The DOI matches its Crossref record; 43 of the 44 references it cites were found in a registry or an index; 23 of the 35 sources its claims cite were checked for support.
Machine learning-driven prediction model for successful weaning of patients from mechanical ventilation in ICU
Open accessAbstract
BACKGROUND: Mechanical ventilation is a critical life support technology in the intensive care unit. However, the weaning process remains complex, making the optimal timing for liberation from ventilation challenging to ascertain and imposing a considerable clinical workload. Additionally, advanced weaning assistance tools that integrate multidimensional clinical factors to help clinical staff make precise decisions during the weaning process are lacking. The aim of this study to develop and validate an interpretable machine learning model that comprehensively evaluates the factors influencing weaning to provide clinical decision support for weaning. METHOD: We collected data from the ICU of Shanghai Tenth People's Hospital and its affiliated hospitals. Ten distinct machine learning algorithms for predicting extubation outcomes in patients receiving mechanical ventilation were developed and internally validated. Model performance was quantified using the area under the receiver operating characteristic curve AUC, overall accuracy, sensitivity, specificity, and F1 score. NRI, IDI, and DCA were used to comprehensively identify the optimal model. The relative contribution of each predictor was ranked and compared through SHAP analysis, and the best-performing model was externally validated. RESULTS: Through univariate and LASSO analyses, 24 predictive variables for machine learning model construction were identified. Comprehensive evaluation showed that among the candidate algorithms, the LGB model demonstrated the highest overall performance. SHAP analysis revealed that the top-ranked features for predicting successful liberation from mechanical ventilation were creatinine levels, lactate levels, the level of consciousness, SpO2, systolic blood pressure, cough reflex, chronic respiratory disease, diastolic blood pressure, and age. CONCLUSIONS: The optimized predictive model, which was developed through the integration of multidimensional predictive factors with diverse machine learning algorithms, exhibits superior predictive accuracy and demonstrates significant clinical potential for determining the optimal timing for weaning patients receiving invasive mechanical ventilation. Trial registration Current Controlled Trials ChiCTR2400093658; registration date: December 10, 2024.
Overview
CoChat Summary
TL;DR
This multicenter study developed and internally/externally validated machine learning models to predict successful weaning from invasive mechanical ventilation in ICU patients, selecting Light Gradient Boosting (LGB) as the best-performing model. The model showed strong discrimination and clinical utility, with external validation AUC 0.972 and key predictors including creatinine, lactate, consciousness level, and SpO2.
Key findings
- From univariate screening and LASSO regression, the authors selected 24 predictor variables for model building.
- Among 10 classifiers, the LGB model was chosen as optimal; its internal validation performance was AUC 0.952, accuracy 0.954, sensitivity 0.980, specificity 0.803, F1 score 0.973, and precision 0.966. The RF model had slightly higher AUC (0.968) but the authors still selected LGB.
- In the external validation set, the final model reached AUC 0.972, and DCA suggested significant net benefits across threshold probabilities.
- SHAP ranked the top features for successful liberation as creatinine, lactate, consciousness level, SpO2, systolic blood pressure, cough reflex, chronic respiratory disease, diastolic blood pressure, and age.
Methods
- Multicenter retrospective cohort using ICU data from Shanghai Tenth People’s Hospital and affiliated hospitals for model development/internal validation (Jan 1, 2021–Jun 30, 2024) and a separate ICU cohort for external validation (Dec 1, 2024–Jul 30, 2025).
- Included adults ≥18 years who were in ICU >48 h and received mechanical ventilation >24 h; excluded ECMO, advanced cancer, and palliative-care cases. The analysis included 1371 patients in the development/internal-validation cohort and 402 in external validation.
- Built and compared 10 machine-learning classifiers (KNN, SVM, NN, XGB, LR, RF, DT, ET, SGD, LGB), using a 7:3 train/test split, LASSO with fivefold cross-validation, SMOTE for imbalance, and evaluation by AUC, accuracy, sensitivity, specificity, F1, NRI, IDI, DCA, and SHAP.
Limitations
- The paper states that it did not include imaging data such as diaphragmatic displacement/thickness, despite their potential relevance to weaning.
- It also states that invasive hemodynamic indicators (e.g., CVP, PCWP, PiCCO) were difficult to obtain, so more accessible proxies were used instead.
- The text makes evident that the study is retrospective and relies on data from a specific hospital system; this may limit generalizability beyond the studied setting (evident from the design and data source, not explicitly framed as a limitation by the authors).
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Figures





Verification
Claims the paper makes
15 Valid · 3 Not valid · 6 Not checked · 40 Own results and interpretations
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“In step three, the sample size needed to reduce model overfitting is calculated, assuming an expected shrinkage rate of 10%, with an expected number of variables P to be included in the final predictive model of 30: S is the expected shrinkage factor, and the noise ratio evaluation index R2cs is 0.3; the calculated sample size is 740 cases.”
§ Machine learning-driven prediction model for successful weaning of patients from mechanical ventilation in ICU > Methods > Sample size calculation
What the sources say
- Partly supportedcites Riley [13] Riley RD, Ensor J, Snell KIE et al (2020) Calculating the sample size required for developing a clinical prediction model. BMJ 368:m441 sourceExperimental — accuracy not yet measured
The paper’s Step 3 section (page not available in our copy) supports calculating sample size to limit overfitting using 10% expected shrinkage, the number of candidate predictor parameters, and anticipated Cox–Snell R². It does not report the claimed calculation of 740 cases for 30 parameters and R²cs of 0.3; its worked examples use different inputs. (LLM, conf=0.86)
plainstates a causen = 740- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
The sources with a verdict support this claim; the rest weren’t checked yet, have no full text we could read, or their verdicts aren’t rated yet.
In the paper
“Furthermore, this model is particularly well equipped to address the challenges posed by imbalanced datasets.”
§ Machine learning-driven prediction model for successful weaning of patients from mechanical ventilation in ICU > Discussion The aim of this respective, multicenter study was to investigate and compare 10 ML models designed to predict successful weaning from mechanical ventilation. Machine learning techniques have shown great potential in the medical field. Compared to previous studies, this research has constructed 10 machine learning models and compared the advantages and disadvantages of each model, thereby better overcoming the limitations of each model [20]. Ultimately, we identified a set of predictive risk factors and built the optimal predictive model. To assess the performance of the models, we utilized two metrics, IDI and NRI, which are typically used to compare the predictive capabilities of different models to assist researchers in selecting the best predictive model. The higher the values of NRI and IDI, the greater the improvement of the new model over the baseline model, indicating stronger predictive capabilities [21]. The results show that the LGB model significantly outperformed other models in classification accuracy, while also being more advantageous in terms of economy, user-friendliness, and adaptability to different clinical scenarios. Additionally, DCA confirmed that this machine learning model has good clinical applicability and acceptance in the weaning context. Finally, SHAP analysis revealed the most influential predictive factors for determining successful weaning of mechanically ventilated patients, enhancing the interpretability and transparency of current machine learning models. Therefore, this study provides valuable insights for personalized and precise weaning predictions for patients receiving invasive mechanical ventilation support. On the basis of univariate screening, we used LASSO regression for feature selection modeling. This approach effectively addresses issues of multicollinearity and overfitting among variables. Key factors that predict the withdrawal of invasive mechanical ventilation support were identified [22]. The use of the LASSO method enhances both the predictive capability and the stability of the model. Studies have shown that consciousness, Glasgow Coma Scale (GCS) score, duration of mechanical ventilation, cough strength, age, Acute Physiology and Chronic Health Evaluation II(APACHEII) score, psychological factors, hemoglobin level, Sequential Organ Failure Assessment(SOFA) score, serum albumin level, and certain respiratory parameters are significant predictors of successful weaning [23–26], consistent with the results of our LASSO analysis. However, the insensitivity and specificity of these indicators have also been noted, and studies have reported that these indicators do not consider the influence of other factors, making it imperative to improve the robustness of clinical decision-making [4, 27]. Jia et al. [9] identified several factors that influence weaning outcomes, including the fraction of inspired Oxygen(FiO2), ventilator mode, peak pressure, positive end-expiratory pressure(PEEP), tidal volume, and mean airway pressure. Otaguro [28] reported that duration of mechanical ventilation, FiO2, PEEP, maximum and mean airway pressures, age, and albumin concentration were important predictors of successful weaning. Lin [29] also reported that peak pressure, respiratory rate, and APACHE II score are significant influencing factors. However, these studies did not comprehensively consider all potential clinical influencing factors. This current study incorporates multiple dimensions of factors affecting weaning, including patient demographic characteristics, comorbidities, vital signs, treatment measures, and laboratory indicators obtained from blood gas and biochemical analyses. This comprehensive approach takes into account the multifaceted influences on weaning, thereby avoiding the biases that may have arisen from a narrow focus on weaning factors in previous research. Current guidelines [24, 30] recommend the use of the spontaneous breathing trial (SBT) as a tool for predicting weaning outcomes. Nevertheless, approximately 10% to 20% of patients who successfully complete an SBT may require reintubation within 48 h after extubation [25]. Various technical methods, including risk prediction nomograms [31], machine learning [20, 32], and a combination of association rule learning and weighted network analysis [33], have been used to analyze and process datasets related to factors or characteristics influencing weaning, leading to the establishment of predictive models. However, researchers often construct risk prediction models on the basis of a single influencing factor, resulting in heterogeneity and a lack of robustness in clinical decision-making. Most models also lack external validation. Furthermore, a few studies have provided explanations of the constructed models. In the ICU, parameters such as vital signs, biochemical indicators, and ventilator-related metrics are readily available. However, integrating these indicators in a way that properly guides weaning remains a challenge in current research. Compared with traditional predictive models, machine learning offers better predictive performance [34]. Prediction of the success of withdrawal from mechanical ventilation involves both continuous and categorical variables, and this often results in the use of imbalanced and sparse datasets. Our research indicates that the LGB model is optimal for this task. One key advantage of LGB is its ability to handle mixed-type features, this helps preserve the true distribution of these variables within the model. Additionally, LGB uses a histogram-based decision tree approach, discretizing continuous features during the data-splitting process, and thereby allowing for effective management of sparse data. Furthermore, this model is particularly well equipped to address the challenges posed by imbalanced datasets. Although we took steps to mitigate the imbalance during data preprocessing, employing a model that is inherently proficient in managing such issues is more suitable for our objectives [35–37]. To improve the interpretability and intuitiveness of ML models, we applied SHAP values to analyze LGB model. SHAP values provide a consistent and locally accurate assessment of the contribution of each feature to the model’s predictions, enabling a more transparent understanding of the relationships between input variables and outcomes [37, 38]. The detailed insights and risk factor explanations derived from the SHAP analysis offer clinicians deeper insights into the decision-making process, allowing them to make more informed clinical judgments rather than relying solely on algorithmic outputs. The results identified several key factors associated with successful weaning from invasive mechanical ventilation, including serum creatinine levels, lactate levels, consciousness state, SpO₂, systolic blood pressure, cough reflex, chronic respiratory disease diastolic blood pressure, and age. In clinical practice, these findings highlight the importance of considering both positive and negative influencing factors revealed by SHAP analysis when making extubation decisions, thereby facilitating more personalized and evidence-based treatment strategies.
What the sources say
- Not checked — no full text of the sourcecites [35] Cai Z, Huang H, Sun G, et al. (2023) Advancing Predictive Models: Unveiling LightGBM Machine Learning for Data AnalysisC. 2023. IEEE. –01–01. source
- Could not verifycites [36] Wang Y, Wang T (2020) Application of improved LightGBM model in blood glucose prediction. Appl Sci 10(9):3227 sourceExperimental — accuracy not yet measured
The Data Preprocessing section describes missing values and the training–test split, while the Introduction compares LightGBM with XGBoost on memory use and speed. Neither passage addresses performance on imbalanced datasets; searches for class imbalance, uneven sample distributions, and LightGBM sampling found no supporting passage in our copy. (LLM, conf=0.72)
- Supportedcites [37] Qi X, Wang S, Fang C et al (2025) Machine learning and SHAP value interpretation for predicting comorbidity of cardiovascular disease and cancer with dietary antioxidants. Redox Biol 79(2):103470 source
The Discussion explicitly says LightGBM is well-suited for handling imbalanced data. It also notes that the study addressed imbalance during preprocessing, so the passage supports the claim about the model’s suitability without establishing that the model alone resolved the imbalance. (LLM, conf=0.94)
boosted- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“SHAP values provide a consistent and locally accurate assessment of the contribution of each feature to the model’s predictions, enabling a more transparent understanding of the relationships between input variables and outcomes [37, 38].”
§ Machine learning-driven prediction model for successful weaning of patients from mechanical ventilation in ICU > Discussion The aim of this respective, multicenter study was to investigate and compare 10 ML models designed to predict successful weaning from mechanical ventilation. Machine learning techniques have shown great potential in the medical field. Compared to previous studies, this research has constructed 10 machine learning models and compared the advantages and disadvantages of each model, thereby better overcoming the limitations of each model [20]. Ultimately, we identified a set of predictive risk factors and built the optimal predictive model. To assess the performance of the models, we utilized two metrics, IDI and NRI, which are typically used to compare the predictive capabilities of different models to assist researchers in selecting the best predictive model. The higher the values of NRI and IDI, the greater the improvement of the new model over the baseline model, indicating stronger predictive capabilities [21]. The results show that the LGB model significantly outperformed other models in classification accuracy, while also being more advantageous in terms of economy, user-friendliness, and adaptability to different clinical scenarios. Additionally, DCA confirmed that this machine learning model has good clinical applicability and acceptance in the weaning context. Finally, SHAP analysis revealed the most influential predictive factors for determining successful weaning of mechanically ventilated patients, enhancing the interpretability and transparency of current machine learning models. Therefore, this study provides valuable insights for personalized and precise weaning predictions for patients receiving invasive mechanical ventilation support. On the basis of univariate screening, we used LASSO regression for feature selection modeling. This approach effectively addresses issues of multicollinearity and overfitting among variables. Key factors that predict the withdrawal of invasive mechanical ventilation support were identified [22]. The use of the LASSO method enhances both the predictive capability and the stability of the model. Studies have shown that consciousness, Glasgow Coma Scale (GCS) score, duration of mechanical ventilation, cough strength, age, Acute Physiology and Chronic Health Evaluation II(APACHEII) score, psychological factors, hemoglobin level, Sequential Organ Failure Assessment(SOFA) score, serum albumin level, and certain respiratory parameters are significant predictors of successful weaning [23–26], consistent with the results of our LASSO analysis. However, the insensitivity and specificity of these indicators have also been noted, and studies have reported that these indicators do not consider the influence of other factors, making it imperative to improve the robustness of clinical decision-making [4, 27]. Jia et al. [9] identified several factors that influence weaning outcomes, including the fraction of inspired Oxygen(FiO2), ventilator mode, peak pressure, positive end-expiratory pressure(PEEP), tidal volume, and mean airway pressure. Otaguro [28] reported that duration of mechanical ventilation, FiO2, PEEP, maximum and mean airway pressures, age, and albumin concentration were important predictors of successful weaning. Lin [29] also reported that peak pressure, respiratory rate, and APACHE II score are significant influencing factors. However, these studies did not comprehensively consider all potential clinical influencing factors. This current study incorporates multiple dimensions of factors affecting weaning, including patient demographic characteristics, comorbidities, vital signs, treatment measures, and laboratory indicators obtained from blood gas and biochemical analyses. This comprehensive approach takes into account the multifaceted influences on weaning, thereby avoiding the biases that may have arisen from a narrow focus on weaning factors in previous research. Current guidelines [24, 30] recommend the use of the spontaneous breathing trial (SBT) as a tool for predicting weaning outcomes. Nevertheless, approximately 10% to 20% of patients who successfully complete an SBT may require reintubation within 48 h after extubation [25]. Various technical methods, including risk prediction nomograms [31], machine learning [20, 32], and a combination of association rule learning and weighted network analysis [33], have been used to analyze and process datasets related to factors or characteristics influencing weaning, leading to the establishment of predictive models. However, researchers often construct risk prediction models on the basis of a single influencing factor, resulting in heterogeneity and a lack of robustness in clinical decision-making. Most models also lack external validation. Furthermore, a few studies have provided explanations of the constructed models. In the ICU, parameters such as vital signs, biochemical indicators, and ventilator-related metrics are readily available. However, integrating these indicators in a way that properly guides weaning remains a challenge in current research. Compared with traditional predictive models, machine learning offers better predictive performance [34]. Prediction of the success of withdrawal from mechanical ventilation involves both continuous and categorical variables, and this often results in the use of imbalanced and sparse datasets. Our research indicates that the LGB model is optimal for this task. One key advantage of LGB is its ability to handle mixed-type features, this helps preserve the true distribution of these variables within the model. Additionally, LGB uses a histogram-based decision tree approach, discretizing continuous features during the data-splitting process, and thereby allowing for effective management of sparse data. Furthermore, this model is particularly well equipped to address the challenges posed by imbalanced datasets. Although we took steps to mitigate the imbalance during data preprocessing, employing a model that is inherently proficient in managing such issues is more suitable for our objectives [35–37]. To improve the interpretability and intuitiveness of ML models, we applied SHAP values to analyze LGB model. SHAP values provide a consistent and locally accurate assessment of the contribution of each feature to the model’s predictions, enabling a more transparent understanding of the relationships between input variables and outcomes [37, 38]. The detailed insights and risk factor explanations derived from the SHAP analysis offer clinicians deeper insights into the decision-making process, allowing them to make more informed clinical judgments rather than relying solely on algorithmic outputs. The results identified several key factors associated with successful weaning from invasive mechanical ventilation, including serum creatinine levels, lactate levels, consciousness state, SpO₂, systolic blood pressure, cough reflex, chronic respiratory disease diastolic blood pressure, and age. In clinical practice, these findings highlight the importance of considering both positive and negative influencing factors revealed by SHAP analysis when making extubation decisions, thereby facilitating more personalized and evidence-based treatment strategies.
What the sources say
- Partly supportedcites [37] Qi X, Wang S, Fang C et al (2025) Machine learning and SHAP value interpretation for predicting comorbidity of cardiovascular disease and cancer with dietary antioxidants. Redox Biol 79(2):103470 sourceExperimental — accuracy not yet measured
The Statistical analysis section says SHAP aggregates local feature contributions and offers insight into each feature’s role in specific predictions. The Discussion likewise says SHAP evaluates each feature’s effect on model outputs. Neither passage establishes the additional claims of consistency or local accuracy. (LLM, conf=0.85)
- Supportedcites [38] Wang K, Tian J, Zheng C et al (2021) Interpretable prediction of 3-year all-cause mortality in patients with heart failure caused by coronary heart disease based on machine learning and SHAP. Comput Biol Med 137:104813 source
The Discussion states that SHAP provides consistent and locally accurate attribution values for each feature in the prediction model. Section 3.8 adds that SHAP values show each feature’s contribution to an individual prediction and help explain it. (LLM, conf=0.94)
plainstates an association- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
The sources with a verdict support this claim; the rest weren’t checked yet, have no full text we could read, or their verdicts aren’t rated yet.
In the paper
“One key advantage of LGB is its ability to handle mixed-type features, this helps preserve the true distribution of these variables within the model.”
§ Machine learning-driven prediction model for successful weaning of patients from mechanical ventilation in ICU > Discussion The aim of this respective, multicenter study was to investigate and compare 10 ML models designed to predict successful weaning from mechanical ventilation. Machine learning techniques have shown great potential in the medical field. Compared to previous studies, this research has constructed 10 machine learning models and compared the advantages and disadvantages of each model, thereby better overcoming the limitations of each model [20]. Ultimately, we identified a set of predictive risk factors and built the optimal predictive model. To assess the performance of the models, we utilized two metrics, IDI and NRI, which are typically used to compare the predictive capabilities of different models to assist researchers in selecting the best predictive model. The higher the values of NRI and IDI, the greater the improvement of the new model over the baseline model, indicating stronger predictive capabilities [21]. The results show that the LGB model significantly outperformed other models in classification accuracy, while also being more advantageous in terms of economy, user-friendliness, and adaptability to different clinical scenarios. Additionally, DCA confirmed that this machine learning model has good clinical applicability and acceptance in the weaning context. Finally, SHAP analysis revealed the most influential predictive factors for determining successful weaning of mechanically ventilated patients, enhancing the interpretability and transparency of current machine learning models. Therefore, this study provides valuable insights for personalized and precise weaning predictions for patients receiving invasive mechanical ventilation support. On the basis of univariate screening, we used LASSO regression for feature selection modeling. This approach effectively addresses issues of multicollinearity and overfitting among variables. Key factors that predict the withdrawal of invasive mechanical ventilation support were identified [22]. The use of the LASSO method enhances both the predictive capability and the stability of the model. Studies have shown that consciousness, Glasgow Coma Scale (GCS) score, duration of mechanical ventilation, cough strength, age, Acute Physiology and Chronic Health Evaluation II(APACHEII) score, psychological factors, hemoglobin level, Sequential Organ Failure Assessment(SOFA) score, serum albumin level, and certain respiratory parameters are significant predictors of successful weaning [23–26], consistent with the results of our LASSO analysis. However, the insensitivity and specificity of these indicators have also been noted, and studies have reported that these indicators do not consider the influence of other factors, making it imperative to improve the robustness of clinical decision-making [4, 27]. Jia et al. [9] identified several factors that influence weaning outcomes, including the fraction of inspired Oxygen(FiO2), ventilator mode, peak pressure, positive end-expiratory pressure(PEEP), tidal volume, and mean airway pressure. Otaguro [28] reported that duration of mechanical ventilation, FiO2, PEEP, maximum and mean airway pressures, age, and albumin concentration were important predictors of successful weaning. Lin [29] also reported that peak pressure, respiratory rate, and APACHE II score are significant influencing factors. However, these studies did not comprehensively consider all potential clinical influencing factors. This current study incorporates multiple dimensions of factors affecting weaning, including patient demographic characteristics, comorbidities, vital signs, treatment measures, and laboratory indicators obtained from blood gas and biochemical analyses. This comprehensive approach takes into account the multifaceted influences on weaning, thereby avoiding the biases that may have arisen from a narrow focus on weaning factors in previous research. Current guidelines [24, 30] recommend the use of the spontaneous breathing trial (SBT) as a tool for predicting weaning outcomes. Nevertheless, approximately 10% to 20% of patients who successfully complete an SBT may require reintubation within 48 h after extubation [25]. Various technical methods, including risk prediction nomograms [31], machine learning [20, 32], and a combination of association rule learning and weighted network analysis [33], have been used to analyze and process datasets related to factors or characteristics influencing weaning, leading to the establishment of predictive models. However, researchers often construct risk prediction models on the basis of a single influencing factor, resulting in heterogeneity and a lack of robustness in clinical decision-making. Most models also lack external validation. Furthermore, a few studies have provided explanations of the constructed models. In the ICU, parameters such as vital signs, biochemical indicators, and ventilator-related metrics are readily available. However, integrating these indicators in a way that properly guides weaning remains a challenge in current research. Compared with traditional predictive models, machine learning offers better predictive performance [34]. Prediction of the success of withdrawal from mechanical ventilation involves both continuous and categorical variables, and this often results in the use of imbalanced and sparse datasets. Our research indicates that the LGB model is optimal for this task. One key advantage of LGB is its ability to handle mixed-type features, this helps preserve the true distribution of these variables within the model. Additionally, LGB uses a histogram-based decision tree approach, discretizing continuous features during the data-splitting process, and thereby allowing for effective management of sparse data. Furthermore, this model is particularly well equipped to address the challenges posed by imbalanced datasets. Although we took steps to mitigate the imbalance during data preprocessing, employing a model that is inherently proficient in managing such issues is more suitable for our objectives [35–37]. To improve the interpretability and intuitiveness of ML models, we applied SHAP values to analyze LGB model. SHAP values provide a consistent and locally accurate assessment of the contribution of each feature to the model’s predictions, enabling a more transparent understanding of the relationships between input variables and outcomes [37, 38]. The detailed insights and risk factor explanations derived from the SHAP analysis offer clinicians deeper insights into the decision-making process, allowing them to make more informed clinical judgments rather than relying solely on algorithmic outputs. The results identified several key factors associated with successful weaning from invasive mechanical ventilation, including serum creatinine levels, lactate levels, consciousness state, SpO₂, systolic blood pressure, cough reflex, chronic respiratory disease diastolic blood pressure, and age. In clinical practice, these findings highlight the importance of considering both positive and negative influencing factors revealed by SHAP analysis when making extubation decisions, thereby facilitating more personalized and evidence-based treatment strategies.
What the sources say
- Not checked — no full text of the sourcecites [35] Cai Z, Huang H, Sun G, et al. (2023) Advancing Predictive Models: Unveiling LightGBM Machine Learning for Data AnalysisC. 2023. IEEE. –01–01. source
- Could not verifycites [36] Wang Y, Wang T (2020) Application of improved LightGBM model in blood glucose prediction. Appl Sci 10(9):3227 sourceExperimental — accuracy not yet measured
The Data Preprocessing section describes the dataset and how missing values were handled, while the LightGBM comparison passage reports model parameters and performance. Searches for mixed-type or categorical features and preservation of their distribution found no passage in our copy addressing the claim. (LLM, conf=0.70)
- Supportedcites [37] Qi X, Wang S, Fang C et al (2025) Machine learning and SHAP value interpretation for predicting comorbidity of cardiovascular disease and cancer with dietary antioxidants. Redox Biol 79(2):103470 source
The Discussion says LightGBM supports the study’s mix of continuous and categorical variables. It adds that avoiding one-hot or label encoding for categorical features helps preserve their true distribution in the model. (LLM, conf=0.94)
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The other important features are lactate levels, the consciousness level, SpO2, systolic blood pressure, cough reflex, chronic respiratory disease, diastolic blood pressure, and age, in that order.”
§ Machine learning-driven prediction model for successful weaning of patients from mechanical ventilation in ICU > Results > SHAP analysis of the final model
plainstates an association- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
19 more claims
Create a free account to see all 64 claims and every source verdict.
- It may state two separate claims as one. 26
- Probably not a substantial claim. 20
Citations
Cites
Calculating the sample size required for developing a clinical prediction model
Riley, R. D., Ensor, J., Snell, K. I. E., Harrell, F. E., Martin, G. P., Reitsma, J. B., Moons, K. G. M., Collins, G., & van Smeden, M. (2020) · BMJ · doi:10.1136/bmj.m441 · Cited by 2,917 · cited 1×
“The sample size was calculated using the formula for sample size estimation in predictive modeling described by Riley [13].”
Choosing Prediction Over Explanation in Psychology: Lessons From Machine Learning
Yarkoni, T., & Westfall, J. (2017) · Perspectives on Psychological Science · doi:10.1177/1745691617693393 · Cited by 1,916 · cited 1×
“Key factors that predict the withdrawal of invasive mechanical ventilation support were identified [22].”
Automated machine learning: Review of the state-of-the-art and opportunities for healthcare
Waring, J., Lindvall, C., & Umeton, R. (2020) · Artificial Intelligence in Medicine · doi:10.1016/j.artmed.2020.101822 · Cited by 834 · cited 1×
“Compared with traditional predictive models, machine learning offers better predictive performance [34].”
Ouellette, D. R., Patel, S., Girard, T. D., Morris, P. E., Schmidt, G. A., Truwit, J. D., Alhazzani, W., Burns, S. M., Epstein, S. K., Esteban, A., et al. (2017) · Chest · doi:10.1016/j.chest.2016.10.036 · Cited by 432 · cited 1×
“Current guidelines [24, 30] recommend the use of the spontaneous breathing trial (SBT) as a tool for predicting weaning outcomes.”
Girard, T. D., Alhazzani, W., Kress, J. P., Ouellette, D. R., Schmidt, G. A., Truwit, J. D., Burns, S. M., Epstein, S. K., Esteban, A., Fan, E., et al. (2017) · American Journal of Respiratory and Critical Care Medicine · doi:10.1164/rccm.201610-2075st · Cited by 387
Wang, K., Tian, J., Zheng, C., Yang, H., Ren, J., Liu, Y., Han, Q., & Zhang, Y. (2021) · Computers in Biology and Medicine · doi:10.1016/j.compbiomed.2021.104813 · Cited by 361 · cited 1×
“SHAP values provide a consistent and locally accurate assessment of the contribution of each feature to the model’s predictions, enabling a more transparent understanding of the relationships between input variables and outcomes [37, 38].”
Peñuelas, O., Frutos-Vivar, F., Fernández, C., Anzueto, A., Epstein, S. K., Apezteguía, C., González, M., Nin, N., Raymondos, K., Tomicic, V., et al. (2011) · American Journal of Respiratory and Critical Care Medicine · doi:10.1164/rccm.201011-1887oc · Cited by 350 · cited 1×
“However, the insensitivity and specificity of these indicators have also been noted, and studies have reported that these indicators do not consider the influence of other factors, making it imperative to improve the robustness of clinical decision-making [4, 27].”
Umbrello, M., Formenti, P., Longhi, D., Galimberti, A., Piva, I., Pezzi, A., Mistraletti, G., Marini, J. J., & Iapichino, G. (2015) · Critical Care · doi:10.1186/s13054-015-0894-9 · Cited by 343 · cited 1×
“Numerous studies [41, 42] have shown that both DD and diaphragmatic thickness fraction (DTF) can accurately reflect diaphragmatic function.”
Goligher, E. C., Laghi, F., Detsky, M. E., Farias, P., Murray, A., Brace, D., Brochard, L. J., Bolz, S.-S., Rubenfeld, G. D., Kavanagh, B. P., et al. (2015) · Intensive Care Medicine · doi:10.1007/s00134-015-3724-2 · Cited by 315
Research works printed but not linked: 1
- Windisch WDDGJ (2020) Prolonged weaning from mechanical ventilation. Dtsch Arztebl Int 117(12):197–204
Cited by
Wang, M., Zhao, C., Hao, Z., Wu, L., Mao, W., & Zhao, J. (2026) · Frontiers in Medicine · doi:10.3389/fmed.2026.1904503
Metadata
- Authors
- Qiu, C., Tang, L., Zhuang, Y., Chi, C., Zheng, K., & Zhu, X.
- Affiliations
- Tongji University (CN)
- Shanghai Tenth People's Hospital (CN)
- Corresponding author
- Zhu, Xiaoping
- Venue
- Intensive Care Medicine Experimental
- Published
- 2026-01-21
- Volume
- 14
- Issue
- 1
- Pages
- 6-6
- ISSN
- 2197-425X (ISSN-L)
- PMC
- PMC12824072
- OpenAlex
- W7125115277
- Cited by
- 1
- References
- 44
- Citations by year
- 2026: 1
- FWCI
- 4.75
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Licence
- cc-by
- creativecommons.org/licenses/by/4.0 (tdm)
- creativecommons.org/licenses/by/4.0 (vor)
- Locations
- Intensive Care Medicine Experimental (publishedVersion, cc-by)
- PubMed (publishedVersion)
- PubMed Central (publishedVersion, cc-by)
- Funders
- the Shanghai Nursing Association Research Project (2024MS-B21)
- National Natural Science Foundation of China (72074168)
- Topics
- Health Sciences › Medicine › Pulmonary and Respiratory Medicine › Respiratory Support and Mechanisms
- Physical Sciences › Computer Science › Artificial Intelligence › Machine Learning in Healthcare
- Health Sciences › Health Professions › Health Information Management › Artificial Intelligence in Healthcare
- Keywords
- weaning from mechanical ventilation, extubation prediction, intensive care unit, clinical decision support, model interpretability, SHAP analysis, LASSO feature selection, gradient boosting, lactate, creatinine, level of consciousness, cough reflex, chronic respiratory disease
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Machine learning-driven prediction model for successful weaning of patients from mechanical ventilation in ICU. https://app.cochat.ai/papers/10.1186/s40635-026-00859-8
- The paper
- Qiu, C., Tang, L., Zhuang, Y., Chi, C., Zheng, K., & Zhu, X. (2026). Machine learning-driven prediction model for successful weaning of patients from mechanical ventilation in ICU. Intensive Care Medicine Experimental. https://doi.org/10.1186/s40635-026-00859-8
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.