of these patients. Specifically, data preprocessing was employed to calculate the time intervals between each examination and select data within 48 hours for analyzing changes in hematoma volume and its percentage. This facilitated the determination of hematoma expansion within the first 48 hours post-onset, with results documented in a dedicated table. Employing the XGBoost model, both the test and training datasets were trained to develop a predictive model for hematoma expansion. Upon evaluation, the model demonstrated a 75% accuracy rate in predicting hematoma expansion across all patients (sub001-sub160). This study underscores the potential of using advanced predictive modeling, such as XGBoost, to enhance the prognosis assessment and clinical decision-making in hemorrhagic stroke care.