# Development and validation of a predictive model for postoperative functional recovery in patients with spontaneous intracerebral hemorrhage

## TL;DR

This study created a tool to predict recovery after brain hemorrhage surgery, using factors like age, blood pressure, and brain imaging features.

## Contribution

A new predictive nomogram was developed and validated for postoperative recovery in spontaneous ICH patients.

## Key findings

- The model included six predictors with strong discriminative performance (AUC of 0.90 in training and 0.83 in validation).
- Calibration and decision curve analysis confirmed the model's clinical utility and accuracy.
- The tool supports individualized rehabilitation planning and efficient resource allocation.

## Abstract

This study aimed to develop and validate a prognostic nomogram for predicting 3-month functional recovery in patients undergoing surgery for spontaneous intracerebral hemorrhage (ICH).

A retrospective cohort of 289 patients diagnosed with spontaneous intracerebral hemorrhage (ICH) underwent surgical management at the Intensive Care Unit of Taizhou Central Hospital between January 2021 and December 2024 was enrolled. Patients were randomly allocated into a training set (n = 203, 70%) and validation set (n = 86, 30%). A prognostic nomogram integrating imaging characteristics and clinical parameters was developed to predict 90-day functional recovery (modified Rankin Scale ≤2). Feature selection employed the Boruta algorithm, followed by multivariable logistic regression. Model discrimination was quantified by area under the ROC curve (AUC), while calibration curve was performed to evaluate model performance. Clinical utility was evaluated through decision curve analysis (DCA).

The multivariable model retained six significant predictors: midline shift (OR:2.09, 95%CI: 1.56–2.79), hematoma volume (OR:1.10, 95%CI: 1.05–1.15), age (OR:1.03, 95%CI: 1.01–1.05), mean arterial pressure (OR:0.93, 95%CI: 0.89–0.98), body mass index (OR:0.78, 95%CI: 0.66–0.92), and Glasgow Coma Scale (GCS) score (OR:0.92, 95%CI: 0.79–1.06). Discriminative performance was robust, with area under the receiver operating characteristic curve (AUC) of 0.90 (95% CI: 0.85–0.96) in the training set and 0.83 (95% CI: 0.73–0.93) in the validation set. Calibration plots demonstrated excellent agreement between predicted and observed probabilities. DCA confirmed the clinical value of the model and its good impact on actual decision-making.

This study developed and validated a pragmatic prognostic nomogram for spontaneous ICH patients undergoing surgical intervention, integrating six clinically actionable predictors: midline shift, hematoma volume, age, MAP, BMI, and GCS. The model demonstrated robust discriminative capacity, calibration and clinical applicability, which provides evidence-based support for the formulation of individualized rehabilitation programs and the optimization of medical resources.