Using ML techniques to predict extubation outcomes for patients with central nervous system injuries in the Yun-Gui Plateau

Using ML techniques to predict extubation outcomes for patients with central nervous system injuries in the Yun-Gui Plateau | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Using ML techniques to predict extubation outcomes for patients with central nervous system injuries in the Yun-Gui Plateau Zihan Chen, Haotian Wu, Yao Zhou, Qian Liu, Hongmei Zhang, Xiaochen Li, and 2 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-5549340/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 22 May, 2025 Read the published version in Scientific Reports → Version 1 posted 6 You are reading this latest preprint version Abstract No predictive models have been reported for tracheostomy extubation success in plateau region rehabilitation departments. Hence, the primary objective of this retrospective study was to evaluate the predictive capabilities of different models for extubation outcomes in CNS injury patients in plateau rehabilitation departments, as well as investigate the influence of clinical features on these outcomes. Data were collected from 501 adult tracheostomy patients in the Department of Rehabilitation Medicine, including 196 successful extubations. Logistic regression was employed to identify the significant features linked to extubation outcomes from a pool of 31 variables. A total of eight independent models and a weighted posterior voting ensemble model were developed. Hyperparameter optimization and 10-fold cross-validation were used to assist in choosing model parameters. Random forest (ACC = 84.15, AUC = 0.85), extra trees (83.17%, 0.87), K-NN (82.18%, 0.85), and gradient boosting (81.19%, 0.85) performed well. An ensemble model (85.15%, 0.87) combining random forest, Gaussian naive Bayes, and K-NN via the WPV method was developed. Dysphagia and low GCS scores have been linked to increased difficulty in extubation, as indicated by SHAP values and previous studies. Moreover, there could be a relationship between chronic inflammation and albumin levels in patients, which may collectively impact extubation success. This study evaluated the effectiveness of conventional models for predicting extubation outcomes and analyzed the factors influencing extubation results at high altitudes, laying the groundwork for clinical use and future research. Nevertheless, further research will see advantages in using multicentric approaches and broadening clinical indicators. Health sciences/Diseases/Neurological disorders/Brain injuries Health sciences/Diseases/Neurological disorders/Spinal cord diseases Health sciences/Diseases/Neurological disorders/Stroke Biological sciences/Computational biology and bioinformatics/Machine learning Airway Extubation Central Nervous System Diseases ML Predict Figures Figure 1 Figure 2 Figure 3 Figure 4 Introduction Although tracheostomy is crucial for severe central nervous system (CNS) injuries, it can also lead to a range of complications, including bleeding, unstable oxygen saturation, infections, and granulation tissue formation 1 . Therefore, extubation assessment is an important step in reducing complications and promoting functional recovery in patients. Factors such as age, swallowing ability, cough reflex, and infection status are associated with successful extubation in patients with CNS injuries 2 . Additionally, some studies have shown that swallowing function is a reliable indicator of extubation outcomes in stroke patients 3 . Most current studies on extubation predictors rely on traditional statistical analyses 4 – 6 , which have provided some clinical insights for extubation decisions. However, the extubation failure rate remains above 10% 7 . In order to improve the accuracy of extubation predictions, further research on new methods is needed. In the Yunnan-Guizhou Plateau, Kunming is situated at a relatively high altitude, approximately around 1900 meters 8 . The unique characteristics of high-altitude environments include low oxygen and low air pressure conditions, which directly impact the respiratory and cardiovascular systems of patients. Wang et al. concluded that long-term hypoxia in high-altitude areas can cause pulmonary vasoconstriction, remodeling, increase the risk of pulmonary hypertension, worsen right ventricular burden, and impact the function of the circulatory and respiratory systems 9 . One common reason for extubation failure in the ICU is cardiovascular system dysfunction 10 . Moreover, high-altitude CNS injury patients are more likely to develop severe brain edema, resulting in higher disability and mortality rates compared to individuals in plain areas 11 . Overall, these patients with poor prognosis may have a lower success rate in extubation and an increased risk of reintubation. However, it is currently unclear which factors will affect the extubation outcomes of CNS injury patients in high-altitude regions. Therefore, one of the aims of this study is to explore factors influencing extubation outcomes in such patients. Both the tracheostomy decannulation process in other departments and in the rehabilitation department emphasize formalized protocols and interprofessional collaboration 12 , 13 . In acute care environments, decannulation decisions are typically driven by factors such as the patient’s ability to tolerate tracheostomy tube capping, the effectiveness of cough, and the control of secretions 14 . The rehabilitation departments, by contrast, incorporate comprehensive assessments that also evaluate functional status, swallowing ability, and the patient’s readiness 15 . Interdisciplinary tracheostomy teams, comprising rehabilitation specialists, respiratory therapists, speech-language pathologists, and nurses, improve decision-making processes and customize decannulation protocols to meet each patient's specific needs 16 . Despite the comprehensive considerations of the rehabilitation department and its emphasis on personalization, there is currently a lack of research using machine learning (ML) methods to predict extubation outcomes. Traditional statistical methods for predicting extubation typically use a small set of predetermined variables and a straightforward model structure based on linear relationships. This may overlook complex nonlinear relationships in clinical data. Modern ML techniques are capable of handling high-dimensional data without strict assumptions. They can capture complex interactions hidden in the data, leading to a significant improvement in prediction accuracy and sensitivity 17 . ML methods can fully utilize data from mechanical ventilators, multi-parameter monitoring devices, and physiological signals, providing a more personalized and dynamic predictive basis. At the same time, another major advantage of ML models is the development of their interpretability tools. SHAP (Shapley Additive exPlanations) can unravel complex models, quantify the impact of each input feature, and help clinical doctors understand the model's decision logic 18 . This approach not only helps identify key factors influencing extubation outcomes, but also provides a theoretical basis for developing clinical interventions based on these factors. However, there is limited research on predictive models for extubation outcomes in high-altitude regions. Therefore, the aim of this retrospective study was to determine the optimal ML models for this patient population and to identify factors associated with successful extubation in tracheostomized patients in the Plateau. There were 8 conventional models developed, as well as a Weighted Posterior Vote (WPV) ensemble model. The predictive performance of the models was subsequently compared using evaluation metrics including accuracy (ACC), area under the receiver operating characteristic curve (AUC), sensitivity, specificity, Youden's index, positive predictive value (PPV), and negative predictive value (NPV). At last, the SHAP analysis was used to evaluate the influence of features on the models and extubation success rate. Materials and methods Study design The study design followed the STROBE guidelines for a retrospective, nonsynchronous study 19 . The data were derived from 501 adult tracheostomy patients (aged over 18) at the Department of Rehabilitation, Second Affiliated Hospital of Kunming Medical University, between August 2016 and November 2022, with 196 successful extubations. Upon admission, patients underwent a functional assessment by skilled Clinical Assessor. Inclusion criteria: (1) Tracheostomy patients; (2) Patients undergoing extubation surgery in our department; (3) with CNS injuries; (4) aged 18 to 85 years old. Exclusion criteria: (1) Patients with unstable conditions or needing to be transferred to other departments for treatment; (2) with incomplete clinical data. Informed consent statement Informed consent to participate was waived by Ethics Committee of the Second Affiliated Hospital of Kunming Medical University (IRB). In cases where participants cannot be located and the research project does not involve personal privacy or commercial interests, informed consent may be waived with approval from the ethics committee, as outlined in the Ethical Review Procedures for Biomedical Research Involving Human Participants by the Chinese National Health Commission. Ethics approval The study design was approved by the Ethics Committee of the Second Affiliated Hospital of Kunming Medical University (Ethics approval number: Shen-PJ-ke-2023-28). The study has been performed in accordance with the Declaration of Helsinki. Data collection The study collected demographic, behavioral, and clinical data from tracheostomy patients: (1) demographic data, such as sex and age; (2) behavioral factors, such as smoking history, primary disease, paralysis, consciousness and cognitive status, and medical history (epilepsy, hypertension, diabetes, heart disease); (3) clinical indicators, including wet rales, dysphagia, pulmonary infection, sputum culture, sputum viscosity, upper respiratory tract structural abnormalities (URSA), and Glasgow Coma Scale (GCS); and (4) blood physiology, including procalcitonin, interleukin 6 (IL-6), high-sensitivity C-reactive protein (hs-CRP), white blood cells, neutrophils, lymphocytes, monocytes, hemoglobin, and albumin. Encode all categorical variables using label encoding. The specific numerical symbols are presented in the Supplementary Material (Table S1). All the data were managed and stored centrally by the hospital's medical records department. Some clinical indicators' diagnostic criteria are detailed in the supplementary materials. The data of all patients included in this study was complete, hence there were no missing values in the complete data. Feature screening To deal with the potential data imbalance, we executed feature selection and standardization. Logistic regression was used to identify the factors related to extubation outcomes. The analysis was conducted using the Statsmodels package in Python 3.12.4 , with a significance level of 0.05. To avoid overlooking hidden information, weakly correlated variables (P ~ 0.05) were also included in the model construction 20 . Next, the selected feature values were standardized. The mean and standard deviation of the training set were calculated, and the training set data was transformed using zero-mean normalization method to have a distribution with a mean of 0 and standard deviation of 1. To prevent information leakage from the test set to the training phase, the test set was normalized using the mean and standard deviation calculated from the training set. Model selection and construction A total of eight models were constructed, including the k-nearest neighbors (K-NN), support vector machine (SVM) 21 , Gaussian naive Bayes (GNB), decision tree (DT), random forest (RF) 22 , extra tree classifier (ET), gradient boosting (GB), and logistic regression (LR). Modeling parameters were selected via cross-validation and hyperparameter optimization. Consistent data splitting was chosen to ensure comparability among multiple models. The 80%-20% split balanced model capacity with evaluation reliability, restrained overfitting, and optimized computational resources 20,23 . Data preprocessing and dataset splitting were carried out using the Numpy and Sklearn packages in Python 3.12.4 24 . The details of hyperparameter optimization are provided in the Supplementary Material (Table S2). The workflow of the model construction is shown in Figure 1. Feature analysis SHAP value was used to assess the significance of each feature. This function, implemented via the SHAP package in Python 3.12.4 25 , employs a game-theoretic approach to evaluate the importance of each feature. SHAP values not only rank features by their importance to the model but also reveal the impact of each feature on patient extubation outcomes 23 . Only models with an AUC greater than 0.85 and an ACC greater than 80% are discussed in the discussion section. Ensemble model construction An ensemble model was constructed using Sklearn packages in Python 3.12.4 . Emerging methods such as WPV offer a unique way to incorporate model confidence levels into the final decision, addressing issues related to class imbalance and overfitting in noisy environments when compared to ensemble methods 26 . We employed the hard classification voting ensemble in our model in detail. The hard voting scheme is suited for predicting distinct class labels, while soft voting is appropriate for predicting continuous values 27 . In theory, each base classifier produces a binary output (1 or 0). WPV conducts voting based on the frequency of each prediction. In the scenario where RF predicts 1, GNB predicts 1, and K-NN predicts 0, the final prediction will be 1 because it is the most common outcome. In model ensemble, the diversity and complementarity of models were taken into account 28 . Results Statistical analysis Logistic regression was performed with extubation outcomes as the independent variable and other factors as the dependent variables. We found that the outcomes of extubation were associated with several factors: disease duration (P = 0.002), underlying condition (P < 0.001), quadriplegia (P < 0.001), GCS score (P = 0.040), swallowing function (P < 0.001), sputum culture pathogens (P = 0.013), interleukin-6 levels (P = 0.023), lymphocyte count (P = 0.046), and airway structural abnormalities (P < 0.001). To avoid omitting relevant information, features with P-values close to 0.05 were also included in the model. These included mental disorders (P = 0.05), pulmonary embolism (P = 0.087), and albumin levels (P = 0.081). The statistical analysis results are presented in Table 1. Table 1 . Logistic regression results for each variable. Categorical variables are presented as n (%), and continuous variables are presented as mean [SD]. The 95% confidence interval is presented as (Lower bound, Upper bound). Variables Successful Extubation (n=196) Failed Extubation (n=305) OR Standard Error CI P -Value Meal gender yes 148 (75.51) 232 (76.06) 0.99 0.332 (-0.672, 0.628) 0.947 no 48 (24.49) 73 (23.93) Wet Rales yes 141 (71.94) 215 (70.49) 0.87 0.302 (-0.728, 0.456) 0.653 no 55 (29.06) 90 (29.51) Smoking yes 123 (62.76) 222 (72.79) 1.25 0.306 (-0.38, 0.819) 0.472 no 73 (37.25) 83 (27.21) Consciousness Disorder yes 122 (62.24) 213 (69.84) 1.28 0.730 (-1.184, 1.676) 0.736 no 74 (37.76) 92 (30.16) Dysphagia yes 103 (52.55) 226 (74.09) 0.13 0.320 (-2.696, -1.441) 0.000 no 93 (47.45) 32 (10.49) Epilepsy yes 167 (85.20) 226 (74.10) 0.66 0.339 (-1.076, 0.252) 0.224 no 29 (14.80) 79 (25.90) Viscous sputum yes 147 (75.0) 265 (86.89) 0.62 0.328 (-1.121, 0.163) 0.144 no 49 (25.00) 40 (13.115) Upper Respiratory Tract Structural Abnormalities (URSA) yes 191 (97.45) 264 (86.56) 0.10 0.642 (-3.584, -1.069) 0.000 no 5 (2.551) 41 (13.44) Hypertension yes 106 (54.08) 160 (52.46) 0.73 0.301 (-0.898, 0.282) 0.306 no 90 (45.918) 145 (47.54) Coronary Heart Disease yes 190 (96.94) 292 (95.74) 1.69 0.696 (-0.839, 1.888) 0.451 no 6 (3.06) 13 (4.26) Diabetes Mellitus yes 166 (84.69) 247 (80.98) 1.07 0.367 (-0.651, 0.787) 0.853 no 30 (15.31) 58 (19.02) Pulmonary Infection yes 167 (85.20) 248 (81.31) 1.78 0.372 (-0.15, 1.306) 0.12 no 29 (14.796) 57 (18.69) Pulmonary Embolism yes 186 (94.90) 274 (89.84) 0.40 0.536 (-1.969, 0.133) 0.087 no 10 (5.10) 10 (3.28) Pulmonary Contusion yes 182 (92.85) 295 (96.72) 1.09 0.594 (-1.079, 1.252) 0.884 no 14 (7.14) 31 (10.16) Mental Disorders yes 175 (89.29) 294 (96.39) 2.99 0.560 (-0.002, 2.193) 0.05 no 21 (10.71) 11 (3.61) Disease Duration DD ≤ 30 days 3 (1.53) 42 (13.77) 1.57 0.150 (0.16, 0.747) 0.002 30 days < DD ≤ 60 days 71 (36.22) 110 (36.07) 60 days 90 days 66 (33.67) 93 (30.49) Primary Disease Cerebral Trauma 69 (35.20) 81 (26.56) 0.78 0.071 (-0.383, -0.106) 0.001 Cerebral Hemorrhage 78 (39.80) 124 (40.66) Cerebral Infarction 21 (10.71) 39 (12.79) Subarachnoid Hemorrhage 4 (2.04) 3 (0.98) Ischemic Hypoxic Brain Disease 5 (2.55) 17 (5.57) Central Nervous System Infectious Diseases 3 (1.53) 7 (2.30) Spinal Cord Injury 7 (3.57) 14 (4.59) Other Diseases 9 (4.59) 20 (6.56) Limb Paralysis Quadriplegia 133 (67.86) 265 (86.89) 0.20 0.351 (-2.319, -0.943) 0.000 Hemiplegia 63 (32.14) 40 (13.11) Paraplegia 0 (0) 0 (0) Sputum Culture Pathogens Pseudomonas aeruginosa 35 (17.85) 76 (24.92) 1.12 0.047 (0.025, 0.21) 0.013 Klebsiella pneumoniae subtype 10 (5.10) 23 (7.54) Acinetobacter baumannii 12 (6.12) 18 (5.90) Escherichia coli 5 (2.55) 9 (2.95) Serratia marcescens 1 (0.51) 9 (2.95) Other Pathogens 12 (6.12) 15 (4.92) Two or more pathogens 64 (32.65) 79 (25.90) None 57 (29.08) 76 (24.92) Age (year) 51.2[17.17] 54.18[16.40] 0.99 0.009 (-0.027, 0.009) 0.342 GCS 12.69[3.17] 9.82[3.81] 1.24 0.106 (0.01, 0.427) 0.040 Procalcitonin (ng/ml) 0.12[0.18] 0.32[1.24] 0.46 0.576 (-1.914, 0.344) 0.173 Interleukin 6 (pg/ml) 16.88[19.67] 41.21[63.13] 0.98 0.007 (-0.028, -0.002) 0.023 High-Sensitivity C-Reactive Protein (mg/L) 18.98[23.04] 33.52[37.04] 1.00 0.006 (-0.012, 0.011) 0.963 White Blood Cell Count (10 9 /L) 8.58[3.03] 10.01[5.54] 1.02 0.073 (-0.119, 0.166) 0.746 Neutrophils (10 9 /L) 6.1[2.78] 7.44[3.43] 0.90 0.090 (-0.28, 0.073) 0.249 Lymphocytes (10 9 /L) 1.68[0.69] 1.53[0.66] 1.56 0.223 (0.008, 0.884) 0.046 Monocytes (10 9 /L) 0.56[0.25] 0.62[0.27] 0.68 0.632 (-1.623, 0.853) 0.542 Hemoglobin (g/L) 111.79[14.48] 107.06[15.75] 1.00 0.010 (-0.019, 0.021) 0.912 Albumin (g/L) 35.89[3.36] 34.04[4.68] 1.07 0.038 (-0.008, 0.141) 0.081 Model performance Following feature selection, all models were developed. Subsequently, hyperparameters of each model were tuned by maximizing classification accuracy. To mitigate overfitting, model parameters were further optimized through 10-fold cross-validation. The ACC and AUC of the model before and after optimization are shown in Figure 1C. The threshold for maximizing the Youden index via ROC curve analysis is: RF (0.5), ET (0.51), K-NN (0.58), GB (0.53) 29 . The performance are presented in Figure 2 and Supplementary Materials (Figs. S1-S3, Table S3). The WPV method was employed for model integration to achieve optimal performance across all evaluation metrics. The RF, GNB, and k-NN models were integrated through this WPV approach. The ensemble model performed best when the weights of the three models were set to a 1:1:1 ratio. The model performance of the WPV is shown in Figure 3. Features contribution analysis The contribution of each feature to the models and its impact on the extubation success rate were determined through the calculation of SHAP values. Swallowing dysfunction was consistently the most influential factor in all models, as patients with swallowing difficulties were often classified as negative (i.e., extubation failure). Lower GCS scores also played a significant role, leading the models to lean towards negative classifications. The SHAP values for each model are showed in Figure 4 and the Supplementary Materials (Figure S4). Discussion In this study, Logistic regression was initially conducted on all variables related to extubation outcomes. Subsequently, eight ML models were developed based on these variables, and their predictive performance for extubation success was evaluated using multiple metrics. The SHAP analysis revealed that among all variables, disease duration, primary disease, tetraplegia, GCS score, swallowing function, pathogens in sputum culture, interleukin-6 levels, lymphocyte count, and abnormal respiratory structures were associated with successful extubation. Under optimal parameter selection, random forest, extra trees, k-NN, gradient boosting, and WPV all performed well. In recent years, the use of ML models in medical prediction has been increasing, but there are significant differences in performance among different models in predicting extubation. For instance, LR is widely used due to its strong interpretability, but it has limitations in modeling complex nonlinear relationships. Wang et al . developed a predictive model using multivariable logistic regression, but its AUC was only 0.793, indicating limitations in handling high-dimensional data 30 . A study compared the performance of RF (AUC = 0.787), linear regression (AUC = 0.762), artificial neural network (AUC = 0.763), and SVM (AUC = 0.740). Nevertheless, none of the models achieved satisfactory results, as all AUC values were less than 0.8 31 . In contrast, Huang et al . achieved an AUC of 0.976 with their RF model built using time series respiratory parameters, which was significantly better than SVM and GNB 32 . Even though it was lower than the previous study, the four models in our study that showed good performance all had AUC values above 0.85. Given the unique physiological parameters in high-altitude areas, we still believe that their performance is clinically significant. Next, they utilized time-series ventilator-derived parameters as features, which may be more advantageous for improving the predictive power of the model. Mixed types of feature values can impact the model's performance. In this study, the modeling features include both continuous and categorical data values. In comparison, mixed type of data is more in line with clinical reality than single type of data. Different ML algorithms have their own advantages and limitations when dealing with mixed features. DT and tree-based models like random forests have the ability to handle categorical variables without the need for extensive preprocessing, demonstrating reliable performance with high-dimensional, heterogeneous datasets 33 . Continuous variables and categorical variables differ in scale, distribution, and information representation. Inputting them directly into data scale-sensitive models such as SVM and K-NN may result in the model being excessively sensitive to one type of feature, while ignoring crucial information from the other type of feature. Therefore, it is often necessary to normalize or standardize continuous variables. Techniques like one-hot encoding, label encoding, and neural embeddings convert categorical variables into a unified numerical representation, effectively addressing scale inconsistency issues during model training 34,35 . Reducing the risk of overfitting and enhancing the generalization of models is crucial. Due to the different ranges and distributions of various features, it is easy to cause data bias and model overfitting issues, especially when the sample size is small 36 . Therefore, the hyperparameter optimization and cross-validation methods is particularly crucial to ensure that the model not only performs excellently but also possesses strong generalization capabilities 37,38 . Additionally, the feature selection process is especially critical for mixed data. This process focuses on retaining the most informative features while eliminating redundant and noisy ones, providing a key pathway to enhance the model's learning efficiency 39 . Hence, The aforementioned methods were adopted to further strengthen the model's performance and learning depth. Furthermore, mitigating model overfitting has also somewhat enhanced the generalization ability of each model 40 . In present study, a WPV was integrated model developed. Bagging, boosting, and stacking are conventional methods that have been widely researched. Bagging functions by training base models on bootstrap samples, ultimately decreasing variance and improving model stability 41,42 . Meanwhile, bagging’s simplicity and ease of implementation make it a robust baseline method, although its uniform voting mechanism may not fully exploit the differing confidence levels of individual models 42 . Although boosting can achieve high accuracy on training data, its performance may degrade on unseen data when noise is present. Consequently, the benefits of boosting are often offset by decreased robustness in noisy environments 43,44 . Finally, stacking entails utilizing a meta-learner to combine the outcomes of different models, thereby capturing more intricate relationships in the data 45 . While stacking often provides better performance in diverse settings, its reliance on an additional meta-learning layer can result in increased complexity in model selection and parameter tuning 45 . In contrast to previously mentioned methods, emerging methods such as WPV offer a new way to incorporate model confidence levels into the final decision, addressing issues related to class imbalance and overfitting in noisy environments 26,46 . This mechanism is a complement to the bagging and boosting frameworks and could improve ensemble performance in multiple medical domains. In present study, the WPV ensemble model showed superior performance in all evaluation dimensions compared to individual models. Given that our dataset was derived from a single center, our model may consequently have reduced generalizability. The WPV method could be utilized to partially alleviate this issue. Firstly, by combining the outputs of complementary models, ensemble models can better capture the underlying distribution of the data, alleviating the risks associated with individual models learning inadequately or underfitting 26 . Next, the voting mechanism has excellent scalability, making it easy to add or remove models integrated into the system. This is extremely beneficial for managing extensive data flows and intricate tasks, enabling the flexible adjustment of the model's scale and structure according to specific requirements 46 . Finally, an analysis using SHAP was conducted to evaluate the impact of each feature on the extubation outcomes. The SHAP values suggested that patients with swallowing dysfunction had a significantly lower extubation success rate. Previous retrospective studies also reported a significant correlation between swallowing function and the extubation success rate 5,47,48 . Damage to the glossopharyngeal (IX), vagus (X), and hypoglossal (XII) nerves could impact the muscles used in swallowing, leading to dysphagia in individuals with CNS injury 49 . Precise coordination of breathing and swallowing is crucial for airway protection 50 . Dysfunction of the swallowing musculature increases the risk of aspiration and pulmonary infection, which are common causes of extubation failure and prolonged intubation in critically ill patients 51,52 . Many studies have emphasized the need for clinicians to prioritize swallowing training after extubation 53,54 . However, the results of our study suggest that swallowing function training before extubation should also be emphasized, as it might help reduce postextubation dysfunction and decrease the likelihood of extubation failure. Training and assessing swallowing are crucial abilities in rehabilitation medicine; early intervention in rehabilitation could enhance the extubation success rate by enhancing swallowing function prior to extubation. Nevertheless, this process must highlight individualization, and specific strategies also require additional investigation. Swallowing difficulties are present in 40% of patients with quadriplegia 49 . They might also have problems like diaphragm imbalance and respiratory issues 55,56 . Given the low oxygen and low pressure environment of the plateau, patients with quadriplegia may need to focus more on their respiratory training. In severe cases, diaphragm pacemakers may be considered as interventions 57 . The influence of consciousness level on extubation success rate remains controversial 58 . Nevertheless, GCS were consistently ranked in the top three contributors in the models' classification. Patients with higher GCS scores showed significantly greater extubation success rates. This finding reinforces previous studies which demonstrate that consciousness level strongly correlates with extubation outcomes 59,60 . Furthermore, cognitive impairment contributes to swallowing disorders, aspiration risks, and subsequent pneumonia, potentially resulting in delayed extubation for patients with tracheostomies 61 . Impaired consciousness leads to prolonged bed rest, thereby increasing the risk of extubation failure 62 . If consciousness disorders are truly related to extubation outcomes, then high-altitude areas need to take this issue seriously. The low pressure and insufficient oxygen supply in high-altitude regions can lead to cerebral tissue hypoxia, resulting in vasodilation, disruption of the blood-brain barrier, and cerebral edema 11,63 . In extreme situations, these alterations may result in a patient quickly transitioning from a state of blurred consciousness to a coma. Therefore, the consciousness of patients with CNS in high-altitude areas should be further emphasized. The results showed that patients with higher ALB levels were more likely to be classified as extubation failures in this both models. Following CNS damage, various forms of cell death, including apoptosis, ferroptosis, and mitochondrial dysfunction, can deplete a substantial amount of protein resources 64,65 . Previous research revealed that patients with a daily protein intake greater than 1.2 g/kg/day effectively presented increased serum albumin levels, and their median duration of mechanical ventilation was significantly shorter than that of those with lower protein intake 66 . Their conclusions emphasized that higher protein intake is beneficial for the successful extubation of patients with prolonged endotracheal intubation 66,67 . These findings suggest that increasing serum albumin levels through enhanced nutritional support or albumin supplementation prior to extubation may be an effective strategy to improve extubation success rates. In our study, lymphocyte and IL-6 levels influenced extubation outcomes. The model is more likely to classify samples with lymphocytopenia and elevated IL-6 levels as having a higher risk of extubation failure. A pathway analysis study highlighted IL-6 as a key biomarker linked to inflammatory processes, with elevated levels serving as a reliable indicator of active inflammation in patients 68 . Moreover, another study has suggested that lymphopenia could indicate a state of chronic inflammation and prolonged immune depression in patients 69 . Although the exact mechanism by which lymphocytopenia impacts extubation outcomes remains unclear, patients with lymphocytopenia and immunosuppression exhibit significantly reduced extubation success rates and elevated in-hospital mortality 70,71 . Patients with lymphopenia were also found to present with thrombocytopenia and hypoalbuminemia, suggesting a state of systemic compromise 72 . This finding suggested that preextubation anti-inflammatory treatment combined with protein intake might be a potential strategy to improve extubation success rates. This study also has several limitations. Firstly, our models demonstrated superior performance compared to previous studies, but they are not yet suitable for clinical use. More clinical research is necessary in this area. Second, the features included in this study may have been limited, and the model's reliability could be enhanced by incorporating a more comprehensive set of features during the modeling process 73 . At the same time, the data collected in this study lack a time dimension, so the results lack explanatory dynamics of the disease course in tracheostomy patients. Moreover, this study did not find any distinctive features specifically associated with high-altitude pulmonary edema outcomes. In future studies, additional indicators such as alveolar oxygen partial pressure (or blood oxygen saturation), plateau pressure, blood urea nitrogen, heart rate, positive end-expiratory pressure, and creatinine will be included 74-76 . Finally, data collection from a single center may reduce the generalization ability of the model. Therefore, in future research, multicenter data collection should be adopted, as it can make the research results more convincing and increase generalizability. Conclusion In summary, the models developed in this study, including the random forest, extra tree classifier, k-nearest neighbor, gradient boosting, and WPV ensemble models, demonstrated strong performance in terms of the AUC and accuracy, indicating good predictive ability for outcome prediction at weaning. According to SHAP analysis, dysphagia, quadriplegia, altered consciousness, low white blood cell levels, and abnormal levels of interleukins and lymphocytes may have a negative impact on extubation outcomes. However, the data was collected from a single center, which may lead to insufficient generalizability. In the future, it is necessary to consider supplementing with data from multiple centers. Furthermore, future studies should incorporate a wider range of clinical indicators, particularly those associated with the respiratory system. Declarations Data availability The datasets used during the current study are available from the corresponding author upon reasonable request. Author contributions Z.H.C., X.Y. and L.Q.Y. conceived of and designed the study. Z.H.C., X.Y., X.C.L., and H.T.W. performed analysis. Z.H.C., X.Y., H.T.W. and L.Q.Y. wrote the manuscript and Z.H.C., X.Y. and L.Q.Y. critically reviewed the manuscript. Z.H.C., X.Y., and Y.Z. generated the figures and tables. Q.L. and H.M.Z. were responsible for the management of public data. All authors have read and approved the manuscript. Competing interests The authors declare no competing interests. Additional information Correspondence and requests for materials should be addressed to L.-Q.Y. Reprints and permissions information is available at www.nature.com/reprints. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. Te images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. References Kumar, S., Patra, A., Deepthi, S. & Biradar, K. Tracheostomy Complications Over 5 Years: Decannulation Analysis. Iran. J. Otorhinolaryngol. 36 , 559–565. 10.22038/ijorl.2024.75652.3533 (2024). Gallice, T., Cugy, E., Branchard, O., Dehail, P. & Moucheboeuf, G. Predictive Factors for Successful Decannulation in Patients with Tracheostomies and Brain Injuries: A Systematic Review. Dysphagia 39 , 552–572. 10.1007/s00455-023-10646-2 (2024). Suntrup-Krueger, S. et al. Extubation Readiness in Critically Ill Stroke Patients. Stroke 50, 1981–1988, (2019). 10.1161/strokeaha.118.024643 Guzatti, N. G. et al. Predictive Factors of Extubation Failure in COVID-19 Mechanically Ventilated Patients. J. Intensive Care Med. 37 , 1250–1255. 10.1177/08850666221093946 (2022). Ghiani, A. et al. Incidence, causes, and predictors of unsuccessful decannulation following prolonged weaning. Therapeutic Adv. chronic disease . 13 , 20406223221109655. 10.1177/20406223221109655 (2022). Chen, Y. et al. Predictors of long-term decannulation in patients with disorders of consciousness. Front. Neurol. 14 10.3389/fneur.2023.1099307 (2023). Thille, A. W., Richard, J. C. & Brochard, L. The decision to extubate in the intensive care unit. Am. J. Respir. Crit Care Med. 187 , 1294–1302. 10.1164/rccm.201208-1523CI (2013). Liu, X. et al. Comparison of Muscle Density in Middle-Aged and Older Chinese Adults Between a High-Altitude Area (Kunming) and a Low-Altitude Area (Beijing). 12 , (2021). 10.3389/fendo.2021.811770 Richalet, J. P., Hermand, E. & Lhuissier, F. J. Cardiovascular physiology and pathophysiology at high altitude. Nat. Rev. Cardiol. 21 , 75–88. 10.1038/s41569-023-00924-9 (2024). Thille, A. W., Cortés-Puch, I. & Esteban, A. Weaning from the ventilator and extubation in ICU. Curr. Opin. Crit. Care . 19 , 57–64. 10.1097/MCC.0b013e32835c5095 (2013). Wei, L. et al. Complement C3 participates in the function and mechanism of traumatic brain injury at simulated high altitude. Brain Res. 1726 , 146423. 10.1016/j.brainres.2019.146423 (2020). Mylne, D., Briggs, N. & Tolé, G. A formalised process for decannulation reduced number of days with tracheostomy for people with an acquired brain injury: a systematic review and meta-analysis %J Brain Impairment. 25, -, doi: (2024). https://doi.org/10.1071/IB23077 Zhou, T. et al. Tracheostomy decannulation protocol in patients with prolonged tracheostomy referred to a rehabilitation hospital: a prospective cohort study. J. Intensive Care . 10 , 34. 10.1186/s40560-022-00626-3 (2022). Kutsukutsa, J., Kuupiel, D., Monori-Kiss, A. & del Rey-Puech, P. & Mashamba-Thompson, T. P. Tracheostomy decannulation methods and procedures for assessing readiness for decannulation in adults: a systematic scoping review. JBI Evid. Implement. 17 (2019). Eskildsen, S. J., Irene, W., Ingrid, P., Aviaja, H. C., Curtis, D. J. & and Rehabilitative intervention for successful decannulation in adult patients with acquired brain injury and tracheostomy: a systematic review. Disabil. Rehabil. 46 , 2464–2476. 10.1080/09638288.2023.2233437 (2024). Ninan, A., Grubb, L. M., Brenner, M. J. & Pandian, V. Effectiveness of interprofessional tracheostomy teams: A systematic review. J. Clin. Nurs. 32 , 6967–6986. 10.1111/jocn.16815 (2023). Park, J. E. et al. Development of a Machine Learning Model for Predicting Weaning Outcomes Based Solely on Continuous Ventilator Parameters during Spontaneous Breathing Trials. 10 , 1163 (2023). Pai, K. C., Su, S. A., Chan, M. C., Wu, C. L. & Chao, W. C. Explainable machine learning approach to predict extubation in critically ill ventilated patients: a retrospective study in central Taiwan. BMC Anesthesiol. 22 , 351. 10.1186/s12871-022-01888-y (2022). Cuschieri, S. The STROBE guidelines. Saudi J. Anaesth. 13 , S31–s34. 10.4103/sja.SJA_543_18 (2019). Mannini, A. et al. Data-driven prediction of decannulation probability and timing in patients with severe acquired brain injury. Comput. Methods Programs Biomed. 209 , 106345. 10.1016/j.cmpb.2021.106345 (2021). Hao, P. Y., Chiang, J. H. & Chen, Y. D. Possibilistic classification by support vector networks. Neural networks: official J. Int. Neural Netw. Soc. 149 , 40–56. 10.1016/j.neunet.2022.02.007 (2022). Becker, T., Rousseau, A. J., Geubbelmans, M., Burzykowski, T. & Valkenborg, D. Decision trees and random forests. American journal of orthodontics and dentofacial orthopedics: official publication of the American Association of Orthodontists, its constituent societies, and the American Board of Orthodontics 164, 894–897, (2023). 10.1016/j.ajodo.2023.09.011 Zhao, Q. Y. et al. Development and Validation of a Machine-Learning Model for Prediction of Extubation Failure in Intensive Care Units. Front. Med. 8 , 676343. 10.3389/fmed.2021.676343 (2021). Abraham, A. et al. Machine learning for neuroimaging with scikit-learn. Front. neuroinformatics . 8 , 14. 10.3389/fninf.2014.00014 (2014). Lundberg, S. M. et al. From Local Explanations to Global Understanding with Explainable AI for Trees. Nat. Mach. Intell. 2 , 56–67. 10.1038/s42256-019-0138-9 (2020). Chalasani, R. D. & Radhika, Y. Prediction of ITK inhibitor kinases activity based on posterior probabilistic weighted average based ensemble voting classification. J. Intell. Fuzzy Syst. 44 , 5837–5846. 10.3233/JIFS-221412 (2023). Mutanga, R. T., Naicker, N. & Olugbara, O. O. J. I. J. o. A. C. S. & Applications (Detecting Hate Speech on Twitter Network using Ensemble Machine Learning, 2022). Kumar, R. et al. Ensemble learning-based early detection of influenza disease. Multimedia tools Appl. 1–21. 10.1007/s11042-023-15848-2 (2023). Schisterman, E. F., Perkins, N. J., Liu, A. & Bondell, H. Optimal Cut-point and Its Corresponding Youden Index to Discriminate Individuals Using Pooled Blood Samples. 16 , 73–81, (2005). 10.1097/01.ede.0000147512.81966.ba Wang, X. et al. Development and validation of a nomogram for decannulation in patients with neurological injury: A prognostic accuracy study. Front. Neurol. 13 10.3389/fneur.2022.979160 (2022). Hur, S. et al. Development and Validation of Unplanned Extubation Prediction Models Using Intensive Care Unit Data: Retrospective, Comparative, Machine Learning Study. J. Med. Internet. Res. 23 , e23508. 10.2196/23508 (2021). Huang, K. Y. et al. Developing a machine-learning model for real-time prediction of successful extubation in mechanically ventilated patients using time-series ventilator-derived parameters. Front. Med. 10 , 1167445. 10.3389/fmed.2023.1167445 (2023). Rodriguez-Galiano, V., Sanchez-Castillo, M., Chica-Olmo, M. & Chica-Rivas, M. Machine learning predictive models for mineral prospectivity: An evaluation of neural networks, random forest, regression trees and support vector machines. Ore Geol. Rev. 71 , 804–818. https://doi.org/10.1016/j.oregeorev.2015.01.001 (2015). Luo, W. et al. Guidelines for Developing and Reporting Machine Learning Predictive Models in Biomedical Research: A Multidisciplinary View. J. Med. Internet. Res. 18 , e323. 10.2196/jmir.5870 (2016). Parish, E. J. & Duraisamy, K. A paradigm for data-driven predictive modeling using field inversion and machine learning. J. Comput. Phys. 305 , 758–774. https://doi.org/10.1016/j.jcp.2015.11.012 (2016). Battineni, G., Sagaro, G. G., Chinatalapudi, N. & Amenta, F. Appl. Mach. Learn. Predictive Models Chronic Disease Diagnosis 10 , 21 (2020). Azodi, C. B. et al. Benchmarking Parametric and Machine Learning Models for Genomic Prediction of Complex Traits. G3 (Bethesda Md) . 9 , 3691–3702. 10.1534/g3.119.400498 (2019). Xiao, Q., Chang, H. H., Geng, G. & Liu, Y. An Ensemble Machine-Learning Model To Predict Historical PM(2.5) Concentrations in China from Satellite Data. Environ. Sci. Technol. 52 , 13260–13269. 10.1021/acs.est.8b02917 (2018). Luo, M. et al. Combination of Feature Selection and CatBoost for Prediction: The First Application to the Estimation of Aboveground Biomass. 12 , 216 (2021). Ghasemzadeh, H., Hillman, R. E. & Mehta, D. D. Toward Generalizable Machine Learning Models in Speech, Language, and Hearing Sciences: Estimating Sample Size and Reducing Overfitting. 67, 753–781, doi: (2024). 10.1044/2023_JSLHR-23-00273 Gneiting, T. & Raftery, A. E. Weather Forecasting with Ensemble Methods. 310 , 248–249, doi: (2005). 10.1126/science.1115255 Mienye, I. D., Sun, Y. A. & Survey of Ensemble Learning: Concepts, Algorithms, Applications, and Prospects. IEEE Access. 10 , 99129–99149. 10.1109/ACCESS.2022.3207287 (2022). Matloob, F. et al. Software Defect Prediction Using Ensemble Learning: A Systematic Literature Review. IEEE Access. 9 , 98754–98771. 10.1109/ACCESS.2021.3095559 (2021). Yuan, X. et al. Applied Machine Learning for Prediction of CO2 Adsorption on Biomass Waste-Derived Porous Carbons. Environ. Sci. Technol. 55 , 11925–11936. 10.1021/acs.est.1c01849 (2021). Chatzimparmpas, A., Martins, R. M., Kucher, K., Kerren, A. & StackGenVis Alignment of Data, Algorithms, and Models for Stacking Ensemble Learning Using Performance Metrics. IEEE Trans. Vis. Comput. Graph. 27 , 1547–1557. 10.1109/TVCG.2020.3030352 (2021). Gomes, H. M., Barddal, J. P., Enembreck, F. & Bifet, A. A Survey on Ensemble Learning for Data Stream Classification. 50 , Article 23, (2017). 10.1145/3054925 Xu, S. S. et al. Development of a Prediction Score for Evaluation of Extubation Readiness in Neurosurgical Patients with Mechanical Ventilation. Anesthesiology 139 , 614–627. 10.1097/aln.0000000000004721 (2023). Park, M. K. & Lee, S. J. Changes in Swallowing and Cough Functions Among Stroke Patients Before and After Tracheostomy Decannulation. Dysphagia 33 , 857–865. 10.1007/s00455-018-9920-9 (2018). Shem, K., Castillo, K., Wong, S. L., Chang, J. & Kolakowsky-Hayner, S. Dysphagia and respiratory care in individuals with tetraplegia: incidence, associated factors, and preventable complications. Top. spinal cord injury rehabilitation . 18 , 15–22. 10.1310/sci1801-15 (2012). Brodsky, M. B. et al. Respiratory-swallow phase patterns and their relationship to swallowing impairment in patients treated for oropharyngeal cancer. Head neck . 32 , 481–489. 10.1002/hed.21209 (2010). Osorio, B., Krakora, R., Epping, G., Wright, D. & Rajwani, K. Aspiration after critical illness: role of endotracheal tube, tracheostomy, swallowing disorders. Semin. Respir. Crit. Care Med. 10.1055/a-2445-9054 (2024). Labeit, B. et al. The assessment of dysphagia after stroke: state of the art and future directions. Lancet Neurol. 22 , 858–870. 10.1016/s1474-4422(23)00153-9 (2023). Chen, L. et al. Interventions for Postextubation Dysphagia in Critically Ill Patients: A Systematic Review and Meta-analysis. Dysphagia 10.1007/s00455-024-10695-1 (2024). Troll, C. et al. A bedside swallowing screen for the identification of post-extubation dysphagia on the intensive care unit - validation of the Gugging Swallowing Screen (GUSS)-ICU. BMC Anesthesiol. 23 10.1186/s12871-023-02072-6 (2023). Catalá-Ripoll, J. V., Monsalve-Naharro, J. & Hernández-Fernández, F. Incidence and predictive factors of diaphragmatic dysfunction in acute stroke. BMC Neurol. 20 10.1186/s12883-020-01664-w (2020). Oo, T., Watt, J. W., Soni, B. M. & Sett, P. K. Delayed diaphragm recovery in 12 patients after high cervical spinal cord injury. A retrospective review of the diaphragm status of 107 patients ventilated after acute spinal cord injury. Spinal cord . 37 , 117–122. 10.1038/sj.sc.3100775 (1999). Kaufman, M. R. et al. Prospective analysis of a surgical algorithm to achieve ventilator weaning in cervical tetraplegia. J. Spinal Cord Med. 45 , 531–535. 10.1080/10790268.2020.1829417 (2022). Chan, L. Y., Jones, A. Y., Chung, R. C. & Hung, K. N. Peak flow rate during induced cough: a predictor of successful decannulation of a tracheotomy tube in neurosurgical patients. Am. J. Crit. care: official publication Am. Association Critical-Care Nurses . 19 , 278–284. 10.4037/ajcc2009575 (2010). Torrini, F. et al. Prediction of extubation outcome in critically ill patients: a systematic review and meta-analysis. Crit. Care. (London, England) . 25 , 391. 10.1186/s13054-021-03802-3 (2021). Huang, H. Y. et al. Clinical outcomes and prognostic factors for prolonged mechanical ventilation in patients with acute stroke and brain trauma. J. Formos. Med. Association = Taiwan. yi zhi . 121 , 162–169. 10.1016/j.jfma.2021.02.011 (2022). Zanata Ide, L., Santos, R. S. & Hirata, G. C. Tracheal decannulation protocol in patients affected by traumatic brain injury. Int. archives Otorhinolaryngol. 18 , 108–114. 10.1055/s-0033-1363467 (2014). Mitchell, R. B. et al. Clinical consensus statement: tracheostomy care. Otolaryngology–head neck surgery: official J. Am. Acad. Otolaryngology-Head Neck Surg. 148 , 6–20. 10.1177/0194599812460376 (2013). Tang, Q. et al. Identification of blood exosomal metabolomic profiling for high-altitude cerebral edema. Sci. Rep. 14 , 11585. 10.1038/s41598-024-62360-0 (2024). Wang, X. et al. CX3CL1/CX3CR1 signal mediates M1-type microglia and accelerates high-altitude-induced forgetting. Front. Cell. Neurosci. 17 10.3389/fncel.2023.1189348 (2023). Zhang, L., Hu, Z., Li, Z. & Lin, Y. Crosstalk among mitophagy, pyroptosis, ferroptosis, and necroptosis in central nervous system injuries. Neural regeneration Res. 19 , 1660–1670. 10.4103/1673-5374.389361 (2024). Huang, S. W. et al. The Impact of Higher Protein Intake in Patients with Prolonged Mechanical Ventilation. Nutrients 14, (2022). 10.3390/nu14204395 Chang, C. H. et al. Higher protein intake may benefit in patients with prolonged mechanical ventilation. Front. Nutr. 11 , 1449240. 10.3389/fnut.2024.1449240 (2024). Misra, S. et al. Common Pathways of Epileptogenesis in Patients With Epilepsy Post-Brain Injury: Findings From a Systematic Review and Meta-analysis. Neurology 101 , e2243–e2256. 10.1212/wnl.0000000000207749 (2023). Adrie, C. et al. Persistent lymphopenia is a risk factor for ICU-acquired infections and for death in ICU patients with sustained hypotension at admission. Ann. Intensiv. Care. 7 10.1186/s13613-017-0242-0 (2017). Liao, T. Y., Chen, Y. L., Chen, Y. L., Kuo, Y. W. & Jerng, J. S. Persistent inflammation and lymphopenia and weaning outcomes of patients with prolonged mechanical ventilation. Respiratory Invest. 62 , 935–941. 10.1016/j.resinv.2024.08.001 (2024). Schneider, B., Oliveira, R. A., Friedman, G. & Moraes, R. B. Association of biomarkers with successful ventilatory weaning in COVID-19 patients: an observational study. Critical care science 36, e20240158en, (2024). 10.62675/2965-2774.20240158-en Soeters, P. B., Wolfe, R. R., Shenkin, A. & Hypoalbuminemia Pathogenesis and Clinical Significance. JPEN J. Parenter. Enteral. Nutr. 43 , 181–193. 10.1002/jpen.1451 (2019). Igarashi, Y. et al. Machine learning for predicting successful extubation in patients receiving mechanical ventilation. Front. Med. 9 , 961252. 10.3389/fmed.2022.961252 (2022). Zeng, Z., Tang, X., Liu, Y., He, Z. & Gong, X. Interpretable recurrent neural network models for dynamic prediction of the extubation failure risk in patients with invasive mechanical ventilation in the intensive care unit. BioData Min. 15 , 21. 10.1186/s13040-022-00309-7 (2022). Kuo, H. J. et al. Improvement in the Prediction of Ventilator Weaning Outcomes by an Artificial Neural Network in a Medical ICU. Respir. Care . 60 , 1560–1569. 10.4187/respcare.03648 (2015). Fabregat, A. et al. A Machine Learning decision-making tool for extubation in Intensive Care Unit patients. Comput. Methods Programs Biomed. 200 10.1016/j.cmpb.2020.105869 (2021). Additional Declarations No competing interests reported. Supplementary Files Supplementarymaterialsv2.0.pdf modellingcode.ipynb Cite Share Download PDF Status: Published Journal Publication published 22 May, 2025 Read the published version in Scientific Reports → Version 1 posted Editorial decision: Accepted 15 Apr, 2025 Reviews received at journal 10 Apr, 2025 Reviewers agreed at journal 10 Apr, 2025 Reviewers invited by journal 08 Apr, 2025 Submission checks completed at journal 07 Apr, 2025 First submitted to journal 26 Mar, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-5549340","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":439934179,"identity":"be66cfd1-41d7-4617-8060-a52a7f47e0ef","order_by":0,"name":"Zihan Chen","email":"","orcid":"","institution":"Department of rehabilitation medicine, the Second Affiliated Hospital of Kunming Medical University","correspondingAuthor":false,"prefix":"","firstName":"Zihan","middleName":"","lastName":"Chen","suffix":""},{"id":439934180,"identity":"4fd13bbd-e603-4fb4-ab03-989b46373ffd","order_by":1,"name":"Haotian Wu","email":"","orcid":"","institution":"Department of rehabilitation medicine, the Second Affiliated Hospital of Kunming Medical University","correspondingAuthor":false,"prefix":"","firstName":"Haotian","middleName":"","lastName":"Wu","suffix":""},{"id":439934181,"identity":"cad9f1b4-3b14-48cc-a553-c0377401a7ca","order_by":2,"name":"Yao Zhou","email":"","orcid":"","institution":"Department of rehabilitation medicine, the Second Affiliated Hospital of Kunming Medical University","correspondingAuthor":false,"prefix":"","firstName":"Yao","middleName":"","lastName":"Zhou","suffix":""},{"id":439934182,"identity":"c8973ef3-766d-47c3-9538-419fc6505989","order_by":3,"name":"Qian Liu","email":"","orcid":"","institution":"Department of rehabilitation medicine, the Second Affiliated Hospital of Kunming Medical University","correspondingAuthor":false,"prefix":"","firstName":"Qian","middleName":"","lastName":"Liu","suffix":""},{"id":439934183,"identity":"290edeb6-c2c5-41ee-a58c-3bb27406f9a4","order_by":4,"name":"Hongmei Zhang","email":"","orcid":"","institution":"Department of rehabilitation medicine, the Second Affiliated Hospital of Kunming Medical University","correspondingAuthor":false,"prefix":"","firstName":"Hongmei","middleName":"","lastName":"Zhang","suffix":""},{"id":439934184,"identity":"63a2f00c-be1c-4dff-b0fa-91b34832177b","order_by":5,"name":"Xiaochen Li","email":"","orcid":"","institution":"Department of rehabilitation medicine, the Second Affiliated Hospital of Kunming Medical University","correspondingAuthor":false,"prefix":"","firstName":"Xiaochen","middleName":"","lastName":"Li","suffix":""},{"id":439934185,"identity":"95280598-e84c-4857-8c60-888cd5f3bb65","order_by":6,"name":"Xue Yang","email":"","orcid":"","institution":"Department of rehabilitation medicine, the Second Affiliated Hospital of Kunming Medical University","correspondingAuthor":false,"prefix":"","firstName":"Xue","middleName":"","lastName":"Yang","suffix":""},{"id":439934186,"identity":"3ba057ee-8032-4272-9afd-d5aa4675eccc","order_by":7,"name":"Liqing Yao","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABAklEQVRIiWNgGAWjYHACxgMJDBIMDOzNx39+MLDh4edvIKwHooXnWIK0REGajOSMA0RoAZMSOQYSPB8O2xg0JOBXbnD88IEDD8os8uQj0hIMJAzO8xgwHGD88DEHj5YzaQkHEs5JFBueeXwgocDgNo85cwOz5MxteLQcyDE4kNgmkbixHahXAqjFsuEAGzMvPi3n30C1NOQYNvAYnOMxOJBAQMsNqC3zOXKMGYDqCWuRvPEM7JfEDTzH0pglDJJ5JGccbMbrF77zyQcf/iirS5zf3nyM8cMfO3t+/uaDHz7i0aJwAESygcIBLsbYgFs9EMg3QLXI41c3CkbBKBgFIxkAALq5XQw+dSsqAAAAAElFTkSuQmCC","orcid":"","institution":"Department of rehabilitation medicine, the Second Affiliated Hospital of Kunming Medical University","correspondingAuthor":true,"prefix":"","firstName":"Liqing","middleName":"","lastName":"Yao","suffix":""}],"badges":[],"createdAt":"2024-11-29 13:08:35","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-5549340/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-5549340/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1038/s41598-025-98861-9","type":"published","date":"2025-05-22T15:58:27+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":80278480,"identity":"e5cd8fbc-dcb2-4074-826a-43eb8d58ae31","added_by":"auto","created_at":"2025-04-10 05:28:21","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":15642676,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eFeature selection and model optimization and development. \u003c/strong\u003e(A) Among all the clinical data collected in this study, the features marked with red stars were those associated with extubation outcomes and were utilized for subsequent modeling. On the left is the distribution plot of the model's test and training sets, along with the proportion of extubation-success patients in each subset. (\u003cstrong\u003eB\u003c/strong\u003e) Flowchart of feature selection and model development. (\u003cstrong\u003eC\u003c/strong\u003e) Comparison of the ACC and AUC before and after optimization for each model. Accuracy (ACC), area under the receiver operating characteristic curve (AUC), K-nearest neighbors (K-NN), support vector machine (SVM), Gaussian naive Bayes (GNB), decision tree (DT), random forest (RF), extra tree classifier (ET), gradient boosting (GB), and logistic regression (LR).\u003c/p\u003e","description":"","filename":"Figure1Flow.png","url":"https://assets-eu.researchsquare.com/files/rs-5549340/v1/64cfda5f638899fa00ed05b0.png"},{"id":80278496,"identity":"211447d8-c7a4-4664-becc-0522aa87caff","added_by":"auto","created_at":"2025-04-10 05:28:22","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":61337095,"visible":true,"origin":"","legend":"\u003cp\u003eThe performance of the four models with an AUC greater than 0.85 and an ACC greater than 80%, including the confusion matrix, AUC curve, and learning curve for each model. Accuracy (ACC), areaunder the receiver operating characteristic curve (AUC), sensitivity (SEN), specificity (SPE), Youden's index(YDI), positive predictive value(PPV), negative predictive value (NPV), and K-nearest neighbors (K-NN) were used.\u003c/p\u003e","description":"","filename":"Figure2Modelperformance.png","url":"https://assets-eu.researchsquare.com/files/rs-5549340/v1/f1ad9456548fafd4e08ac12b.png"},{"id":80278479,"identity":"c528615e-dd46-4a73-9d81-5dc7896e17ba","added_by":"auto","created_at":"2025-04-10 05:28:21","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":2262171,"visible":true,"origin":"","legend":"\u003cp\u003eConfusion matrices and AUCsof the WPV ensemble model and its submodels. Three confusion matrices of the independent models that make up the ensemble model (top), their AUC curves (bottom), evaluation metrics of the ensemble model (left middle), and the confusion matrix and AUC curve of the ensemble model (right middle). Weighted Posteriors Voting (WPV), Accuracy (ACC), Area Under the Receiver Operating Characteristic Curve (AUC), Sensitivity (SEN), Specificity (SPE), Youden's Index (YDI), Positive Predictive Value (PPV), Negative Predictive Value (NPV), and K-Nearest Neighbors (K-NN).\u003c/p\u003e","description":"","filename":"Figure3WPV.png","url":"https://assets-eu.researchsquare.com/files/rs-5549340/v1/ea40ba88bfa36374c937f4a1.png"},{"id":80278483,"identity":"a9dcf04e-5627-4d1b-b97e-78847243e55e","added_by":"auto","created_at":"2025-04-10 05:28:21","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":25673134,"visible":true,"origin":"","legend":"\u003cp\u003eBar summary plots and SHAP summary plots of the four models with AUCs greater than 0.85 and ACCs greater than 80%. In the bar chart, the top three features contributing to the model are highlighted in red. In the SHAP plot, points to the left of the zero coordinate represent failed extubation, while points to the right represent successful extubation. Shapley additive explanations (SHAP), Glasgow Coma Scale (GCS), Upper Respiratory Tract Structural Abnormalities (URSA), Interleukin 6 (IL-6), Albumin (ALB), Pulmonary Embolism (PE), K-Nearest Neighbors (K-NN).\u003c/p\u003e","description":"","filename":"Figure4SHAP.png","url":"https://assets-eu.researchsquare.com/files/rs-5549340/v1/d3598f5a31c5da0b1d7745f5.png"},{"id":80279581,"identity":"ad411f7e-74dc-440d-a1f7-1b8bb628af97","added_by":"auto","created_at":"2025-04-10 05:36:21","extension":"pdf","order_by":5,"title":"","display":"","copyAsset":false,"role":"supplement","size":1018494,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementarymaterialsv2.0.pdf","url":"https://assets-eu.researchsquare.com/files/rs-5549340/v1/d2fff0ebeef76988590ea5f9.pdf"},{"id":80278476,"identity":"37722343-b709-4aab-95a4-e175da559da7","added_by":"auto","created_at":"2025-04-10 05:28:20","extension":"ipynb","order_by":6,"title":"","display":"","copyAsset":false,"role":"supplement","size":579200,"visible":true,"origin":"","legend":"","description":"","filename":"modellingcode.ipynb","url":"https://assets-eu.researchsquare.com/files/rs-5549340/v1/4690406232e047e2be57945a.ipynb"}],"financialInterests":"No competing interests reported.","formattedTitle":"Using ML techniques to predict extubation outcomes for patients with central nervous system injuries in the Yun-Gui Plateau","fulltext":[{"header":"Introduction","content":"\u003cp\u003eAlthough tracheostomy is crucial for severe central nervous system (CNS) injuries, it can also lead to a range of complications, including bleeding, unstable oxygen saturation, infections, and granulation tissue formation\u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003e. Therefore, extubation assessment is an important step in reducing complications and promoting functional recovery in patients. Factors such as age, swallowing ability, cough reflex, and infection status are associated with successful extubation in patients with CNS injuries\u003csup\u003e\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e. Additionally, some studies have shown that swallowing function is a reliable indicator of extubation outcomes in stroke patients\u003csup\u003e\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u003c/sup\u003e. Most current studies on extubation predictors rely on traditional statistical analyses\u003csup\u003e\u003cspan additionalcitationids=\"CR5\" citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u003c/sup\u003e, which have provided some clinical insights for extubation decisions. However, the extubation failure rate remains above 10%\u003csup\u003e7\u003c/sup\u003e. In order to improve the accuracy of extubation predictions, further research on new methods is needed.\u003c/p\u003e \u003cp\u003eIn the Yunnan-Guizhou Plateau, Kunming is situated at a relatively high altitude, approximately around 1900 meters\u003csup\u003e\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u003c/sup\u003e. The unique characteristics of high-altitude environments include low oxygen and low air pressure conditions, which directly impact the respiratory and cardiovascular systems of patients. Wang \u003cem\u003eet al.\u003c/em\u003e concluded that long-term hypoxia in high-altitude areas can cause pulmonary vasoconstriction, remodeling, increase the risk of pulmonary hypertension, worsen right ventricular burden, and impact the function of the circulatory and respiratory systems\u003csup\u003e\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e\u003c/sup\u003e. One common reason for extubation failure in the ICU is cardiovascular system dysfunction\u003csup\u003e\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e\u003c/sup\u003e. Moreover, high-altitude CNS injury patients are more likely to develop severe brain edema, resulting in higher disability and mortality rates compared to individuals in plain areas\u003csup\u003e\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u003c/sup\u003e. Overall, these patients with poor prognosis may have a lower success rate in extubation and an increased risk of reintubation. However, it is currently unclear which factors will affect the extubation outcomes of CNS injury patients in high-altitude regions. Therefore, one of the aims of this study is to explore factors influencing extubation outcomes in such patients.\u003c/p\u003e \u003cp\u003eBoth the tracheostomy decannulation process in other departments and in the rehabilitation department emphasize formalized protocols and interprofessional collaboration\u003csup\u003e\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e,\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u003c/sup\u003e. In acute care environments, decannulation decisions are typically driven by factors such as the patient\u0026rsquo;s ability to tolerate tracheostomy tube capping, the effectiveness of cough, and the control of secretions\u003csup\u003e\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u003c/sup\u003e. The rehabilitation departments, by contrast, incorporate comprehensive assessments that also evaluate functional status, swallowing ability, and the patient\u0026rsquo;s readiness\u003csup\u003e\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e\u003c/sup\u003e. Interdisciplinary tracheostomy teams, comprising rehabilitation specialists, respiratory therapists, speech-language pathologists, and nurses, improve decision-making processes and customize decannulation protocols to meet each patient's specific needs\u003csup\u003e\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u003c/sup\u003e. Despite the comprehensive considerations of the rehabilitation department and its emphasis on personalization, there is currently a lack of research using machine learning (ML) methods to predict extubation outcomes.\u003c/p\u003e \u003cp\u003eTraditional statistical methods for predicting extubation typically use a small set of predetermined variables and a straightforward model structure based on linear relationships. This may overlook complex nonlinear relationships in clinical data. Modern ML techniques are capable of handling high-dimensional data without strict assumptions. They can capture complex interactions hidden in the data, leading to a significant improvement in prediction accuracy and sensitivity\u003csup\u003e\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e\u003c/sup\u003e. ML methods can fully utilize data from mechanical ventilators, multi-parameter monitoring devices, and physiological signals, providing a more personalized and dynamic predictive basis. At the same time, another major advantage of ML models is the development of their interpretability tools. SHAP (Shapley Additive exPlanations) can unravel complex models, quantify the impact of each input feature, and help clinical doctors understand the model's decision logic\u003csup\u003e\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e\u003c/sup\u003e. This approach not only helps identify key factors influencing extubation outcomes, but also provides a theoretical basis for developing clinical interventions based on these factors.\u003c/p\u003e \u003cp\u003eHowever, there is limited research on predictive models for extubation outcomes in high-altitude regions. Therefore, the aim of this retrospective study was to determine the optimal ML models for this patient population and to identify factors associated with successful extubation in tracheostomized patients in the Plateau. There were 8 conventional models developed, as well as a Weighted Posterior Vote (WPV) ensemble model. The predictive performance of the models was subsequently compared using evaluation metrics including accuracy (ACC), area under the receiver operating characteristic curve (AUC), sensitivity, specificity, Youden's index, positive predictive value (PPV), and negative predictive value (NPV). At last, the SHAP analysis was used to evaluate the influence of features on the models and extubation success rate.\u003c/p\u003e"},{"header":"Materials and methods","content":"\u003cp\u003e\u003cstrong\u003eStudy design\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe study design followed the STROBE guidelines for a retrospective, nonsynchronous study \u003csup\u003e19\u003c/sup\u003e. The data were derived from 501 adult tracheostomy patients (aged over 18) at the Department of Rehabilitation, Second Affiliated Hospital of Kunming Medical University, between August 2016 and November 2022, with 196 successful extubations. Upon admission, patients underwent a functional assessment by skilled Clinical Assessor. Inclusion criteria: (1) Tracheostomy patients; (2) Patients undergoing extubation surgery in our department; (3) with CNS injuries; (4) aged 18 to 85 years old. Exclusion criteria: (1) Patients with unstable conditions or needing to be transferred to other departments for treatment; (2) with incomplete clinical data.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eInformed consent statement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eInformed consent to participate was waived by Ethics Committee of the Second Affiliated Hospital of Kunming Medical University (IRB). In cases where participants cannot be located and the research project does not involve personal privacy or commercial interests, informed consent may be waived with approval from the ethics committee, as outlined in the Ethical Review Procedures for Biomedical Research Involving Human Participants by the Chinese National Health Commission.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics approval\u003c/strong\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe study design was approved by the Ethics Committee of the Second Affiliated Hospital of Kunming Medical University (Ethics approval number: Shen-PJ-ke-2023-28). The study has been performed in accordance with the Declaration of Helsinki.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData collection\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe study collected demographic, behavioral, and clinical data from tracheostomy patients: (1) demographic data, such as sex and age; (2) behavioral factors, such as smoking history, primary disease, paralysis, consciousness and cognitive status, and medical history (epilepsy, hypertension, diabetes, heart disease); (3) clinical indicators, including wet rales, dysphagia, pulmonary infection, sputum culture, sputum viscosity,\u0026nbsp;upper respiratory tract structural abnormalities (URSA), and Glasgow Coma Scale (GCS); and (4) blood physiology, including procalcitonin,\u0026nbsp;interleukin\u0026nbsp;6\u0026nbsp;(IL-6),\u0026nbsp;high-sensitivity C-reactive protein\u0026nbsp;(hs-CRP), white blood cells, neutrophils, lymphocytes, monocytes, hemoglobin,\u0026nbsp;and\u0026nbsp;albumin. Encode all categorical variables using label encoding. The specific numerical symbols\u0026nbsp;are\u0026nbsp;presented in\u0026nbsp;the\u0026nbsp;Supplementary Material (Table S1). All\u0026nbsp;the\u0026nbsp;data were managed and stored centrally by the hospital\u0026apos;s medical records department. Some clinical indicators\u0026apos; diagnostic criteria are detailed in the supplementary materials. The data of all patients included in this study was complete, hence there were no missing values in the complete data.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFeature screening\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo deal with the potential data imbalance, we executed feature selection and standardization. Logistic regression was used to identify the factors related to extubation outcomes. The analysis was conducted using the \u003cem\u003eStatsmodels\u003c/em\u003e package in \u003cem\u003ePython 3.12.4\u003c/em\u003e, with a significance level of 0.05. To avoid overlooking hidden information, weakly correlated variables (P ~ 0.05) were also included in the model construction\u003csup\u003e20\u003c/sup\u003e. Next, the selected feature values were standardized. The mean and standard deviation of the training set were calculated, and the training set data was transformed using zero-mean normalization method to have a distribution with a mean of 0 and standard deviation of 1. To prevent information leakage from the test set to the training phase, the test set was normalized using the mean and standard deviation calculated from the training set.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eModel selection and construction\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eA total of eight models were constructed, including\u0026nbsp;the k-nearest neighbors (K-NN), support vector machine (SVM)\u003csup\u003e21\u003c/sup\u003e, Gaussian naive Bayes (GNB), decision tree (DT), random forest (RF)\u003csup\u003e22\u003c/sup\u003e, extra tree classifier (ET), gradient boosting (GB), and logistic regression (LR). Modeling parameters were selected\u0026nbsp;via\u0026nbsp;cross-validation and hyperparameter optimization. Consistent data splitting was chosen to ensure comparability among multiple models. The 80%-20% split balanced model capacity with evaluation reliability, restrained overfitting, and optimized computational resources\u003csup\u003e20,23\u003c/sup\u003e. Data preprocessing and dataset splitting were carried out using the \u003cem\u003eNumpy\u003c/em\u003e and \u003cem\u003eSklearn\u003c/em\u003e packages in \u003cem\u003ePython 3.12.4\u003c/em\u003e\u003csup\u003e24\u003c/sup\u003e. The details of hyperparameter optimization are provided in the Supplementary Material (Table S2). The workflow of the model construction is shown in Figure 1.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFeature analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eSHAP value was used to assess the significance of each feature. This function, implemented via the \u003cem\u003eSHAP\u003c/em\u003e package in \u003cem\u003ePython 3.12.4\u003c/em\u003e\u003csup\u003e25\u003c/sup\u003e, employs\u0026nbsp;a game-theoretic approach to evaluate the importance of each feature. SHAP values not only rank features by their importance to the model but also reveal the impact of each feature on patient extubation outcomes\u003csup\u003e23\u003c/sup\u003e. Only models with an AUC greater than 0.85 and an ACC greater than 80% are discussed in the discussion section.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEnsemble model construction\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAn ensemble model was constructed using \u003cem\u003eSklearn\u003c/em\u003e packages in \u003cem\u003ePython 3.12.4\u003c/em\u003e. Emerging methods such as WPV offer a unique way to incorporate model confidence levels into the final decision, addressing issues related to class imbalance and overfitting in noisy environments when compared to ensemble methods\u003csup\u003e26\u003c/sup\u003e. We employed the hard classification voting ensemble in our model in detail. The hard voting scheme is suited for predicting distinct class labels, while soft voting is appropriate for predicting continuous values\u003csup\u003e27\u003c/sup\u003e.\u0026nbsp;In theory, each base classifier produces a binary output (1 or 0). WPV conducts voting based on the frequency of each prediction. In the scenario where RF predicts 1, GNB predicts 1, and K-NN predicts 0, the final prediction will be 1 because it is the most common outcome. In model ensemble, the diversity and complementarity of models were taken into account\u003csup\u003e28\u003c/sup\u003e.\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003e\u003cstrong\u003eStatistical analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eLogistic regression was performed with extubation outcomes as the independent variable and other factors as the dependent variables. We found that the outcomes of extubation were associated with several factors: disease duration (P = 0.002), underlying condition (P \u0026lt; 0.001), quadriplegia (P \u0026lt; 0.001), GCS score (P = 0.040), swallowing function (P \u0026lt; 0.001), sputum culture pathogens (P = 0.013), interleukin-6 levels (P = 0.023), lymphocyte count (P = 0.046), and airway structural abnormalities (P \u0026lt; 0.001). To avoid omitting relevant information, features with P-values close to 0.05 were also included in the model. These included mental disorders (P = 0.05), pulmonary embolism (P = 0.087), and albumin levels (P = 0.081). The statistical analysis results are presented in Table 1.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 1\u003c/strong\u003e.\u0026nbsp;\u003cstrong\u003eLogistic regression results for each variable.\u003c/strong\u003e Categorical variables are presented as n (%), and continuous variables are presented as mean [SD]. The 95% confidence interval is presented as (Lower bound, Upper bound).\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"95%\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 74px;\"\u003e\n \u003cp\u003eVariables\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003eSuccessful Extubation \u0026nbsp;(n=196)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003eFailed Extubation \u0026nbsp;(n=305)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47px;\"\u003e\n \u003cp\u003eOR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 50px;\"\u003e\n \u003cp\u003eStandard Error\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 75px;\"\u003e\n \u003cp\u003eCI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 50px;\"\u003e\n \u003cp\u003eP -Value\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"2\" style=\"width: 74px;\"\u003e\n \u003cp\u003eMeal gender\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003eyes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e148 (75.51)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e232 (76.06)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0.99\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.332\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 75px;\"\u003e\n \u003cp\u003e(-0.672, 0.628)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.947\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003eno\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e48 (24.49)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e73 (23.93)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"2\" style=\"width: 74px;\"\u003e\n \u003cp\u003eWet Rales\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003eyes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e141 (71.94)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e215 (70.49)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0.87\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.302\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 75px;\"\u003e\n \u003cp\u003e(-0.728, 0.456)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.653\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003eno\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e55 (29.06)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e90 (29.51)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"2\" style=\"width: 74px;\"\u003e\n \u003cp\u003eSmoking\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003eyes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e123 (62.76)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e222 (72.79)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 47px;\"\u003e\n \u003cp\u003e1.25\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.306\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 75px;\"\u003e\n \u003cp\u003e(-0.38, 0.819)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.472\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003eno\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e73 (37.25)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e83 (27.21)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"2\" style=\"width: 74px;\"\u003e\n \u003cp\u003eConsciousness Disorder\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003eyes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e122 (62.24)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e213 (69.84)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 47px;\"\u003e\n \u003cp\u003e1.28\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.730\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 75px;\"\u003e\n \u003cp\u003e(-1.184, 1.676)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.736\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003eno\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e74 (37.76)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e92 (30.16)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"2\" style=\"width: 74px;\"\u003e\n \u003cp\u003eDysphagia\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003eyes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e103 (52.55)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e226 (74.09)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0.13\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.320\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 75px;\"\u003e\n \u003cp\u003e(-2.696, -1.441)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003eno\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e93 (47.45)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e32 (10.49)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"2\" style=\"width: 74px;\"\u003e\n \u003cp\u003eEpilepsy\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003eyes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e167 (85.20)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e226 (74.10)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0.66\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.339\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 75px;\"\u003e\n \u003cp\u003e(-1.076, 0.252)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.224\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003eno\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e29 (14.80)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e79 (25.90)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"2\" style=\"width: 74px;\"\u003e\n \u003cp\u003eViscous sputum\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003eyes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e147 (75.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e265 (86.89)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0.62\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.328\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 75px;\"\u003e\n \u003cp\u003e(-1.121, 0.163)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 50px;\"\u003e\n \u003cp\u003e0.144\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003eno\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e49 (25.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e40 (13.115)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 50px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"2\" style=\"width: 74px;\"\u003e\n \u003cp\u003eUpper Respiratory Tract Structural Abnormalities (URSA)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003eyes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e191 (97.45)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e264 (86.56)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47px;\"\u003e\n \u003cp\u003e0.10\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 50px;\"\u003e\n \u003cp\u003e0.642\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 75px;\"\u003e\n \u003cp\u003e(-3.584, -1.069)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 50px;\"\u003e\n \u003cp\u003e0.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003eno\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e5 (2.551)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e41 (13.44)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 50px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 75px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 50px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"2\" style=\"width: 74px;\"\u003e\n \u003cp\u003eHypertension\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003eyes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e106 (54.08)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e160 (52.46)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0.73\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.301\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 75px;\"\u003e\n \u003cp\u003e(-0.898, 0.282)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.306\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003eno\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e90 (45.918)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e145 (47.54)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"2\" style=\"width: 74px;\"\u003e\n \u003cp\u003eCoronary Heart Disease\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003eyes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e190 (96.94)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e292 (95.74)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 47px;\"\u003e\n \u003cp\u003e1.69\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.696\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 75px;\"\u003e\n \u003cp\u003e(-0.839, 1.888)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.451\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003eno\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e6 (3.06)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e13 (4.26)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"2\" style=\"width: 74px;\"\u003e\n \u003cp\u003eDiabetes Mellitus\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003eyes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e166 (84.69)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e247 (80.98)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 47px;\"\u003e\n \u003cp\u003e1.07\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.367\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 75px;\"\u003e\n \u003cp\u003e(-0.651, 0.787)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.853\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003eno\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e30 (15.31)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e58 (19.02)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"2\" style=\"width: 74px;\"\u003e\n \u003cp\u003ePulmonary Infection\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003eyes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e167 (85.20)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e248 (81.31)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 47px;\"\u003e\n \u003cp\u003e1.78\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.372\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 75px;\"\u003e\n \u003cp\u003e(-0.15, 1.306)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.12\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003eno\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e29 (14.796)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e57 (18.69)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"2\" style=\"width: 74px;\"\u003e\n \u003cp\u003ePulmonary Embolism\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003eyes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e186 (94.90)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e274 (89.84)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0.40\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.536\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 75px;\"\u003e\n \u003cp\u003e(-1.969, 0.133)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.087\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003eno\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e10 (5.10)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e10 (3.28)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"2\" style=\"width: 74px;\"\u003e\n \u003cp\u003ePulmonary Contusion\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003eyes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e182 (92.85)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e295 (96.72)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 47px;\"\u003e\n \u003cp\u003e1.09\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.594\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 75px;\"\u003e\n \u003cp\u003e(-1.079, 1.252)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.884\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003eno\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e14 (7.14)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e31 (10.16)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"2\" style=\"width: 74px;\"\u003e\n \u003cp\u003eMental Disorders\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003eyes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e175 (89.29)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e294 (96.39)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 47px;\"\u003e\n \u003cp\u003e2.99\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.560\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 75px;\"\u003e\n \u003cp\u003e(-0.002, 2.193)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.05\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003eno\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e21 (10.71)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e11 (3.61)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"4\" style=\"width: 74px;\"\u003e\n \u003cp\u003eDisease Duration\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003eDD \u0026le; 30 days\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e3 (1.53)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e42 (13.77)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"4\" style=\"width: 47px;\"\u003e\n \u003cp\u003e1.57\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"4\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.150\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"4\" style=\"width: 75px;\"\u003e\n \u003cp\u003e\u0026nbsp;(0.16, 0.747)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"4\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.002\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e30 days \u0026lt; DD \u0026le; 60 days\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e71 (36.22)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e110 (36.07)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e60 days \u0026lt; DD \u0026le; 90 days\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e56 (28.57)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e60 (19.67)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003eDD \u0026gt; 90 days\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e66 (33.67)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e93 (30.49)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"8\" style=\"width: 74px;\"\u003e\n \u003cp\u003ePrimary Disease\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003eCerebral Trauma\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e69 (35.20)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e81 (26.56)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"8\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0.78\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"8\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.071\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"8\" style=\"width: 75px;\"\u003e\n \u003cp\u003e\u0026nbsp;(-0.383, -0.106)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"8\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003eCerebral Hemorrhage\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e78 (39.80)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e124 (40.66)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003eCerebral Infarction\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e21 (10.71)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e39 (12.79)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003eSubarachnoid Hemorrhage\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e4 (2.04)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e3 (0.98)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003eIschemic Hypoxic Brain Disease\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e5 (2.55)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e17 (5.57)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003eCentral Nervous System Infectious Diseases\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e3 (1.53)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e7 (2.30)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003eSpinal Cord Injury\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e7 (3.57)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e14 (4.59)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003eOther Diseases\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e9 (4.59)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e20 (6.56)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"3\" style=\"width: 74px;\"\u003e\n \u003cp\u003eLimb Paralysis\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003eQuadriplegia\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e133 (67.86)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e265 (86.89)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"3\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0.20\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"3\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.351\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"3\" style=\"width: 75px;\"\u003e\n \u003cp\u003e\u0026nbsp;(-2.319, -0.943)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"3\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003eHemiplegia\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e63 (32.14)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e40 (13.11)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003eParaplegia\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e0 (0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e0 (0)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"8\" style=\"width: 74px;\"\u003e\n \u003cp\u003eSputum Culture Pathogens\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e\u003cem\u003ePseudomonas aeruginosa\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e35 (17.85)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e76 (24.92)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"8\" style=\"width: 47px;\"\u003e\n \u003cp\u003e1.12\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"8\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.047\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"8\" style=\"width: 75px;\"\u003e\n \u003cp\u003e\u0026nbsp;(0.025, 0.21)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"8\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.013\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e\u003cem\u003eKlebsiella pneumoniae\u003c/em\u003e subtype\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e10 (5.10)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e23 (7.54)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e\u003cem\u003eAcinetobacter baumannii\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e12 (6.12)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e18 (5.90)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e\u003cem\u003eEscherichia coli\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e5 (2.55)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e9 (2.95)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e\u003cem\u003eSerratia marcescens\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e1 (0.51)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e9 (2.95)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003eOther Pathogens\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e12 (6.12)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e15 (4.92)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003eTwo or more pathogens\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e64 (32.65)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e79 (25.90)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003eNone\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e57 (29.08)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e76 (24.92)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 74px;\"\u003e\n \u003cp\u003eAge \u0026nbsp; \u0026nbsp; (year)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e51.2[17.17]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e54.18[16.40]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0.99\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.009\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 75px;\"\u003e\n \u003cp\u003e\u0026nbsp;(-0.027, 0.009)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.342\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 74px;\"\u003e\n \u003cp\u003eGCS\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e12.69[3.17]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e9.82[3.81]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 47px;\"\u003e\n \u003cp\u003e1.24\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.106\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 75px;\"\u003e\n \u003cp\u003e\u0026nbsp;(0.01, 0.427)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.040\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 74px;\"\u003e\n \u003cp\u003eProcalcitonin \u0026nbsp;(ng/ml)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e0.12[0.18]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e0.32[1.24]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0.46\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.576\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 75px;\"\u003e\n \u003cp\u003e\u0026nbsp;(-1.914, 0.344)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.173\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 74px;\"\u003e\n \u003cp\u003eInterleukin 6 (pg/ml)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e16.88[19.67]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e41.21[63.13]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0.98\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.007\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 75px;\"\u003e\n \u003cp\u003e\u0026nbsp;(-0.028, -0.002)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.023\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 74px;\"\u003e\n \u003cp\u003eHigh-Sensitivity C-Reactive Protein \u0026nbsp;(mg/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e18.98[23.04]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e33.52[37.04]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 47px;\"\u003e\n \u003cp\u003e1.00\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.006\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 75px;\"\u003e\n \u003cp\u003e\u0026nbsp;(-0.012, 0.011)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.963\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 74px;\"\u003e\n \u003cp\u003eWhite Blood Cell Count \u0026nbsp;(10\u003csup\u003e9\u003c/sup\u003e/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e8.58[3.03]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e10.01[5.54]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 47px;\"\u003e\n \u003cp\u003e1.02\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.073\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 75px;\"\u003e\n \u003cp\u003e\u0026nbsp;(-0.119, 0.166)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.746\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 74px;\"\u003e\n \u003cp\u003eNeutrophils \u0026nbsp;(10\u003csup\u003e9\u003c/sup\u003e/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e6.1[2.78]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e7.44[3.43]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0.90\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.090\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 75px;\"\u003e\n \u003cp\u003e\u0026nbsp;(-0.28, 0.073)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.249\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 74px;\"\u003e\n \u003cp\u003eLymphocytes \u0026nbsp;(10\u003csup\u003e9\u003c/sup\u003e/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e1.68[0.69]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e1.53[0.66]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 47px;\"\u003e\n \u003cp\u003e1.56\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.223\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 75px;\"\u003e\n \u003cp\u003e\u0026nbsp;(0.008, 0.884)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.046\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 74px;\"\u003e\n \u003cp\u003eMonocytes \u0026nbsp;(10\u003csup\u003e9\u003c/sup\u003e/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e0.56[0.25]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e0.62[0.27]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0.68\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.632\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 75px;\"\u003e\n \u003cp\u003e\u0026nbsp;(-1.623, 0.853)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.542\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 74px;\"\u003e\n \u003cp\u003eHemoglobin \u0026nbsp;(g/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e111.79[14.48]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e107.06[15.75]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 47px;\"\u003e\n \u003cp\u003e1.00\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.010\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 75px;\"\u003e\n \u003cp\u003e\u0026nbsp;(-0.019, 0.021)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.912\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 74px;\"\u003e\n \u003cp\u003eAlbumin \u0026nbsp;(g/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e35.89[3.36]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 85px;\"\u003e\n \u003cp\u003e34.04[4.68]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 47px;\"\u003e\n \u003cp\u003e1.07\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.038\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 75px;\"\u003e\n \u003cp\u003e\u0026nbsp;(-0.008, 0.141)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0.081\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cstrong\u003eModel performance\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eFollowing feature selection, all models were developed. Subsequently, hyperparameters of each model were tuned by maximizing classification accuracy. To mitigate overfitting, model parameters were further optimized through 10-fold cross-validation.\u0026nbsp;The ACC and AUC of the model before and after optimization are shown in Figure 1C. The threshold for maximizing the Youden index via ROC curve analysis is: RF (0.5), ET (0.51), K-NN (0.58), GB (0.53)\u003csup\u003e29\u003c/sup\u003e. The performance are presented in Figure 2 and Supplementary Materials (Figs. S1-S3, Table S3).\u003c/p\u003e\n\u003cp\u003eThe WPV method was employed for model integration to achieve optimal performance across all evaluation metrics. The RF, GNB, and k-NN models were integrated through this WPV approach. The ensemble model performed best when the weights of the three models were set to a 1:1:1 ratio. The model performance of the WPV is shown in Figure 3.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFeatures contribution analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe contribution of each feature to the models and its impact on the extubation success rate were determined through the calculation of SHAP values. Swallowing dysfunction was consistently the most influential factor in all models, as patients with swallowing difficulties were often classified as negative (i.e., extubation failure). Lower GCS scores also played a significant role, leading the models to lean towards negative classifications. The SHAP values for each model are showed in Figure 4 and the Supplementary Materials (Figure S4).\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eIn this study, Logistic regression was initially conducted on all variables related to extubation outcomes. Subsequently, eight ML models were developed based on these variables, and their predictive performance for extubation success was evaluated using multiple metrics. The SHAP analysis revealed that among all variables, disease duration, primary disease, tetraplegia, GCS score, swallowing function, pathogens in sputum culture, interleukin-6 levels, lymphocyte count, and abnormal respiratory structures were associated with successful extubation.\u0026nbsp;Under optimal parameter selection, random forest, extra trees, k-NN, gradient boosting, and WPV\u0026nbsp;all performed well.\u003c/p\u003e\n\u003cp\u003eIn recent years, the use of ML models in medical prediction has been increasing, but there are significant differences in performance among different models in predicting extubation. For instance, LR is widely used due to its strong interpretability, but it has limitations in modeling complex nonlinear relationships. Wang\u003cem\u003e\u0026nbsp;et al\u003c/em\u003e. developed a predictive model using multivariable logistic regression, but its AUC was only 0.793, indicating limitations in handling high-dimensional data\u003csup\u003e30\u003c/sup\u003e. A study compared the performance of RF (AUC = 0.787), linear regression (AUC = 0.762), artificial neural network (AUC = 0.763), and SVM (AUC = 0.740). Nevertheless, none of the models achieved satisfactory results, as all AUC values were less than 0.8\u003csup\u003e31\u003c/sup\u003e. In contrast, Huang \u003cem\u003eet al\u003c/em\u003e. achieved an AUC of 0.976 with their RF model built using time series respiratory parameters, which was significantly better than SVM and GNB\u003csup\u003e32\u003c/sup\u003e. Even though it was lower than the previous study, the four models in our study that showed good performance all had AUC values above 0.85. Given the unique physiological parameters in high-altitude areas, we still believe that their performance is clinically significant. Next, they utilized time-series ventilator-derived parameters as features, which may be more advantageous for improving the predictive power of the model.\u003c/p\u003e\n\u003cp\u003eMixed types of feature values can impact the model's performance. In this study, the modeling features include both continuous and categorical data values. In comparison, mixed type of data is more in line with clinical reality than single type of data. Different ML algorithms have their own advantages and limitations when dealing with mixed features. DT and tree-based models like random forests have the ability to handle categorical variables without the need for extensive preprocessing, demonstrating reliable performance with high-dimensional, heterogeneous datasets\u003csup\u003e33\u003c/sup\u003e. Continuous variables and categorical variables differ in scale, distribution, and information representation. Inputting them directly into data scale-sensitive models such as SVM and K-NN may result in the model being excessively sensitive to one type of feature, while ignoring crucial information from the other type of feature. Therefore, it is often necessary to normalize or standardize continuous variables. Techniques like one-hot encoding, label encoding, and neural embeddings convert categorical variables into a unified numerical representation, effectively addressing scale inconsistency issues during model training\u003csup\u003e34,35\u003c/sup\u003e.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eReducing the risk of overfitting and enhancing the generalization of models is crucial. Due to the different ranges and distributions of various features, it is easy to cause data bias and model overfitting issues, especially when the sample size is small\u003csup\u003e36\u003c/sup\u003e. Therefore, the hyperparameter optimization and cross-validation methods is particularly crucial to ensure that the model not only performs excellently but also possesses strong generalization capabilities\u003csup\u003e37,38\u003c/sup\u003e. Additionally, the feature selection process is especially critical for mixed data. This process focuses on retaining the most informative features while eliminating redundant and noisy ones, providing a key pathway to enhance the model's learning efficiency\u003csup\u003e39\u003c/sup\u003e. Hence, The aforementioned methods were adopted to further strengthen the model's performance and learning depth.\u0026nbsp;Furthermore, mitigating model overfitting has also somewhat enhanced the generalization ability of each model\u003csup\u003e40\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003eIn present study, a WPV was integrated model developed. Bagging, boosting, and stacking are conventional methods that have been widely researched. Bagging functions by training base models on bootstrap samples, ultimately decreasing variance and improving model stability\u003csup\u003e41,42\u003c/sup\u003e. Meanwhile, bagging’s simplicity and ease of implementation make it a robust baseline method, although its uniform voting mechanism may not fully exploit the differing confidence levels of individual models\u003csup\u003e42\u003c/sup\u003e. Although boosting can achieve high accuracy on training data, its performance may degrade on unseen data when noise is present. Consequently, the benefits of boosting are often offset by decreased robustness in noisy environments\u003csup\u003e43,44\u003c/sup\u003e.\u0026nbsp;Finally, stacking entails utilizing a meta-learner to combine the outcomes of different models, thereby capturing more intricate relationships in the data\u003csup\u003e45\u003c/sup\u003e. While stacking often provides better performance in diverse settings, its reliance on an additional meta-learning layer can result in increased complexity in model selection and parameter tuning\u003csup\u003e45\u003c/sup\u003e.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eIn contrast to previously mentioned methods, emerging methods such as WPV offer a new way to incorporate model confidence levels into the final decision, addressing issues related to class imbalance and overfitting in noisy environments\u003csup\u003e26,46\u003c/sup\u003e. This mechanism is a complement to the bagging and boosting frameworks and could improve ensemble performance in multiple medical domains.\u0026nbsp;In present study, the WPV\u0026nbsp;ensemble\u0026nbsp;model showed superior performance in all evaluation dimensions compared to individual models. Given that our dataset was derived from a single center, our model may consequently have reduced generalizability. The WPV method could be utilized to partially alleviate this issue. Firstly, by combining the outputs of complementary models,\u0026nbsp;ensemble\u0026nbsp;models can better capture the underlying distribution of the data, alleviating the risks associated with individual models learning inadequately or underfitting\u003csup\u003e26\u003c/sup\u003e. Next, the voting mechanism has excellent scalability, making it easy to add or remove models integrated into the system. This is extremely beneficial for managing extensive data flows and intricate tasks, enabling the flexible adjustment of the model's scale and structure according to specific requirements\u003csup\u003e46\u003c/sup\u003e.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eFinally, an analysis using SHAP was conducted to evaluate the impact of each feature on the extubation outcomes. The SHAP values suggested that patients with swallowing dysfunction had a significantly lower extubation success rate. Previous retrospective studies also reported a significant correlation between swallowing function and the extubation success rate\u003csup\u003e5,47,48\u003c/sup\u003e. Damage to the glossopharyngeal (IX), vagus (X), and hypoglossal (XII) nerves could impact the muscles used in swallowing, leading to dysphagia in individuals with CNS injury\u003csup\u003e49\u003c/sup\u003e. Precise coordination of breathing and swallowing is crucial for airway protection\u003csup\u003e50\u003c/sup\u003e. Dysfunction of the swallowing musculature increases the risk of aspiration and pulmonary infection, which are common causes of extubation failure and prolonged intubation in critically ill patients\u003csup\u003e51,52\u003c/sup\u003e. Many studies\u0026nbsp;have\u0026nbsp;emphasized the need for clinicians to prioritize swallowing training after extubation\u003csup\u003e53,54\u003c/sup\u003e. However, the results of our study\u0026nbsp;suggest\u0026nbsp;that swallowing function training before extubation should also be emphasized, as it\u0026nbsp;might\u0026nbsp;help reduce\u0026nbsp;postextubation\u0026nbsp;dysfunction and decrease the likelihood of extubation failure. Training and assessing swallowing are crucial abilities in rehabilitation medicine; early intervention in rehabilitation could enhance the extubation success rate by enhancing swallowing function prior to extubation. Nevertheless, this process must highlight individualization, and specific strategies also require additional investigation. Swallowing difficulties are present in 40% of patients with quadriplegia\u003csup\u003e49\u003c/sup\u003e.\u0026nbsp;They might also have problems like diaphragm imbalance and respiratory issues\u003csup\u003e55,56\u003c/sup\u003e.\u0026nbsp;Given the low oxygen and low pressure environment of the plateau, patients with quadriplegia may need to focus more on their respiratory training. In severe cases, diaphragm pacemakers may be considered as interventions\u003csup\u003e57\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003eThe influence of consciousness level on extubation success rate remains controversial\u003csup\u003e58\u003c/sup\u003e.\u0026nbsp;Nevertheless, GCS were consistently ranked in the top three contributors in the models' classification.\u0026nbsp;Patients with higher GCS scores showed significantly greater extubation success rates.\u0026nbsp;This finding reinforces previous studies\u0026nbsp;which\u0026nbsp;demonstrate that consciousness level strongly correlates with extubation outcomes\u003csup\u003e59,60\u003c/sup\u003e.\u0026nbsp;Furthermore, cognitive impairment contributes to swallowing disorders, aspiration risks, and subsequent pneumonia, potentially resulting in delayed extubation for patients with tracheostomies\u003csup\u003e61\u003c/sup\u003e.\u0026nbsp;Impaired consciousness leads to prolonged bed rest, thereby increasing the risk of extubation failure\u003csup\u003e62\u003c/sup\u003e. If consciousness disorders are truly related to extubation outcomes, then high-altitude areas need to take this issue seriously. The low pressure and insufficient oxygen supply in high-altitude regions can lead to cerebral tissue hypoxia, resulting in vasodilation, disruption of the blood-brain barrier, and cerebral edema\u003csup\u003e11,63\u003c/sup\u003e. In extreme situations, these alterations may result in a patient quickly transitioning from a state of blurred consciousness to a coma. Therefore, the consciousness of patients with CNS in high-altitude areas should be further emphasized.\u003c/p\u003e\n\u003cp\u003eThe results showed that patients with higher ALB levels were more likely to be classified as extubation failures in this both models. Following CNS damage, various forms of cell death, including apoptosis, ferroptosis, and mitochondrial dysfunction, can deplete a substantial amount of protein resources\u003csup\u003e64,65\u003c/sup\u003e. Previous research\u0026nbsp;revealed\u0026nbsp;that patients with a daily protein intake greater than 1.2 g/kg/day effectively\u0026nbsp;presented\u0026nbsp;increased serum albumin levels, and their median duration of mechanical ventilation was significantly shorter\u0026nbsp;than that of\u0026nbsp;those with lower protein intake\u003csup\u003e66\u003c/sup\u003e. Their conclusions emphasized that higher protein intake is beneficial for the successful extubation of patients with prolonged endotracheal intubation\u003csup\u003e66,67\u003c/sup\u003e. These findings suggest that increasing serum albumin levels through enhanced nutritional support or albumin supplementation prior to extubation may be an effective strategy to improve extubation success rates.\u003c/p\u003e\n\u003cp\u003eIn our study, lymphocyte and IL-6 levels influenced extubation outcomes. The model is more likely to classify samples with lymphocytopenia and elevated IL-6 levels as having a higher risk of extubation failure. A pathway analysis study highlighted IL-6 as a key biomarker linked to inflammatory processes, with elevated levels serving as a reliable indicator of active inflammation in patients\u003csup\u003e68\u003c/sup\u003e. Moreover, another study has\u0026nbsp;suggested that lymphopenia could indicate a state of chronic inflammation and prolonged immune depression in patients\u003csup\u003e69\u003c/sup\u003e. Although the exact mechanism by which lymphocytopenia impacts extubation outcomes remains unclear, patients with lymphocytopenia and immunosuppression exhibit significantly reduced extubation success rates and elevated in-hospital mortality\u003csup\u003e70,71\u003c/sup\u003e. Patients with lymphopenia were also found to present with thrombocytopenia and hypoalbuminemia, suggesting a state of systemic compromise\u003csup\u003e72\u003c/sup\u003e. This finding suggested that preextubation anti-inflammatory treatment combined with protein intake\u0026nbsp;might\u0026nbsp;be a potential strategy to improve extubation success rates.\u003c/p\u003e\n\u003cp\u003eThis study also has several limitations.\u0026nbsp;Firstly, our models demonstrated superior performance compared to previous studies, but they are not yet suitable for clinical use. More clinical research is necessary in this area.\u0026nbsp;Second, the features included in this study may have been limited, and the model's reliability could be enhanced by incorporating a more comprehensive set of features during the modeling process\u003csup\u003e73\u003c/sup\u003e. At the same time, the data collected in this study lack a time dimension, so the results lack explanatory dynamics of the disease course in tracheostomy patients. Moreover, this study did not find any distinctive features specifically associated with high-altitude pulmonary edema outcomes. In future studies, additional indicators such as alveolar oxygen partial pressure (or blood oxygen saturation), plateau pressure, blood urea nitrogen, heart rate, positive end-expiratory pressure, and creatinine will\u0026nbsp;be included\u003csup\u003e74-76\u003c/sup\u003e. Finally, data collection from a single center may reduce the generalization ability of the model. Therefore, in future research, multicenter data collection should be adopted, as it can make the research results more convincing and increase generalizability.\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eIn summary, the models developed in this study, including the random forest, extra tree classifier, k-nearest neighbor, gradient boosting, and WPV ensemble models, demonstrated strong performance in terms of the AUC and accuracy, indicating good predictive ability for outcome prediction at weaning. According to SHAP analysis, dysphagia, quadriplegia, altered consciousness, low white blood cell levels, and abnormal levels of interleukins and lymphocytes may have a negative impact on extubation outcomes. However, the data was collected from a single center, which may lead to insufficient generalizability. In the future, it is necessary to consider supplementing with data from multiple centers. Furthermore, future studies should incorporate a wider range of clinical indicators, particularly those associated with the respiratory system.\u003c/p\u003e\n\u003cp\u003e\u003cbr\u003e\u003c/p\u003e"},{"header":"Declarations","content":"\u003ch2\u003eData availability\u003c/h2\u003e\n\u003cp\u003eThe datasets used during the current study are available from the corresponding author upon reasonable request.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor contributions\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eZ.H.C., X.Y. and L.Q.Y. conceived of and designed the study. Z.H.C., X.Y., X.C.L., and H.T.W. performed analysis. Z.H.C., X.Y., H.T.W. and L.Q.Y. wrote the manuscript and Z.H.C., X.Y. and L.Q.Y. critically reviewed the manuscript. Z.H.C., X.Y., and Y.Z. generated the figures and tables. Q.L. and H.M.Z. were responsible for the management of public data. All authors have read and approved the manuscript.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare no competing interests.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAdditional information\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCorrespondence\u003c/strong\u003e and requests for materials should be addressed to L.-Q.Y.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eReprints and permissions information\u003c/strong\u003e is available at www.nature.com/reprints.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eOpen Access\u003c/strong\u003e This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. Te images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eKumar, S., Patra, A., Deepthi, S. \u0026amp; Biradar, K. Tracheostomy Complications Over 5 Years: Decannulation Analysis. \u003cem\u003eIran. J. Otorhinolaryngol.\u003c/em\u003e \u003cb\u003e36\u003c/b\u003e, 559\u0026ndash;565. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.22038/ijorl.2024.75652.3533\u003c/span\u003e\u003cspan address=\"10.22038/ijorl.2024.75652.3533\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGallice, T., Cugy, E., Branchard, O., Dehail, P. \u0026amp; Moucheboeuf, G. Predictive Factors for Successful Decannulation in Patients with Tracheostomies and Brain Injuries: A Systematic Review. \u003cem\u003eDysphagia\u003c/em\u003e \u003cb\u003e39\u003c/b\u003e, 552\u0026ndash;572. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1007/s00455-023-10646-2\u003c/span\u003e\u003cspan address=\"10.1007/s00455-023-10646-2\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSuntrup-Krueger, S. et al. Extubation Readiness in Critically Ill Stroke Patients. \u003cem\u003eStroke\u003c/em\u003e 50, 1981\u0026ndash;1988, (2019). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1161/strokeaha.118.024643\u003c/span\u003e\u003cspan address=\"10.1161/strokeaha.118.024643\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGuzatti, N. G. et al. Predictive Factors of Extubation Failure in COVID-19 Mechanically Ventilated Patients. \u003cem\u003eJ. Intensive Care Med.\u003c/em\u003e \u003cb\u003e37\u003c/b\u003e, 1250\u0026ndash;1255. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1177/08850666221093946\u003c/span\u003e\u003cspan address=\"10.1177/08850666221093946\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGhiani, A. et al. Incidence, causes, and predictors of unsuccessful decannulation following prolonged weaning. \u003cem\u003eTherapeutic Adv. chronic disease\u003c/em\u003e. \u003cb\u003e13\u003c/b\u003e, 20406223221109655. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1177/20406223221109655\u003c/span\u003e\u003cspan address=\"10.1177/20406223221109655\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen, Y. et al. Predictors of long-term decannulation in patients with disorders of consciousness. \u003cem\u003eFront. Neurol.\u003c/em\u003e \u003cb\u003e14\u003c/b\u003e \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3389/fneur.2023.1099307\u003c/span\u003e\u003cspan address=\"10.3389/fneur.2023.1099307\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eThille, A. W., Richard, J. C. \u0026amp; Brochard, L. The decision to extubate in the intensive care unit. \u003cem\u003eAm. J. Respir. Crit Care Med.\u003c/em\u003e \u003cb\u003e187\u003c/b\u003e, 1294\u0026ndash;1302. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1164/rccm.201208-1523CI\u003c/span\u003e\u003cspan address=\"10.1164/rccm.201208-1523CI\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2013).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLiu, X. et al. Comparison of Muscle Density in Middle-Aged and Older Chinese Adults Between a High-Altitude Area (Kunming) and a Low-Altitude Area (Beijing). \u003cb\u003e12\u003c/b\u003e, (2021). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3389/fendo.2021.811770\u003c/span\u003e\u003cspan address=\"10.3389/fendo.2021.811770\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRichalet, J. P., Hermand, E. \u0026amp; Lhuissier, F. J. Cardiovascular physiology and pathophysiology at high altitude. \u003cem\u003eNat. Rev. Cardiol.\u003c/em\u003e \u003cb\u003e21\u003c/b\u003e, 75\u0026ndash;88. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41569-023-00924-9\u003c/span\u003e\u003cspan address=\"10.1038/s41569-023-00924-9\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eThille, A. W., Cort\u0026eacute;s-Puch, I. \u0026amp; Esteban, A. Weaning from the ventilator and extubation in ICU. \u003cem\u003eCurr. Opin. Crit. Care\u003c/em\u003e. \u003cb\u003e19\u003c/b\u003e, 57\u0026ndash;64. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1097/MCC.0b013e32835c5095\u003c/span\u003e\u003cspan address=\"10.1097/MCC.0b013e32835c5095\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2013).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWei, L. et al. Complement C3 participates in the function and mechanism of traumatic brain injury at simulated high altitude. \u003cem\u003eBrain Res.\u003c/em\u003e \u003cb\u003e1726\u003c/b\u003e, 146423. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.brainres.2019.146423\u003c/span\u003e\u003cspan address=\"10.1016/j.brainres.2019.146423\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMylne, D., Briggs, N. \u0026amp; Tol\u0026eacute;, G. A formalised process for decannulation reduced number of days with tracheostomy for people with an acquired brain injury: a systematic review and meta-analysis %J Brain Impairment. 25, -, doi: (2024). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1071/IB23077\u003c/span\u003e\u003cspan address=\"10.1071/IB23077\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhou, T. et al. Tracheostomy decannulation protocol in patients with prolonged tracheostomy referred to a rehabilitation hospital: a prospective cohort study. \u003cem\u003eJ. Intensive Care\u003c/em\u003e. \u003cb\u003e10\u003c/b\u003e, 34. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s40560-022-00626-3\u003c/span\u003e\u003cspan address=\"10.1186/s40560-022-00626-3\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKutsukutsa, J., Kuupiel, D., Monori-Kiss, A. \u0026amp; del Rey-Puech, P. \u0026amp; Mashamba-Thompson, T. P. Tracheostomy decannulation methods and procedures for assessing readiness for decannulation in adults: a systematic scoping review. \u003cem\u003eJBI Evid. Implement.\u003c/em\u003e \u003cb\u003e17\u003c/b\u003e (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEskildsen, S. J., Irene, W., Ingrid, P., Aviaja, H. C., Curtis, D. J. \u0026amp; and Rehabilitative intervention for successful decannulation in adult patients with acquired brain injury and tracheostomy: a systematic review. \u003cem\u003eDisabil. Rehabil.\u003c/em\u003e \u003cb\u003e46\u003c/b\u003e, 2464\u0026ndash;2476. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1080/09638288.2023.2233437\u003c/span\u003e\u003cspan address=\"10.1080/09638288.2023.2233437\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNinan, A., Grubb, L. M., Brenner, M. J. \u0026amp; Pandian, V. Effectiveness of interprofessional tracheostomy teams: A systematic review. \u003cem\u003eJ. Clin. Nurs.\u003c/em\u003e \u003cb\u003e32\u003c/b\u003e, 6967\u0026ndash;6986. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1111/jocn.16815\u003c/span\u003e\u003cspan address=\"10.1111/jocn.16815\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePark, J. E. et al. Development of a Machine Learning Model for Predicting Weaning Outcomes Based Solely on Continuous Ventilator Parameters during Spontaneous Breathing Trials. \u003cb\u003e10\u003c/b\u003e, 1163 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePai, K. C., Su, S. A., Chan, M. C., Wu, C. L. \u0026amp; Chao, W. C. Explainable machine learning approach to predict extubation in critically ill ventilated patients: a retrospective study in central Taiwan. \u003cem\u003eBMC Anesthesiol.\u003c/em\u003e \u003cb\u003e22\u003c/b\u003e, 351. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s12871-022-01888-y\u003c/span\u003e\u003cspan address=\"10.1186/s12871-022-01888-y\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCuschieri, S. The STROBE guidelines. \u003cem\u003eSaudi J. Anaesth.\u003c/em\u003e \u003cb\u003e13\u003c/b\u003e, S31\u0026ndash;s34. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.4103/sja.SJA_543_18\u003c/span\u003e\u003cspan address=\"10.4103/sja.SJA_543_18\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMannini, A. et al. Data-driven prediction of decannulation probability and timing in patients with severe acquired brain injury. \u003cem\u003eComput. Methods Programs Biomed.\u003c/em\u003e \u003cb\u003e209\u003c/b\u003e, 106345. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.cmpb.2021.106345\u003c/span\u003e\u003cspan address=\"10.1016/j.cmpb.2021.106345\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHao, P. Y., Chiang, J. H. \u0026amp; Chen, Y. D. Possibilistic classification by support vector networks. \u003cem\u003eNeural networks: official J. Int. Neural Netw. Soc.\u003c/em\u003e \u003cb\u003e149\u003c/b\u003e, 40\u0026ndash;56. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.neunet.2022.02.007\u003c/span\u003e\u003cspan address=\"10.1016/j.neunet.2022.02.007\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBecker, T., Rousseau, A. J., Geubbelmans, M., Burzykowski, T. \u0026amp; Valkenborg, D. Decision trees and random forests. \u003cem\u003eAmerican journal of orthodontics and dentofacial orthopedics: official publication of the American Association of Orthodontists, its constituent societies, and the American Board of Orthodontics\u003c/em\u003e 164, 894\u0026ndash;897, (2023). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.ajodo.2023.09.011\u003c/span\u003e\u003cspan address=\"10.1016/j.ajodo.2023.09.011\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhao, Q. Y. et al. Development and Validation of a Machine-Learning Model for Prediction of Extubation Failure in Intensive Care Units. \u003cem\u003eFront. Med.\u003c/em\u003e \u003cb\u003e8\u003c/b\u003e, 676343. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3389/fmed.2021.676343\u003c/span\u003e\u003cspan address=\"10.3389/fmed.2021.676343\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAbraham, A. et al. Machine learning for neuroimaging with scikit-learn. \u003cem\u003eFront. neuroinformatics\u003c/em\u003e. \u003cb\u003e8\u003c/b\u003e, 14. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3389/fninf.2014.00014\u003c/span\u003e\u003cspan address=\"10.3389/fninf.2014.00014\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2014).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLundberg, S. M. et al. From Local Explanations to Global Understanding with Explainable AI for Trees. \u003cem\u003eNat. Mach. Intell.\u003c/em\u003e \u003cb\u003e2\u003c/b\u003e, 56\u0026ndash;67. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s42256-019-0138-9\u003c/span\u003e\u003cspan address=\"10.1038/s42256-019-0138-9\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChalasani, R. D. \u0026amp; Radhika, Y. Prediction of ITK inhibitor kinases activity based on posterior probabilistic weighted average based ensemble voting classification. \u003cem\u003eJ. Intell. Fuzzy Syst.\u003c/em\u003e \u003cb\u003e44\u003c/b\u003e, 5837\u0026ndash;5846. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3233/JIFS-221412\u003c/span\u003e\u003cspan address=\"10.3233/JIFS-221412\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMutanga, R. T., Naicker, N. \u0026amp; Olugbara, O. O. \u003cem\u003eJ. I. J. o. A. C. S. \u0026amp; Applications\u003c/em\u003e (Detecting Hate Speech on Twitter Network using Ensemble Machine Learning, 2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKumar, R. et al. Ensemble learning-based early detection of influenza disease. \u003cem\u003eMultimedia tools Appl.\u003c/em\u003e 1\u0026ndash;21. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1007/s11042-023-15848-2\u003c/span\u003e\u003cspan address=\"10.1007/s11042-023-15848-2\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSchisterman, E. F., Perkins, N. J., Liu, A. \u0026amp; Bondell, H. Optimal Cut-point and Its Corresponding Youden Index to Discriminate Individuals Using Pooled Blood Samples. \u003cb\u003e16\u003c/b\u003e, 73\u0026ndash;81, (2005). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1097/01.ede.0000147512.81966.ba\u003c/span\u003e\u003cspan address=\"10.1097/01.ede.0000147512.81966.ba\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang, X. et al. Development and validation of a nomogram for decannulation in patients with neurological injury: A prognostic accuracy study. \u003cem\u003eFront. Neurol.\u003c/em\u003e \u003cb\u003e13\u003c/b\u003e \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3389/fneur.2022.979160\u003c/span\u003e\u003cspan address=\"10.3389/fneur.2022.979160\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHur, S. et al. Development and Validation of Unplanned Extubation Prediction Models Using Intensive Care Unit Data: Retrospective, Comparative, Machine Learning Study. \u003cem\u003eJ. Med. Internet. Res.\u003c/em\u003e \u003cb\u003e23\u003c/b\u003e, e23508. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.2196/23508\u003c/span\u003e\u003cspan address=\"10.2196/23508\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHuang, K. Y. et al. Developing a machine-learning model for real-time prediction of successful extubation in mechanically ventilated patients using time-series ventilator-derived parameters. \u003cem\u003eFront. Med.\u003c/em\u003e \u003cb\u003e10\u003c/b\u003e, 1167445. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3389/fmed.2023.1167445\u003c/span\u003e\u003cspan address=\"10.3389/fmed.2023.1167445\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRodriguez-Galiano, V., Sanchez-Castillo, M., Chica-Olmo, M. \u0026amp; Chica-Rivas, M. Machine learning predictive models for mineral prospectivity: An evaluation of neural networks, random forest, regression trees and support vector machines. \u003cem\u003eOre Geol. Rev.\u003c/em\u003e \u003cb\u003e71\u003c/b\u003e, 804\u0026ndash;818. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.oregeorev.2015.01.001\u003c/span\u003e\u003cspan address=\"10.1016/j.oregeorev.2015.01.001\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2015).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLuo, W. et al. Guidelines for Developing and Reporting Machine Learning Predictive Models in Biomedical Research: A Multidisciplinary View. \u003cem\u003eJ. Med. Internet. Res.\u003c/em\u003e \u003cb\u003e18\u003c/b\u003e, e323. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.2196/jmir.5870\u003c/span\u003e\u003cspan address=\"10.2196/jmir.5870\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2016).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eParish, E. J. \u0026amp; Duraisamy, K. A paradigm for data-driven predictive modeling using field inversion and machine learning. \u003cem\u003eJ. Comput. Phys.\u003c/em\u003e \u003cb\u003e305\u003c/b\u003e, 758\u0026ndash;774. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.jcp.2015.11.012\u003c/span\u003e\u003cspan address=\"10.1016/j.jcp.2015.11.012\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2016).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBattineni, G., Sagaro, G. G., Chinatalapudi, N. \u0026amp; Amenta, F. \u003cem\u003eAppl. Mach. Learn. Predictive Models Chronic Disease Diagnosis\u003c/em\u003e \u003cb\u003e10\u003c/b\u003e, 21 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAzodi, C. B. et al. Benchmarking Parametric and Machine Learning Models for Genomic Prediction of Complex Traits. \u003cem\u003eG3 (Bethesda Md)\u003c/em\u003e. \u003cb\u003e9\u003c/b\u003e, 3691\u0026ndash;3702. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1534/g3.119.400498\u003c/span\u003e\u003cspan address=\"10.1534/g3.119.400498\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eXiao, Q., Chang, H. H., Geng, G. \u0026amp; Liu, Y. An Ensemble Machine-Learning Model To Predict Historical PM(2.5) Concentrations in China from Satellite Data. \u003cem\u003eEnviron. Sci. Technol.\u003c/em\u003e \u003cb\u003e52\u003c/b\u003e, 13260\u0026ndash;13269. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1021/acs.est.8b02917\u003c/span\u003e\u003cspan address=\"10.1021/acs.est.8b02917\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2018).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLuo, M. et al. Combination of Feature Selection and CatBoost for Prediction: The First Application to the Estimation of Aboveground Biomass. \u003cb\u003e12\u003c/b\u003e, 216 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGhasemzadeh, H., Hillman, R. E. \u0026amp; Mehta, D. D. Toward Generalizable Machine Learning Models in Speech, Language, and Hearing Sciences: Estimating Sample Size and Reducing Overfitting. 67, 753\u0026ndash;781, doi: (2024). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1044/2023_JSLHR-23-00273\u003c/span\u003e\u003cspan address=\"10.1044/2023_JSLHR-23-00273\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGneiting, T. \u0026amp; Raftery, A. E. Weather Forecasting with Ensemble Methods. \u003cb\u003e310\u003c/b\u003e, 248\u0026ndash;249, doi: (2005). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1126/science.1115255\u003c/span\u003e\u003cspan address=\"10.1126/science.1115255\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMienye, I. D., Sun, Y. A. \u0026amp; Survey of Ensemble Learning: Concepts, Algorithms, Applications, and Prospects. \u003cem\u003eIEEE Access.\u003c/em\u003e \u003cb\u003e10\u003c/b\u003e, 99129\u0026ndash;99149. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1109/ACCESS.2022.3207287\u003c/span\u003e\u003cspan address=\"10.1109/ACCESS.2022.3207287\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMatloob, F. et al. Software Defect Prediction Using Ensemble Learning: A Systematic Literature Review. \u003cem\u003eIEEE Access.\u003c/em\u003e \u003cb\u003e9\u003c/b\u003e, 98754\u0026ndash;98771. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1109/ACCESS.2021.3095559\u003c/span\u003e\u003cspan address=\"10.1109/ACCESS.2021.3095559\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYuan, X. et al. Applied Machine Learning for Prediction of CO2 Adsorption on Biomass Waste-Derived Porous Carbons. \u003cem\u003eEnviron. Sci. Technol.\u003c/em\u003e \u003cb\u003e55\u003c/b\u003e, 11925\u0026ndash;11936. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1021/acs.est.1c01849\u003c/span\u003e\u003cspan address=\"10.1021/acs.est.1c01849\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChatzimparmpas, A., Martins, R. M., Kucher, K., Kerren, A. \u0026amp; StackGenVis Alignment of Data, Algorithms, and Models for Stacking Ensemble Learning Using Performance Metrics. \u003cem\u003eIEEE Trans. Vis. Comput. Graph.\u003c/em\u003e \u003cb\u003e27\u003c/b\u003e, 1547\u0026ndash;1557. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1109/TVCG.2020.3030352\u003c/span\u003e\u003cspan address=\"10.1109/TVCG.2020.3030352\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGomes, H. M., Barddal, J. P., Enembreck, F. \u0026amp; Bifet, A. A Survey on Ensemble Learning for Data Stream Classification. \u003cb\u003e50\u003c/b\u003e, Article 23, (2017). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1145/3054925\u003c/span\u003e\u003cspan address=\"10.1145/3054925\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eXu, S. S. et al. Development of a Prediction Score for Evaluation of Extubation Readiness in Neurosurgical Patients with Mechanical Ventilation. \u003cem\u003eAnesthesiology\u003c/em\u003e \u003cb\u003e139\u003c/b\u003e, 614\u0026ndash;627. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1097/aln.0000000000004721\u003c/span\u003e\u003cspan address=\"10.1097/aln.0000000000004721\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePark, M. K. \u0026amp; Lee, S. J. Changes in Swallowing and Cough Functions Among Stroke Patients Before and After Tracheostomy Decannulation. \u003cem\u003eDysphagia\u003c/em\u003e \u003cb\u003e33\u003c/b\u003e, 857\u0026ndash;865. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1007/s00455-018-9920-9\u003c/span\u003e\u003cspan address=\"10.1007/s00455-018-9920-9\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2018).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eShem, K., Castillo, K., Wong, S. L., Chang, J. \u0026amp; Kolakowsky-Hayner, S. Dysphagia and respiratory care in individuals with tetraplegia: incidence, associated factors, and preventable complications. \u003cem\u003eTop. spinal cord injury rehabilitation\u003c/em\u003e. \u003cb\u003e18\u003c/b\u003e, 15\u0026ndash;22. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1310/sci1801-15\u003c/span\u003e\u003cspan address=\"10.1310/sci1801-15\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2012).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBrodsky, M. B. et al. Respiratory-swallow phase patterns and their relationship to swallowing impairment in patients treated for oropharyngeal cancer. \u003cem\u003eHead neck\u003c/em\u003e. \u003cb\u003e32\u003c/b\u003e, 481\u0026ndash;489. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1002/hed.21209\u003c/span\u003e\u003cspan address=\"10.1002/hed.21209\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2010).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOsorio, B., Krakora, R., Epping, G., Wright, D. \u0026amp; Rajwani, K. Aspiration after critical illness: role of endotracheal tube, tracheostomy, swallowing disorders. \u003cem\u003eSemin. Respir. Crit. Care Med.\u003c/em\u003e \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1055/a-2445-9054\u003c/span\u003e\u003cspan address=\"10.1055/a-2445-9054\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLabeit, B. et al. The assessment of dysphagia after stroke: state of the art and future directions. \u003cem\u003eLancet Neurol.\u003c/em\u003e \u003cb\u003e22\u003c/b\u003e, 858\u0026ndash;870. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/s1474-4422(23)00153-9\u003c/span\u003e\u003cspan address=\"10.1016/s1474-4422(23)00153-9\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen, L. et al. Interventions for Postextubation Dysphagia in Critically Ill Patients: A Systematic Review and Meta-analysis. \u003cem\u003eDysphagia\u003c/em\u003e \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1007/s00455-024-10695-1\u003c/span\u003e\u003cspan address=\"10.1007/s00455-024-10695-1\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTroll, C. et al. A bedside swallowing screen for the identification of post-extubation dysphagia on the intensive care unit - validation of the Gugging Swallowing Screen (GUSS)-ICU. \u003cem\u003eBMC Anesthesiol.\u003c/em\u003e \u003cb\u003e23\u003c/b\u003e \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s12871-023-02072-6\u003c/span\u003e\u003cspan address=\"10.1186/s12871-023-02072-6\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCatal\u0026aacute;-Ripoll, J. V., Monsalve-Naharro, J. \u0026amp; Hern\u0026aacute;ndez-Fern\u0026aacute;ndez, F. Incidence and predictive factors of diaphragmatic dysfunction in acute stroke. \u003cem\u003eBMC Neurol.\u003c/em\u003e \u003cb\u003e20\u003c/b\u003e \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s12883-020-01664-w\u003c/span\u003e\u003cspan address=\"10.1186/s12883-020-01664-w\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOo, T., Watt, J. W., Soni, B. M. \u0026amp; Sett, P. K. Delayed diaphragm recovery in 12 patients after high cervical spinal cord injury. A retrospective review of the diaphragm status of 107 patients ventilated after acute spinal cord injury. \u003cem\u003eSpinal cord\u003c/em\u003e. \u003cb\u003e37\u003c/b\u003e, 117\u0026ndash;122. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/sj.sc.3100775\u003c/span\u003e\u003cspan address=\"10.1038/sj.sc.3100775\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (1999).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKaufman, M. R. et al. Prospective analysis of a surgical algorithm to achieve ventilator weaning in cervical tetraplegia. \u003cem\u003eJ. Spinal Cord Med.\u003c/em\u003e \u003cb\u003e45\u003c/b\u003e, 531\u0026ndash;535. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1080/10790268.2020.1829417\u003c/span\u003e\u003cspan address=\"10.1080/10790268.2020.1829417\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChan, L. Y., Jones, A. Y., Chung, R. C. \u0026amp; Hung, K. N. Peak flow rate during induced cough: a predictor of successful decannulation of a tracheotomy tube in neurosurgical patients. \u003cem\u003eAm. J. Crit. care: official publication Am. Association Critical-Care Nurses\u003c/em\u003e. \u003cb\u003e19\u003c/b\u003e, 278\u0026ndash;284. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.4037/ajcc2009575\u003c/span\u003e\u003cspan address=\"10.4037/ajcc2009575\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2010).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTorrini, F. et al. Prediction of extubation outcome in critically ill patients: a systematic review and meta-analysis. \u003cem\u003eCrit. Care. (London, England)\u003c/em\u003e. \u003cb\u003e25\u003c/b\u003e, 391. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s13054-021-03802-3\u003c/span\u003e\u003cspan address=\"10.1186/s13054-021-03802-3\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHuang, H. Y. et al. Clinical outcomes and prognostic factors for prolonged mechanical ventilation in patients with acute stroke and brain trauma. \u003cem\u003eJ. Formos. Med. Association = Taiwan. yi zhi\u003c/em\u003e. \u003cb\u003e121\u003c/b\u003e, 162\u0026ndash;169. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.jfma.2021.02.011\u003c/span\u003e\u003cspan address=\"10.1016/j.jfma.2021.02.011\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZanata Ide, L., Santos, R. S. \u0026amp; Hirata, G. C. Tracheal decannulation protocol in patients affected by traumatic brain injury. \u003cem\u003eInt. archives Otorhinolaryngol.\u003c/em\u003e \u003cb\u003e18\u003c/b\u003e, 108\u0026ndash;114. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1055/s-0033-1363467\u003c/span\u003e\u003cspan address=\"10.1055/s-0033-1363467\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2014).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMitchell, R. B. et al. Clinical consensus statement: tracheostomy care. \u003cem\u003eOtolaryngology\u0026ndash;head neck surgery: official J. Am. Acad. Otolaryngology-Head Neck Surg.\u003c/em\u003e \u003cb\u003e148\u003c/b\u003e, 6\u0026ndash;20. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1177/0194599812460376\u003c/span\u003e\u003cspan address=\"10.1177/0194599812460376\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2013).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTang, Q. et al. Identification of blood exosomal metabolomic profiling for high-altitude cerebral edema. \u003cem\u003eSci. Rep.\u003c/em\u003e \u003cb\u003e14\u003c/b\u003e, 11585. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41598-024-62360-0\u003c/span\u003e\u003cspan address=\"10.1038/s41598-024-62360-0\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang, X. et al. CX3CL1/CX3CR1 signal mediates M1-type microglia and accelerates high-altitude-induced forgetting. \u003cem\u003eFront. Cell. Neurosci.\u003c/em\u003e \u003cb\u003e17\u003c/b\u003e \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3389/fncel.2023.1189348\u003c/span\u003e\u003cspan address=\"10.3389/fncel.2023.1189348\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang, L., Hu, Z., Li, Z. \u0026amp; Lin, Y. Crosstalk among mitophagy, pyroptosis, ferroptosis, and necroptosis in central nervous system injuries. \u003cem\u003eNeural regeneration Res.\u003c/em\u003e \u003cb\u003e19\u003c/b\u003e, 1660\u0026ndash;1670. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.4103/1673-5374.389361\u003c/span\u003e\u003cspan address=\"10.4103/1673-5374.389361\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHuang, S. W. et al. The Impact of Higher Protein Intake in Patients with Prolonged Mechanical Ventilation. \u003cem\u003eNutrients\u003c/em\u003e 14, (2022). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3390/nu14204395\u003c/span\u003e\u003cspan address=\"10.3390/nu14204395\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChang, C. H. et al. Higher protein intake may benefit in patients with prolonged mechanical ventilation. \u003cem\u003eFront. Nutr.\u003c/em\u003e \u003cb\u003e11\u003c/b\u003e, 1449240. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3389/fnut.2024.1449240\u003c/span\u003e\u003cspan address=\"10.3389/fnut.2024.1449240\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMisra, S. et al. Common Pathways of Epileptogenesis in Patients With Epilepsy Post-Brain Injury: Findings From a Systematic Review and Meta-analysis. \u003cem\u003eNeurology\u003c/em\u003e \u003cb\u003e101\u003c/b\u003e, e2243\u0026ndash;e2256. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1212/wnl.0000000000207749\u003c/span\u003e\u003cspan address=\"10.1212/wnl.0000000000207749\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAdrie, C. et al. Persistent lymphopenia is a risk factor for ICU-acquired infections and for death in ICU patients with sustained hypotension at admission. \u003cem\u003eAnn. Intensiv. Care.\u003c/em\u003e \u003cb\u003e7\u003c/b\u003e \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s13613-017-0242-0\u003c/span\u003e\u003cspan address=\"10.1186/s13613-017-0242-0\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLiao, T. Y., Chen, Y. L., Chen, Y. L., Kuo, Y. W. \u0026amp; Jerng, J. S. Persistent inflammation and lymphopenia and weaning outcomes of patients with prolonged mechanical ventilation. \u003cem\u003eRespiratory Invest.\u003c/em\u003e \u003cb\u003e62\u003c/b\u003e, 935\u0026ndash;941. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.resinv.2024.08.001\u003c/span\u003e\u003cspan address=\"10.1016/j.resinv.2024.08.001\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSchneider, B., Oliveira, R. A., Friedman, G. \u0026amp; Moraes, R. B. Association of biomarkers with successful ventilatory weaning in COVID-19 patients: an observational study. \u003cem\u003eCritical care science\u003c/em\u003e 36, e20240158en, (2024). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.62675/2965-2774.20240158-en\u003c/span\u003e\u003cspan address=\"10.62675/2965-2774.20240158-en\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSoeters, P. B., Wolfe, R. R., Shenkin, A. \u0026amp; Hypoalbuminemia Pathogenesis and Clinical Significance. \u003cem\u003eJPEN J. Parenter. Enteral. Nutr.\u003c/em\u003e \u003cb\u003e43\u003c/b\u003e, 181\u0026ndash;193. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1002/jpen.1451\u003c/span\u003e\u003cspan address=\"10.1002/jpen.1451\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eIgarashi, Y. et al. Machine learning for predicting successful extubation in patients receiving mechanical ventilation. \u003cem\u003eFront. Med.\u003c/em\u003e \u003cb\u003e9\u003c/b\u003e, 961252. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3389/fmed.2022.961252\u003c/span\u003e\u003cspan address=\"10.3389/fmed.2022.961252\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZeng, Z., Tang, X., Liu, Y., He, Z. \u0026amp; Gong, X. Interpretable recurrent neural network models for dynamic prediction of the extubation failure risk in patients with invasive mechanical ventilation in the intensive care unit. \u003cem\u003eBioData Min.\u003c/em\u003e \u003cb\u003e15\u003c/b\u003e, 21. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s13040-022-00309-7\u003c/span\u003e\u003cspan address=\"10.1186/s13040-022-00309-7\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKuo, H. J. et al. Improvement in the Prediction of Ventilator Weaning Outcomes by an Artificial Neural Network in a Medical ICU. \u003cem\u003eRespir. Care\u003c/em\u003e. \u003cb\u003e60\u003c/b\u003e, 1560\u0026ndash;1569. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.4187/respcare.03648\u003c/span\u003e\u003cspan address=\"10.4187/respcare.03648\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2015).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFabregat, A. et al. A Machine Learning decision-making tool for extubation in Intensive Care Unit patients. \u003cem\u003eComput. Methods Programs Biomed.\u003c/em\u003e \u003cb\u003e200\u003c/b\u003e \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.cmpb.2020.105869\u003c/span\u003e\u003cspan address=\"10.1016/j.cmpb.2020.105869\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2021).\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Airway Extubation, Central Nervous System Diseases, ML, Predict","lastPublishedDoi":"10.21203/rs.3.rs-5549340/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-5549340/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eNo predictive models have been reported for tracheostomy extubation success in plateau region rehabilitation departments. Hence, the primary objective of this retrospective study was to evaluate the predictive capabilities of different models for extubation outcomes in CNS injury patients in plateau rehabilitation departments, as well as investigate the influence of clinical features on these outcomes. Data were collected from 501 adult tracheostomy patients in the Department of Rehabilitation Medicine, including 196 successful extubations. Logistic regression was employed to identify the significant features linked to extubation outcomes from a pool of 31 variables. A total of eight independent models and a weighted posterior voting ensemble model were developed. Hyperparameter optimization and 10-fold cross-validation were used to assist in choosing model parameters. Random forest (ACC\u0026thinsp;=\u0026thinsp;84.15, AUC\u0026thinsp;=\u0026thinsp;0.85), extra trees (83.17%, 0.87), K-NN (82.18%, 0.85), and gradient boosting (81.19%, 0.85) performed well. An ensemble model (85.15%, 0.87) combining random forest, Gaussian naive Bayes, and K-NN via the WPV method was developed. Dysphagia and low GCS scores have been linked to increased difficulty in extubation, as indicated by SHAP values and previous studies. Moreover, there could be a relationship between chronic inflammation and albumin levels in patients, which may collectively impact extubation success. This study evaluated the effectiveness of conventional models for predicting extubation outcomes and analyzed the factors influencing extubation results at high altitudes, laying the groundwork for clinical use and future research. Nevertheless, further research will see advantages in using multicentric approaches and broadening clinical indicators.\u003c/p\u003e","manuscriptTitle":"Using ML techniques to predict extubation outcomes for patients with central nervous system injuries in the Yun-Gui Plateau","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-04-10 05:27:26","doi":"10.21203/rs.3.rs-5549340/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Accepted","date":"2025-04-15T07:31:33+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-04-10T12:50:08+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"112934400972140796498900222184680555954","date":"2025-04-10T12:49:19+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-04-08T07:30:00+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-04-07T10:58:00+00:00","index":"","fulltext":""},{"type":"submitted","content":"Scientific Reports","date":"2025-03-26T12:15:30+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"f4c8525a-26aa-4a7a-9941-ff0f5f0807b3","owner":[],"postedDate":"April 10th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[{"id":46833512,"name":"Health sciences/Diseases/Neurological disorders/Brain injuries"},{"id":46833513,"name":"Health sciences/Diseases/Neurological disorders/Spinal cord diseases"},{"id":46833514,"name":"Health sciences/Diseases/Neurological disorders/Stroke"},{"id":46833515,"name":"Biological sciences/Computational biology and bioinformatics/Machine learning"}],"tags":[],"updatedAt":"2025-05-26T16:04:32+00:00","versionOfRecord":{"articleIdentity":"rs-5549340","link":"https://doi.org/10.1038/s41598-025-98861-9","journal":{"identity":"scientific-reports","isVorOnly":false,"title":"Scientific Reports"},"publishedOn":"2025-05-22 15:58:27","publishedOnDateReadable":"May 22nd, 2025"},"versionCreatedAt":"2025-04-10 05:27:26","video":"","vorDoi":"10.1038/s41598-025-98861-9","vorDoiUrl":"https://doi.org/10.1038/s41598-025-98861-9","workflowStages":[]},"version":"v1","identity":"rs-5549340","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-5549340","identity":"rs-5549340","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}