Related Experiment Video
Updated: Aug 5, 2026
08:25
Polar Histogram Visualization of Acute Stress Disorder Scale Scores for Comprehensive Clinical Assessment
Published on: December 6, 2024
Comparison Study of Multiple Machine Learning Models for Predicting Anxiety Among Neurosurgical ICU Family Members
Feng Zhang1, Cunyi Huo1, Ping Xu1
1Department of Neurosurgery, The First Affiliated Hospital of Wannan Medical University (Yijishan Hospital of Wannan Medical University), Zheshan West Road on the 2nd, Wuhu 241000, Anhui, China.
Journal of Nursing Management
|July 30, 2026
Summary
Machine learning models effectively predict anxiety in neurosurgical ICU families, identifying key unmet needs. The Support Vector Machine (SVM) model shows promise for rapid screening and targeted nursing interventions.
Area of Science:
- Utilizes machine learning for psychological distress prediction in critical care.
- Focuses on neurosurgical intensive care unit (NICU) patient relatives' well-being.
Background:
- Relatives of neurosurgical ICU patients experience high rates of moderate to severe anxiety.
- Anxiety is linked to specific stressors and unmet needs within the neurosurgical ICU environment.
- Existing methods may not adequately screen for family anxiety or identify core unmet demands.
Purpose of the Study:
- To develop machine learning (ML) models predicting anxiety in neurosurgical ICU relatives.
- To identify critical unmet family needs strongly correlated with psychological distress.
- To select an optimal ML model for rapid clinical anxiety screening and targeted nursing interventions.
Main Methods:
- 1000 relatives of neurosurgical ICU patients (Glasgow Coma Scale ≤ 8) were enrolled.
- Data collected included general clinical information, HADS, and Critical Care Family Needs Inventory (CCFNI) scores.
- Machine learning models (SVM, Random Forest, ANN) were developed and validated using a 7:3 train-test split and 10-fold cross-validation.
Main Results:
- A high incidence of moderate to severe anxiety (76.8%) was observed among relatives.
- The Support Vector Machine (SVM) model demonstrated superior predictive efficacy (Accuracy: 0.8863, AUC: 0.9364).
- Core unmet needs identified by the SVM model included CCFNI7, CCFNI12, and CCFNI9.
Conclusions:
- Machine learning algorithms, particularly SVM, are effective for predicting family anxiety and screening unmet needs in neurosurgical ICUs.
- The SVM model can serve as a reliable clinical tool for precise nursing interventions and improving family-centered care.
- Findings support the development of personalized intervention protocols to enhance the care experience for patients and relatives.