Yongli Wu

The First People's Hospital of Foshan, China

Title: Respiratory Variability in Disorders of Consciousness: Relationship and Prognostic Value

---

Biography

Yongli Wu is affiliated with the Department of Rehabilitation Medicine, The First People’s Hospital of Foshan (Foshan Hospital Affiliated to Southern University of Science and Technology), School of Medicine, Southern University of Science and Technology, Foshan, China. His research interests include rehabilitation medicine, disorders of consciousness, respiratory variability, autonomic nervous system function, and machine learning-based clinical prediction. He has published studies on respiratory cycle measurement using inertial measurement units, respiratory variability in disorders of consciousness, and prognostic assessment in prolonged disorders of consciousness. His recent work focuses on non-invasive physiological biomarkers for consciousness assessment and clinical decision support.

Abstract

Objective: To investigate the relationship between RV and level of consciousness and to evaluate RV-based analysis for predicting clinical outcomes in DOC patients.
Methods: Patients with DOC and healthy controls were assessed using inertial measurement unit sensors to record triaxial acceleration signals. Respiratory waveforms were extracted using Butterworth filtering and moving average. RV indicators, including standard deviation, coefficient of variation, root mean square of successive differences, triangular index, and Poincaré plot metrics, were derived from 300-s respiratory parameters. A generalized additive model adjusted for age, sex, body temperature, heart rate, and pulse pressure difference. Group differences were evaluated using 1000-round bootstrap resampling and Mann–Whitney U tests with FDR correction. Machine learning models—random forest, elastic net, support vector machine, and partial least squares regression—were used to predict Coma Recovery Scale–Revised (CRS-R) scores and clinical outcomes.
Results: 44 DOC patients and 44 healthy controls were analyzed. Multiple RV indicators differed significantly between groups (p < 0.05). The DOC group showed higher inspiratory–expiratory ratio SD2, indicating increased complexity of respiratory rhythm, whereas respiratory rate area and triangular indices of exhalation and respiratory rate were lower in controls, reflecting reduced RV. The random forest model achieved the best performance for CRS-R prediction (MSE = 3.76, R² = 0.76) and clinical outcome prediction (AUC = 0.74, sensitivity = 0.86), outperforming other models.
Conclusions: RV, particularly when analyzed using random forest modeling, shows strong potential for prognostic assessment and clinical decision support in DOC. RV may represent a non-invasive biomarker reflecting neural control of respiration and consciousness state.