AS
A. Schoe
info
Please Note
<p>This page displays the records of the person named above and is not linked to a unique person identifier. This record may need to be merged to a profile.</p>
3 records found
1
Objective
The primary aim of this study was to develop and validate a machine learning prediction model for respiratory deterioration in mechanically ventilated Intensive Care Unit (ICU) patients. The secondary aim was to identify physiological parameters associated with respiratory failure during mechanical ventilation.
Methods
Two distinct prediction models were developed using data from ICU patients admitted to the Leiden University Medical Centre (LUMC) between 2018 and 2023. Patients receiving invasive mechanical ventilation (IMV) for at least 48 hours with a PaO2/FiO2 ratio below 40 kPa were included and allocated to COVID training, COVID test, or non-COVID test sets. Model 1 predicts respiratory deterioration within six hours after switching from controlled to assisted ventilation. Model 2 is an hourly updating model predicting respiratory deterioration occurring more than six hours after this switch. XGBoost models were cross-validated on the COVID training set to identify the optimal observation windows and prediction horizons, after which feature selection and hyperparameter optimisation were performed. Model 1 was optimised for the area under the receiver operating characteristic (AUROC) and Model 2 for the area under the precision-recall curve (AUPRC). Discriminative performance, generalisability, and clinical utility were evaluated on the COVID and non-COVID test sets.
Results
A total of 296 patients were included in the COVID training set, 78 in the COVID test set, and 755 to the non-COVID test set. For Model 1, a one-hour observation window was selected. The most important features were the mean fraction of inspired oxygen (FiO2), propofol infusion rate, and peripheral oxygen saturation (SpO2). This model achieved an AUROC of 0.78 on the COVID test and 0.76 on the non-COVID test set. For model 2, a two-hour observation window and a six-hour prediction horizon were selected, with the SpO2/FiO2 ratio as the most important input feature. This model achieved an AUPRC of 0.05 on the COVID test set and 0.03 on the non-COVID test set.
Conclusion
Model 1 demonstrated moderate discriminative performance but limited clinical utility at relevant operating points. Model 2 showed very limited predictive value, primarily due to extreme class imbalance. Consequently, neither model is currently suitable for clinical implementation. With larger datasets and more advanced modelling techniques, Model 1 may have the potential to become a clinically useful decision support tool to support decisions on switching from controlled to assisted ventilation. ...
The primary aim of this study was to develop and validate a machine learning prediction model for respiratory deterioration in mechanically ventilated Intensive Care Unit (ICU) patients. The secondary aim was to identify physiological parameters associated with respiratory failure during mechanical ventilation.
Methods
Two distinct prediction models were developed using data from ICU patients admitted to the Leiden University Medical Centre (LUMC) between 2018 and 2023. Patients receiving invasive mechanical ventilation (IMV) for at least 48 hours with a PaO2/FiO2 ratio below 40 kPa were included and allocated to COVID training, COVID test, or non-COVID test sets. Model 1 predicts respiratory deterioration within six hours after switching from controlled to assisted ventilation. Model 2 is an hourly updating model predicting respiratory deterioration occurring more than six hours after this switch. XGBoost models were cross-validated on the COVID training set to identify the optimal observation windows and prediction horizons, after which feature selection and hyperparameter optimisation were performed. Model 1 was optimised for the area under the receiver operating characteristic (AUROC) and Model 2 for the area under the precision-recall curve (AUPRC). Discriminative performance, generalisability, and clinical utility were evaluated on the COVID and non-COVID test sets.
Results
A total of 296 patients were included in the COVID training set, 78 in the COVID test set, and 755 to the non-COVID test set. For Model 1, a one-hour observation window was selected. The most important features were the mean fraction of inspired oxygen (FiO2), propofol infusion rate, and peripheral oxygen saturation (SpO2). This model achieved an AUROC of 0.78 on the COVID test and 0.76 on the non-COVID test set. For model 2, a two-hour observation window and a six-hour prediction horizon were selected, with the SpO2/FiO2 ratio as the most important input feature. This model achieved an AUPRC of 0.05 on the COVID test set and 0.03 on the non-COVID test set.
Conclusion
Model 1 demonstrated moderate discriminative performance but limited clinical utility at relevant operating points. Model 2 showed very limited predictive value, primarily due to extreme class imbalance. Consequently, neither model is currently suitable for clinical implementation. With larger datasets and more advanced modelling techniques, Model 1 may have the potential to become a clinically useful decision support tool to support decisions on switching from controlled to assisted ventilation. ...
Objective
The primary aim of this study was to develop and validate a machine learning prediction model for respiratory deterioration in mechanically ventilated Intensive Care Unit (ICU) patients. The secondary aim was to identify physiological parameters associated with respiratory failure during mechanical ventilation.
Methods
Two distinct prediction models were developed using data from ICU patients admitted to the Leiden University Medical Centre (LUMC) between 2018 and 2023. Patients receiving invasive mechanical ventilation (IMV) for at least 48 hours with a PaO2/FiO2 ratio below 40 kPa were included and allocated to COVID training, COVID test, or non-COVID test sets. Model 1 predicts respiratory deterioration within six hours after switching from controlled to assisted ventilation. Model 2 is an hourly updating model predicting respiratory deterioration occurring more than six hours after this switch. XGBoost models were cross-validated on the COVID training set to identify the optimal observation windows and prediction horizons, after which feature selection and hyperparameter optimisation were performed. Model 1 was optimised for the area under the receiver operating characteristic (AUROC) and Model 2 for the area under the precision-recall curve (AUPRC). Discriminative performance, generalisability, and clinical utility were evaluated on the COVID and non-COVID test sets.
Results
A total of 296 patients were included in the COVID training set, 78 in the COVID test set, and 755 to the non-COVID test set. For Model 1, a one-hour observation window was selected. The most important features were the mean fraction of inspired oxygen (FiO2), propofol infusion rate, and peripheral oxygen saturation (SpO2). This model achieved an AUROC of 0.78 on the COVID test and 0.76 on the non-COVID test set. For model 2, a two-hour observation window and a six-hour prediction horizon were selected, with the SpO2/FiO2 ratio as the most important input feature. This model achieved an AUPRC of 0.05 on the COVID test set and 0.03 on the non-COVID test set.
Conclusion
Model 1 demonstrated moderate discriminative performance but limited clinical utility at relevant operating points. Model 2 showed very limited predictive value, primarily due to extreme class imbalance. Consequently, neither model is currently suitable for clinical implementation. With larger datasets and more advanced modelling techniques, Model 1 may have the potential to become a clinically useful decision support tool to support decisions on switching from controlled to assisted ventilation.
The primary aim of this study was to develop and validate a machine learning prediction model for respiratory deterioration in mechanically ventilated Intensive Care Unit (ICU) patients. The secondary aim was to identify physiological parameters associated with respiratory failure during mechanical ventilation.
Methods
Two distinct prediction models were developed using data from ICU patients admitted to the Leiden University Medical Centre (LUMC) between 2018 and 2023. Patients receiving invasive mechanical ventilation (IMV) for at least 48 hours with a PaO2/FiO2 ratio below 40 kPa were included and allocated to COVID training, COVID test, or non-COVID test sets. Model 1 predicts respiratory deterioration within six hours after switching from controlled to assisted ventilation. Model 2 is an hourly updating model predicting respiratory deterioration occurring more than six hours after this switch. XGBoost models were cross-validated on the COVID training set to identify the optimal observation windows and prediction horizons, after which feature selection and hyperparameter optimisation were performed. Model 1 was optimised for the area under the receiver operating characteristic (AUROC) and Model 2 for the area under the precision-recall curve (AUPRC). Discriminative performance, generalisability, and clinical utility were evaluated on the COVID and non-COVID test sets.
Results
A total of 296 patients were included in the COVID training set, 78 in the COVID test set, and 755 to the non-COVID test set. For Model 1, a one-hour observation window was selected. The most important features were the mean fraction of inspired oxygen (FiO2), propofol infusion rate, and peripheral oxygen saturation (SpO2). This model achieved an AUROC of 0.78 on the COVID test and 0.76 on the non-COVID test set. For model 2, a two-hour observation window and a six-hour prediction horizon were selected, with the SpO2/FiO2 ratio as the most important input feature. This model achieved an AUPRC of 0.05 on the COVID test set and 0.03 on the non-COVID test set.
Conclusion
Model 1 demonstrated moderate discriminative performance but limited clinical utility at relevant operating points. Model 2 showed very limited predictive value, primarily due to extreme class imbalance. Consequently, neither model is currently suitable for clinical implementation. With larger datasets and more advanced modelling techniques, Model 1 may have the potential to become a clinically useful decision support tool to support decisions on switching from controlled to assisted ventilation.
Exploring Multimodal Respiratory Monitoring and Ultrasound-based Quantification of Diaphragm Functio
Towards personalized ventilation strategies in critically ill patients
Mechanical ventilation (MV) in the intensive care unit is inherently associated with both life-saving benefits and the risk of lung and diaphragm injury, making the timing of ventilator detachment a well-studied clinical challenge. This creates a need for reliable monitoring of respiratory effort to tailor MV to individual patient requirements. Diaphragm ultrasound is a non-invasive technique with respiratory monitoring potential. However, its application is limited by operator dependency and the lack of automated, continuous analysis. To help overcome these limitations, the Sonoskin project aims to develop a wearable and operator-independent ultrasound-based respiratory monitoring system, integrating diaphragm ultrasound with surface electromyography (sEMG) as a multimodal monitoring approach. The European consortium Sonoskin forms the framework of this thesis.
The primary objective of this thesis was to evaluate the feasibility of continuous ultrasound-based diaphragm monitoring and to develop automated methods for extracting clinically relevant diaphragm parameters, with a specific focus on diaphragm thickening fraction (DTF). This thesis consists of two parts.
In Part I, a physiological study design was developed to enable synchronized acquisition of diaphragm ultrasound, sEMG, esophageal pressure, ventilator data, and gas-exchange parameters in healthy volunteers.
In Part II, a semi-automated algorithm for diaphragm parameter extraction was developed using a Radon transform-based adaptive M-mode approach. This method explicitly accounts for changes in diaphragm orientation and enables robust tracking of both pleural and peritoneal interfaces. When applied to 110 ultrasound recordings from ten healthy volunteers, the Radon-based method showed markedly improved line-tracking performance and superior image quality compared with a conventional static M-mode approach. These improvements were reflected by significantly higher contrast and smoother, more traceable signals in both spatial and intensity domains. Automated DTF measurements showed a moderate correlation between the left and right hemidiaphragms. Absolute diaphragm thickness and DTF values were within the lower range of normal values reported in the literature, supporting the need for further evaluation and potential redefinition of reference values when using automated Radon-based tracking methods.
In addition to diaphragm thickening, angular motion of the diaphragm was quantified through analysis of the diaphragmatic angle (θd). Cyclic changes in θd were consistently detected but showed only a moderate relationship with DTF, and in approximately half of the recordings the dominant frequency of θd and of DTF was similar. This cautiously suggests that angular motion captures complementary mechanical information about diaphragm behavior. Both DTF- and θd-derived parameters exhibited substantial inter-individual variability, while no significant differences were observed across controlled conditions with varying levels of breathing effort in healthy volunteers.
In conclusion, this thesis demonstrates that automated ultrasound-based quantification of diaphragm thickening is feasible using a Radon-based approach and outlines a physiological study to further explore the relevance and interpretation of diaphragm ultrasound and sEMG monitoring. Parameters derived from diaphragmatic angular motion represent a novel measure that warrants further investigation to determine potential clinical relevance. By improving tracking robustness and signal quality compared with conventional methods, this work provides a methodological foundation for future development of continuous, operator-independent, and multimodal respiratory monitoring systems to support personalized ventilation strategies.
...
The primary objective of this thesis was to evaluate the feasibility of continuous ultrasound-based diaphragm monitoring and to develop automated methods for extracting clinically relevant diaphragm parameters, with a specific focus on diaphragm thickening fraction (DTF). This thesis consists of two parts.
In Part I, a physiological study design was developed to enable synchronized acquisition of diaphragm ultrasound, sEMG, esophageal pressure, ventilator data, and gas-exchange parameters in healthy volunteers.
In Part II, a semi-automated algorithm for diaphragm parameter extraction was developed using a Radon transform-based adaptive M-mode approach. This method explicitly accounts for changes in diaphragm orientation and enables robust tracking of both pleural and peritoneal interfaces. When applied to 110 ultrasound recordings from ten healthy volunteers, the Radon-based method showed markedly improved line-tracking performance and superior image quality compared with a conventional static M-mode approach. These improvements were reflected by significantly higher contrast and smoother, more traceable signals in both spatial and intensity domains. Automated DTF measurements showed a moderate correlation between the left and right hemidiaphragms. Absolute diaphragm thickness and DTF values were within the lower range of normal values reported in the literature, supporting the need for further evaluation and potential redefinition of reference values when using automated Radon-based tracking methods.
In addition to diaphragm thickening, angular motion of the diaphragm was quantified through analysis of the diaphragmatic angle (θd). Cyclic changes in θd were consistently detected but showed only a moderate relationship with DTF, and in approximately half of the recordings the dominant frequency of θd and of DTF was similar. This cautiously suggests that angular motion captures complementary mechanical information about diaphragm behavior. Both DTF- and θd-derived parameters exhibited substantial inter-individual variability, while no significant differences were observed across controlled conditions with varying levels of breathing effort in healthy volunteers.
In conclusion, this thesis demonstrates that automated ultrasound-based quantification of diaphragm thickening is feasible using a Radon-based approach and outlines a physiological study to further explore the relevance and interpretation of diaphragm ultrasound and sEMG monitoring. Parameters derived from diaphragmatic angular motion represent a novel measure that warrants further investigation to determine potential clinical relevance. By improving tracking robustness and signal quality compared with conventional methods, this work provides a methodological foundation for future development of continuous, operator-independent, and multimodal respiratory monitoring systems to support personalized ventilation strategies.
...
Mechanical ventilation (MV) in the intensive care unit is inherently associated with both life-saving benefits and the risk of lung and diaphragm injury, making the timing of ventilator detachment a well-studied clinical challenge. This creates a need for reliable monitoring of respiratory effort to tailor MV to individual patient requirements. Diaphragm ultrasound is a non-invasive technique with respiratory monitoring potential. However, its application is limited by operator dependency and the lack of automated, continuous analysis. To help overcome these limitations, the Sonoskin project aims to develop a wearable and operator-independent ultrasound-based respiratory monitoring system, integrating diaphragm ultrasound with surface electromyography (sEMG) as a multimodal monitoring approach. The European consortium Sonoskin forms the framework of this thesis.
The primary objective of this thesis was to evaluate the feasibility of continuous ultrasound-based diaphragm monitoring and to develop automated methods for extracting clinically relevant diaphragm parameters, with a specific focus on diaphragm thickening fraction (DTF). This thesis consists of two parts.
In Part I, a physiological study design was developed to enable synchronized acquisition of diaphragm ultrasound, sEMG, esophageal pressure, ventilator data, and gas-exchange parameters in healthy volunteers.
In Part II, a semi-automated algorithm for diaphragm parameter extraction was developed using a Radon transform-based adaptive M-mode approach. This method explicitly accounts for changes in diaphragm orientation and enables robust tracking of both pleural and peritoneal interfaces. When applied to 110 ultrasound recordings from ten healthy volunteers, the Radon-based method showed markedly improved line-tracking performance and superior image quality compared with a conventional static M-mode approach. These improvements were reflected by significantly higher contrast and smoother, more traceable signals in both spatial and intensity domains. Automated DTF measurements showed a moderate correlation between the left and right hemidiaphragms. Absolute diaphragm thickness and DTF values were within the lower range of normal values reported in the literature, supporting the need for further evaluation and potential redefinition of reference values when using automated Radon-based tracking methods.
In addition to diaphragm thickening, angular motion of the diaphragm was quantified through analysis of the diaphragmatic angle (θd). Cyclic changes in θd were consistently detected but showed only a moderate relationship with DTF, and in approximately half of the recordings the dominant frequency of θd and of DTF was similar. This cautiously suggests that angular motion captures complementary mechanical information about diaphragm behavior. Both DTF- and θd-derived parameters exhibited substantial inter-individual variability, while no significant differences were observed across controlled conditions with varying levels of breathing effort in healthy volunteers.
In conclusion, this thesis demonstrates that automated ultrasound-based quantification of diaphragm thickening is feasible using a Radon-based approach and outlines a physiological study to further explore the relevance and interpretation of diaphragm ultrasound and sEMG monitoring. Parameters derived from diaphragmatic angular motion represent a novel measure that warrants further investigation to determine potential clinical relevance. By improving tracking robustness and signal quality compared with conventional methods, this work provides a methodological foundation for future development of continuous, operator-independent, and multimodal respiratory monitoring systems to support personalized ventilation strategies.
The primary objective of this thesis was to evaluate the feasibility of continuous ultrasound-based diaphragm monitoring and to develop automated methods for extracting clinically relevant diaphragm parameters, with a specific focus on diaphragm thickening fraction (DTF). This thesis consists of two parts.
In Part I, a physiological study design was developed to enable synchronized acquisition of diaphragm ultrasound, sEMG, esophageal pressure, ventilator data, and gas-exchange parameters in healthy volunteers.
In Part II, a semi-automated algorithm for diaphragm parameter extraction was developed using a Radon transform-based adaptive M-mode approach. This method explicitly accounts for changes in diaphragm orientation and enables robust tracking of both pleural and peritoneal interfaces. When applied to 110 ultrasound recordings from ten healthy volunteers, the Radon-based method showed markedly improved line-tracking performance and superior image quality compared with a conventional static M-mode approach. These improvements were reflected by significantly higher contrast and smoother, more traceable signals in both spatial and intensity domains. Automated DTF measurements showed a moderate correlation between the left and right hemidiaphragms. Absolute diaphragm thickness and DTF values were within the lower range of normal values reported in the literature, supporting the need for further evaluation and potential redefinition of reference values when using automated Radon-based tracking methods.
In addition to diaphragm thickening, angular motion of the diaphragm was quantified through analysis of the diaphragmatic angle (θd). Cyclic changes in θd were consistently detected but showed only a moderate relationship with DTF, and in approximately half of the recordings the dominant frequency of θd and of DTF was similar. This cautiously suggests that angular motion captures complementary mechanical information about diaphragm behavior. Both DTF- and θd-derived parameters exhibited substantial inter-individual variability, while no significant differences were observed across controlled conditions with varying levels of breathing effort in healthy volunteers.
In conclusion, this thesis demonstrates that automated ultrasound-based quantification of diaphragm thickening is feasible using a Radon-based approach and outlines a physiological study to further explore the relevance and interpretation of diaphragm ultrasound and sEMG monitoring. Parameters derived from diaphragmatic angular motion represent a novel measure that warrants further investigation to determine potential clinical relevance. By improving tracking robustness and signal quality compared with conventional methods, this work provides a methodological foundation for future development of continuous, operator-independent, and multimodal respiratory monitoring systems to support personalized ventilation strategies.