Derivation and Validation of an Ensemble Model for the Prediction of Agitation in Mechanically Ventilated Patients Maintained Under Light Sedation

Predicting Agitation-Sedation Levels in Intensive Care Unit Patients: Development of an Ensemble Model

Background

Agitation and sedation management is critical in intensive care as it affects patient safety. Traditional nursing assessments suffer from low frequency and subjectivity. Automating these assessments can boost intensive care unit (ICU) efficiency, treatment capacity, and patient safety.

objectives

The aim of this study was to develop a machine-learning based assessment of agitation and sedation.

Methods

Using data from the Taichung Veterans General Hospital ICU database (2020), an ensemble learning model was developed for classifying the levels of agitation and sedation. Different ensemble learning model sequences were compared. In addition, an interpretable artificial intelligence approach, SHAP (Shapley additive explanations), was employed for explanatory analysis.

Results

With 20 features and 121,303 data points, the random forest model achieved high area under the curve values across all models (sedation classification: 0.97; agitation classification: 0.88). The ensemble learning model enhanced agitation sensitivity (0.82) while maintaining high AUC values across all categories (all >0.82). The model explanations aligned with clinical experience.

Conclusions

This study proposes an ICU agitation-sedation assessment automation using machine learning, enhancing efficiency and safety. Ensemble learning improves agitation sensitivity while maintaining accuracy. Real-time monitoring and future digital integration have the potential for advancements in intensive care.

Dynamic mortality prediction in critically Ill children during interhospital transports to PICUs using explainable AI

Critically ill children who require inter-hospital transfers to paediatric intensive care units are sicker than other admissions and have higher mortality rates. Current transport practice primarily relies on early clinical assessments within the initial hours of transport. Real-time mortality risk during transport is lacking due to the absence of data-driven assessment tools. Addressing this gap, our research introduces the PROMPT (Patient-centred Real-time Outcome monitoring and Mortality PredicTion), an explainable end-to-end machine learning pipeline to forecast 30-day mortality risks. The PROMPT integrates continuous time-series vital signs and medical records with episode-specific transport data to provide real-time mortality prediction. The results demonstrated that with PROMPT, both the random forest and logistic regression models achieved the best performance with AUROC 0.83 (95% CI: 0.79-0.86) and 0.81 (95% CI: 0.76-0.85), respectively. The proposed model has demonstrated proof-of-principle in predicting mortality risk in transported children and providing individual-level model interpretability during inter-hospital transports.

Improving cardiovascular risk prediction with machine learning: a focus on perivascular adipose tissue characteristics

BACKGROUND

Timely prevention of major adverse cardiovascular events (MACEs) is imperative for reducing cardiovascular diseases-related mortality. Perivascular adipose tissue (PVAT), the adipose tissue surrounding coronary arteries, has attracted increased amounts of attention. Developing a model for predicting the incidence of MACE utilizing machine learning (ML) integrating clinical and PVAT features may facilitate targeted preventive interventions and improve patient outcomes.

METHODS

From January 2017 to December 2019, we analyzed a cohort of 1077 individuals who underwent coronary CT scanning at our facility. Clinical features were collected alongside imaging features, such as coronary artery calcium (CAC) scores and perivascular adipose tissue (PVAT) characteristics. Logistic regression (LR), Framingham Risk Score, and ML algorithms were employed for MACE prediction.

RESULTS

We screened seven critical features to improve the practicability of the model. MACE patients tended to be older, smokers, and hypertensive. Imaging biomarkers such as CAC scores and PVAT characteristics differed significantly between patients with and without a 3-year MACE risk in a population that did not exhibit disparities in laboratory results. The ensemble model, which leverages multiple ML algorithms, demonstrated superior predictive performance compared with the other models. Finally, the ensemble model was used for risk stratification prediction to explore its clinical application value.

CONCLUSIONS

The developed ensemble model effectively predicted MACE incidence based on clinical and imaging features, highlighting the potential of ML algorithms in cardiovascular risk prediction and personalized medicine. Early identification of high-risk patients may facilitate targeted preventive interventions and improve patient outcomes.

Acupuncture reduces mechanical ventilation time in critically ill patients: A systematic review and meta-analysis of randomized control trials

BACKGROUND

Mechanical Ventilation (MV) is an essential life support machine, frequently utilized in an Intensive Care Unit (ICU). Recently, a growing number of clinical trials have investigated the effect of acupuncture treatment on MV outcomes.

OBJECTIVES

This study investigated the safety and efficacy of acupuncture treatment for critically ill patients under MV.

METHODS

In this systematic review and meta-analysis of randomized controlled trials, the efficacy of acupuncture related interventions was compared to routine ICU treatments, and sham/control acupuncture as control interventions applied to ICU patients undergoing MV. The databases of PubMed, Cochrane Library, and Web of Science were extensively searched in the month of April 2022. The primary outcome measurements were defined as total MV time, ICU length of stay, and mortality. The Cochrane Collaboration risk of bias tool was employed to analyze the severity of bias. The meta-analysis was conducted using Review Manager 5.3 software. The quality of evidence was evaluated according to the GRADE approach.

RESULTS

A total of 10 clinical trials were included in this investigation. When comparing the performance of acupuncture-related interventions to that of the reported control interventions, the results of the meta-analysis revealed a significant reduction in the total number of MV days as well as the duration of ICU length of stay following acupuncture treatment (MD -2.06 [-3.33, -0.79] P = 0.001, I2 = 55 %, MD-1.26 [-2.00, -0.53] P = 0.0008, I2 = 77 %, respectively). A reduction in the total mortality was similarly observed (RR = 0.67 [0.47, 0.94] P = 0.02, I2 = 0 %).

CONCLUSION

This systematic review and meta-analysis identified a noteworthy reduction in the total MV days, time spent in the ICU, as well as the total mortality following acupuncture related interventions. However, the small sample size, risk of bias and existing heterogeneity should be taken into consideration. The results of this study are promising and further investigations in this field are warranted.

Comparison of logistic regression and machine learning methods for predicting postoperative delirium in elderly patients: A retrospective study

AIMS

To compare the performance of logistic regression and machine learning methods in predicting postoperative delirium (POD) in elderly patients.

METHOD

This was a retrospective study of perioperative medical data from patients undergoing non-cardiac and non-neurology surgery over 65 years old from January 2014 to August 2019. Forty-six perioperative variables were used to predict POD. A traditional logistic regression and five machine learning models (Random Forest, GBM, AdaBoost, XGBoost, and a stacking ensemble model) were compared by the area under the receiver operating characteristic curve (AUC-ROC), sensitivity, specificity, and precision.

RESULTS

In total, 29,756 patients were enrolled, and the incidence of POD was 3.22% after variable screening. AUCs were 0.783 (0.765-0.8) for the logistic regression method, 0.78 for random forest, 0.76 for GBM, 0.74 for AdaBoost, 0.73 for XGBoost, and 0.77 for the stacking ensemble model. The respective sensitivities for the 6 aforementioned models were 74.2%, 72.2%, 76.8%, 63.6%, 71.6%, and 67.4%. The respective specificities for the 6 aforementioned models were 70.7%, 99.8%, 96.5%, 98.8%, 96.5%, and 96.1%. The respective precision values for the 6 aforementioned models were 7.8%, 52.3%, 55.6%, 57%, 54.5%, and 56.4%.

CONCLUSIONS

The optimal application of the logistic regression model could provide quick and convenient POD risk identification to help improve the perioperative management of surgical patients because of its better sensitivity, fewer variables, and easier interpretability than the machine learning model.

Developing an ensemble machine learning model for early prediction of sepsis-associated acute kidney injury

Sepsis-associated acute kidney injury (S-AKI) is very common and early prediction is beneficial. This study aiming to develop an accurate ensemble model to predict the risk of S-AKI based on easily available clinical information. Patients with sepsis from the United States (US) database Medical Information Mart for Intensive Care-IV were used as a modeling cohort to predict the occurrence of AKI by combining Support Vector Machine, Random Forest, Neural Network, and Extreme Gradient Boost as four first-level learners via stacking algorithm. The external validation databases were the eICU Collaborative Research Database from US and Critical Care Database comprising infection patients at Zigong Fourth People's Hospital from China, whose AUROC values for the ensemble model 48-12 h before the onset of AKI were 0.774-0.788 and 0.756-0.813, respectively. In this study, an ensemble model for early prediction of S-AKI onset was developed and it demonstrated good performance in multicenter external datasets.

Comparison and Clinical Value of Ciprofol and Propofol in Intraoperative Adverse Reactions, Operation, Resuscitation, and Satisfaction of Patients under Painless Gastroenteroscopy Anesthesia

Objective

To investigate the comparison and clinical value of ciprofol and propofol for painless gastroenteroscopy anesthesia in terms of intraoperative adverse reactions, operation, resuscitation, and satisfaction of patients.

Methods

A total of 96 patients who underwent painless gastroenteroscopy anesthesia in our hospital from June 2021 to January 2022 were enrolled. The cases were randomly assigned into research group and control group. The control group received propofol anesthesia (n = 49), and the research group received ciprofol anesthesia (n = 47). The patients, physician satisfaction, vital signs, incidence of adverse reactions, anesthetic first dose, additional time, additional dose, total dose, induction time, insertion time, operation time, awake time, orientation recovery time, leaving room time, and injection pain score were compared.

Results

The overall satisfaction of the study group was higher than that of the control group (p < 0.05). After taking medicine, the score of 1 min and MAP in the study group were higher than those in the control group. The incidence of adverse reactions in the study group was lower than that in the control group (p < 0.05). The satisfaction of doctors in the study group was higher than that in the control group (p < 0.05). The anesthesia induction time, intubation time, operation time, awake time, orientation recovery time, and leaving room time in the study group were significantly longer than those in the control group (p < 0.05). The incidence and degree of injection pain in the propofol group were significantly lower than those in the propofol group (p < 0.05).

Conclusion

In painless gastroenteroscopy, compared with propofol, ciprofol is equally safe and effective for patients and will not cause early cognitive dysfunction after operation, which is a good choice in painless gastroenteroscopy anesthesia. In addition, ciprofol has significant advantages in patient and physician satisfaction, especially in injection pain. This trial is registered with ChiCTR2100045400.

Development of artificial neural networks for early prediction of intestinal perforation in preterm infants

Intestinal perforation (IP) in preterm infants is a life-threatening condition that may result in serious complications and increased mortality. Early Prediction of IP in infants is important, but challenging due to its multifactorial and complex nature of the disease. Thus, there are no reliable tools to predict IP in infants. In this study, we developed new machine learning (ML) models for predicting IP in very low birth weight (VLBW) infants and compared their performance to that of classic ML methods. We developed artificial neural networks (ANNs) using VLBW infant data from a nationwide cohort and prospective web-based registry. The new ANN models, which outperformed all other classic ML methods, showed an area under the receiver operating characteristic curve (AUROC) of 0.8832 for predicting IP associated with necrotizing enterocolitis (NEC-IP) and 0.8797 for spontaneous IP (SIP). We tested these algorithms using patient data from our institution, which were not included in the training dataset, and obtained an AUROC of 1.0000 for NEC-IP and 0.9364 for SIP. NEC-IP and SIP in VLBW infants can be predicted at an excellent performance level with these newly developed ML models. https://github.com/kdhRick2222/Early-Prediction-of-Intestinal-Perforation-in-Preterm-Infants.

Factors Associated With Deep Sedation Practice in Mechanically Ventilated Patients: A Post hoc Analysis of a Cross-Sectional Survey Combined With a Questionnaire for Physicians on Sedation Practices

Purpose

The study aimed to explore factors associated with deep sedation practice in intensive care units (ICUs).

Materials and Methods

A post hoc analysis was conducted for a cross–sectional survey on sedation practices in mechanically ventilated (MV) patients, combined with a questionnaire for physicians regarding their preferences for light sedation (P–pls Score) in 92 Chinese ICUs.

Results

There were 457 and 127 eligible MV patients in the light and deep sedation groups respectively. A multivariable logistic regression analysis demonstrated that the control mode of mechanical ventilation, plasma lactate level, and the Sequential Organ Failure Assessment (SOFA) score were independent risk factors for deep sedation practice (p <0.01). Notably, the adjusted odds ratio (95% CI) of the average P–pls score in the ICU ≤ 2 for deep sedation practice was 1.861 (1.163, 2.978, p = 0.01). In addition, the areas under curves of receiver operating characteristics (AUC–ROC) of the model to predict the probability of deep sedation practice were 0.753 (0.699, 0.806) and 0.772 (0.64, 0.905) in the training set and the validation set, respectively. The 28–day mortality was increased in patients with exposure to deep sedation practice but not significantly.

Conclusion

Both factors related to stressful stimuli and the ICU physicians' perception of patient tolerability in mechanical ventilation were likely associated with deep sedation practice in MV patients.

Considerations for the implementation of machine learning into acute care settings

INTRODUCTION

Management of patients in the acute care setting requires accurate diagnosis and rapid initiation of validated treatments; therefore, this setting is likely to be an environment in which cognitive augmentation of the clinician's provision of care with technology rooted in artificial intelligence, such as machine learning (ML), is likely to eventuate.

SOURCES OF DATA

PubMed and Google Scholar with search terms that included ML, intensive/critical care unit, electronic health records (EHR), anesthesia information management systems and clinical decision support were the primary sources for this report.

AREAS OF AGREEMENT

Different categories of learning of large clinical datasets, often contained in EHRs, are used for training in ML. Supervised learning uses algorithm-based models, including support vector machines, to pair patients' attributes with an expected outcome. Unsupervised learning uses clustering algorithms to define to which disease grouping a patient's attributes most closely approximates. Reinforcement learning algorithms use ongoing environmental feedback to deterministically pursue likely patient outcome.

AREAS OF CONTROVERSY

Application of ML can result in undesirable outcomes over concerns related to fairness, transparency, privacy and accountability. Whether these ML technologies irrevocably change the healthcare workforce remains unresolved.

GROWING POINTS

Well-resourced Learning Health Systems are likely to exploit ML technology to gain the fullest benefits for their patients. How these clinical advantages can be extended to patients in health systems that are neither well-endowed, nor have the necessary data gathering technologies, needs to be urgently addressed to avoid further disparities in healthcare.

Accuracy of approximations to recover incompletely reported logistic regression models depended on other available information

OBJECTIVE

To provide approximations to recover the full regression equation across different scenarios of incompletely reported prediction models that were developed from binary logistic regression.

STUDY DESIGN AND SETTING

In a case study, we considered four common scenarios and illustrated their corresponding approximations: (A) Missing: the intercept, Available: the regression coefficients of predictors, overall frequency of the outcome and descriptive statistics of the predictors; (B) Missing: regression coefficients and the intercept, Available: a simplified score; (C) Missing: regression coefficients and the intercept, Available: a nomogram; (D) Missing: regression coefficients and the intercept, Available: a web calculator.

RESULTS

In the scenario A, a simplified approach based on the predicted probability corresponding to the average linear predictor was inaccurate. An approximation based on the overall outcome frequency and an approximation of the linear predictor distribution was more accurate, however, the appropriateness of the underlying assumptions cannot be verified in practice. In the scenario B, the recovered equation was inaccurate due to rounding and categorization of risk scores. In the scenarios C and D, the full regression equation could be recovered with minimal error.

CONCLUSION

The accuracy of the approximations in recovering the regression equation varied depending on the available information.

Evaluation of the Need for Intensive Care in Children With Pneumonia: Machine Learning Approach

Background

Timely decision-making regarding intensive care unit (ICU) admission for children with pneumonia is crucial for a better prognosis. Despite attempts to establish a guideline or triage system for evaluating ICU care needs, no clinically applicable paradigm is available.

Objective

The aim of this study was to develop machine learning (ML) algorithms to predict ICU care needs for pediatric pneumonia patients within 24 hours of admission, evaluate their performance, and identify clinical indices for making decisions for pediatric pneumonia patients.

Methods

Pneumonia patients admitted to National Taiwan University Hospital from January 2010 to December 2019 aged under 18 years were enrolled. Their underlying diseases, clinical manifestations, and laboratory data at admission were collected. The outcome of interest was ICU transfer within 24 hours of hospitalization. We compared clinically relevant features between early ICU transfer patients and patients without ICU care. ML algorithms were developed to predict ICU admission. The performance of the algorithms was evaluated using sensitivity, specificity, area under the receiver operating characteristic curve (AUC), and average precision. The relative feature importance of the best-performing algorithm was compared with physician-rated feature importance for explainability.

Results

A total of 8464 pediatric hospitalizations due to pneumonia were recorded, and 1166 (1166/8464, 13.8%) hospitalized patients were transferred to the ICU within 24 hours. Early ICU transfer patients were younger (P<.001), had higher rates of underlying diseases (eg, cardiovascular, neuropsychological, and congenital anomaly/genetic disorders; P<.001), had abnormal laboratory data, had higher pulse rates (P<.001), had higher breath rates (P<.001), had lower oxygen saturation (P<.001), and had lower peak body temperature (P<.001) at admission than patients without ICU transfer. The random forest (RF) algorithm achieved the best performance (sensitivity 0.94, 95% CI 0.92-0.95; specificity 0.94, 95% CI 0.92-0.95; AUC 0.99, 95% CI 0.98-0.99; and average precision 0.93, 95% CI 0.90-0.96). The lowest systolic blood pressure and presence of cardiovascular and neuropsychological diseases ranked in the top 10 in both RF relative feature importance and clinician judgment.

Conclusions

The ML approach could provide a clinically applicable triage algorithm and identify important clinical indices, such as age, underlying diseases, abnormal vital signs, and laboratory data for evaluating the need for intensive care in children with pneumonia.

Added value of systemic inflammation markers in predicting pulmonary infection in stroke patients: A retrospective study by machine learning analysis

Exploring candidate markers to predict the clinical outcomes of pulmonary infection in stroke patients have a high unmet need. This study aimed to develop machine learning (ML)-based predictive models for pulmonary infection.Between January 2008 and April 2021, a retrospective analysis of 1397 stroke patients who had CT angiography from skull to diaphragm (including CT of the chest) within 24 hours of symptom onset. A total of 21 variables were included, and the prediction model of pulmonary infection was established by multiple ML-based algorithms. Risk factors for pulmonary infection were determined by the feature selection method. Area under the curve (AUC) and decision curve analysis were used to determine the model with the best resolution and to assess the net clinical benefits associated with the use of predictive models, respectively.A total of 889 cases were included in this study as a training group, while 508 cases were as a validation group. The feature selection indicated the top 6 predictors were procalcitonin, C-reactive protein, soluble interleukin-2 receptor, consciousness disorder, dysphagia, and invasive procedure. The AUCs of the 5 models ranged from 0.78 to 0.87 in the training cohort. When the ML-based models were applied to the validation set, the results also remained reconcilable, and the AUC was between 0.891 and 0.804. The decision curve analysis also showed performed better than positive line and negative line, indicating the favorable predictive performance and clinical values of the models.By incorporating clinical characteristics and systemic inflammation markers, it is feasible to develop ML-based models for the presence and consequences of signs of pulmonary infection in stroke patients, and the use of the model may be greatly beneficial to clinicians in risk stratification and management decisions.

Predicting Central Serous Chorioretinopathy Recurrence Using Machine Learning

Purpose: To predict central serous chorioretinopathy (CSC) recurrence 3 and 6 months after laser treatment by using machine learning.

Methods: Clinical and imaging features of 461 patients (480 eyes) with CSC were collected at Zhongshan Ophthalmic Center (ZOC) and Xiamen Eye Center (XEC). The ZOC data (416 eyes of 401 patients) were used as the training dataset and the internal test dataset, while the XEC data (64 eyes of 60 patients) were used as the external test dataset. Six different machine learning algorithms and an ensemble model were trained to predict recurrence in patients with CSC. After completing the initial detailed investigation, we designed a simplified model using only clinical data and OCT features.

Results: The ensemble model exhibited the best performance among the six algorithms, with accuracies of 0.941 (internal test dataset) and 0.970 (external test dataset) at 3 months and 0.903 (internal test dataset) and 1.000 (external test dataset) at 6 months. The simplified model showed a comparable level of predictive power.

Conclusion: Machine learning achieves high accuracies in predicting the recurrence of CSC patients. The application of an intelligent recurrence prediction model for patients with CSC can potentially facilitate recurrence factor identification and precise individualized interventions.

Artificial intelligence model comparison for risk factor analysis of patent ductus arteriosus in nationwide very low birth weight infants cohort

Despite the many comorbidities and high mortality rate in preterm infants with patent ductus arteriosus (PDA), therapeutic strategies vary depending on the clinical setting, and most studies of the related risk factors are based on small sample populations. We aimed to compare the performance of artificial intelligence (AI) analysis with that of conventional analysis to identify risk factors associated with symptomatic PDA (sPDA) in very low birth weight infants. This nationwide cohort study included 8369 very low birth weight (VLBW) infants. The participants were divided into an sPDA group and an asymptomatic PDA or spontaneously close PDA (nPDA) group. The sPDA group was further divided into treated and untreated subgroups. A total of 47 perinatal risk factors were collected and analyzed. Multiple logistic regression was used as a standard analytic tool, and five AI algorithms were used to identify the factors associated with sPDA. Combining a large database of risk factors from nationwide registries and AI techniques achieved higher accuracy and better performance of the PDA prediction tasks, and the ensemble methods showed the best performances.

Tools Are Needed to Promote Sedation Practices for Mechanically Ventilated Patients

Suboptimal sedation practices continue to be frequent, although the updated guidelines for management of pain, agitation, and delirium in mechanically ventilated (MV) patients have been published for several years. Causes of low adherence to the recommended minimal sedation protocol are multifactorial. However, the barriers to translation of these protocols into standard care for MV patients have yet to be analyzed. In our view, it is necessary to develop fresh insights into the interaction between the patients' responses to nociceptive stimuli and individualized regulation of patients' tolerance when using analgesics and sedatives. By better understanding this interaction, development of novel tools to assess patient pain tolerance and to define and predict oversedation or delirium may promote better sedation practices in the future.

Factors associated with delirium among survivors of acute respiratory distress syndrome: a nationwide cohort study

Background

The prevalence of delirium, its associated factors, and its impact on long-term mortality among survivors of acute respiratory distress syndrome (ARDS) is unclear.

Methods

Since this was a population-based study, data were extracted from the National Health Insurance database in South Korea. All adults who were admitted to intensive care units with a diagnosis of ARDS between January 1, 2010, and December 31, 2019, and who survived for ≥ 60 days were included. The International Statistical Classification of Diseases and Related Health Problems, tenth revision code of delirium (F05) was used to extract delirium cases during hospitalization.

Results

A total of 6809 ARDS survivors were included in the analysis, and 319 patients (4.7%) were diagnosed with delirium during hospitalization. In the multivariable logistic regression analysis after covariate adjustment, male sex (odds ratio [OR] 1.60, 95% confidence interval [CI] 1.23, 2.08; P < 0.001), longer duration of hospitalization (OR 1.02, 95% CI 1.01, 1.03; P < 0.001), neuromuscular blockade use (OR 1.50, 95% CI 1.12, 2.01; P = 0.006), benzodiazepine (OR 1.55, 95% CI 1.13, 2.13; P = 0.007) and propofol (OR 1.48, 95% CI 1.01, 2.17; P = 0.046) continuous infusion, and concurrent depression (OR 1.31, 95% CI 1.01, 1.71; P = 0.044) were associated with a higher prevalence of delirium among ARDS survivors. In the multivariable Cox regression analysis after adjustment for covariates, the occurrence of delirium was not significantly associated with 1-year all-cause mortality, when compared to the other survivors who did not develop delirium (hazard ratio: 0.85, 95% CI 1.01, 1.71; P = 0.044).

Conclusions

In South Korea, 4.7% of ARDS survivors were diagnosed with delirium during hospitalization in South Korea. Some factors were potential risk factors for the development of delirium, but the occurrence of delirium might not affect 1-year all-cause mortality among ARDS survivors.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12890-021-01714-0.

Machine learning using the extreme gradient boosting (XGBoost) algorithm predicts 5-day delta of SOFA score at ICU admission in COVID-19 patients

Background

Accurate risk stratification of critically ill patients with coronavirus disease 2019 (COVID-19) is essential for optimizing resource allocation, delivering targeted interventions, and maximizing patient survival probability. Machine learning (ML) techniques are attracting increased interest for the development of prediction models as they excel in the analysis of complex signals in data-rich environments such as critical care.

Methods

We retrieved data on patients with COVID-19 admitted to an intensive care unit (ICU) between March and October 2020 from the RIsk Stratification in COVID-19 patients in the Intensive Care Unit (RISC-19-ICU) registry. We applied the Extreme Gradient Boosting (XGBoost) algorithm to the data to predict as a binary outcome the increase or decrease in patients' Sequential Organ Failure Assessment (SOFA) score on day 5 after ICU admission. The model was iteratively cross-validated in different subsets of the study cohort.

Results

The final study population consisted of 675 patients. The XGBoost model correctly predicted a decrease in SOFA score in 320/385 (83%) critically ill COVID-19 patients, and an increase in the score in 210/290 (72%) patients. The area under the mean receiver operating characteristic curve for XGBoost was significantly higher than that for the logistic regression model (0.86 vs. 0.69, P < 0.01 [paired t-test with 95% confidence interval]).

Conclusions

The XGBoost model predicted the change in SOFA score in critically ill COVID-19 patients admitted to the ICU and can guide clinical decision support systems (CDSSs) aimed at optimizing available resources.

Validation of the National Early Warning Score (NEWS)-2 for adults in the emergency department in a tertiary-level clinic in Colombia: Cohort study

The National Early Warning Score (NEWS)-2 is an early warning scale that is used in emergency departments to identify patients at risk of clinical deterioration and to help establish rapid and timely management. The objective of this study was to determine the validity and prediction of mortality using the NEWS2 scale for adults in the emergency department of a tertiary clinic in Colombia.A prospective observational study was conducted between August 2018 and June 2019 at the Universidad de La Sabana Clinic.The nursing staff in the triage classified the patients admitted to the emergency room according to Emergency Severity Index and NEWS2. Demographic data, physiological variables, admission diagnosis, mortality outcome, and comorbidities were extracted.Three thousand nine hundred eighty-six patients were included in the study. Ninety-two (2%) patients required intensive care unit management, with a mean NEWS2 score of 7. A total of 158 patients died in hospital, of which 63 were women (40%). Of these 65 patients required intensive care unit management. The receiver operating characteristic curve for NEWS2 had an area of 0.90 (CI 95%: 0.87-0.92). A classification and score equivalency analysis was performed between triage and the NEWS2 scale in terms of mortality. Of the patients classified as triage I, 32.3% died, and those who obtained a NEWS2 score greater than or equal to 10 had a mortality of 38.6%.Among our population, NEWS2 was not inferior in its area under the receiver operating characteristic curve when predicting mortality than triage, and the cutoff point for NEWS2 to predict in-hospital mortality was higher.

A case of paroxysmal complete atrioventricular block in a COVID-19 patient

Many types of cardiac arrhythmias can occur in people with COVID-19, and these arrhythmias can affect the patient's outcomes. We have experienced paroxysmal complete atrioventricular block in a patient with COVID-19 and would like to share the course of treatment.

Efficacy and Safety of Ciprofol Sedation in ICU Patients with Mechanical Ventilation: A Clinical Trial Study Protocol

INTRODUCTION

From previous studies of pharmacodynamic data in mice, rats, beagle dogs and mini pigs, frequently in direct comparison to induction doses of propofol, ciprofol produced a rapid onset of anesthesia/sedation.

METHODS

A phase 1 study suggested potential clinical advantages of ciprofol as a sedation/anesthetic agent, with no evidence of drug-related toxicity. However, the sedation effects and safety of ciprofol in intensive care unit (ICU) patients with mechanical ventilation should be further confirmed in a phase 3 study with a larger cohort of patients. During a phase 3, non-inferiority, multicenter, single-blind, randomized, propofol controlled trial, Chinese ICU patients undergoing mechanical ventilation and requiring endotracheal intubation will be sedated for 6-24 h after randomization. Considering a success rate for ICU sedation of 99% for ciprofol and the positive control drug propofol, a total sample size of 120 subjects with mechanical ventilation will be required to achieve 80% power to determine non-inferiority with a margin of 8%. Finally, taking into account 10% losses, 135 patients will be enrolled and randomly assigned to ciprofol (90 cases) and propofol (45 cases) groups in a 2:1 ratio. The primary outcome will be the success rate of sedation satisfied by the following conditions: the time within the range of Richmond Agitation and Sedation Score (+ 1 ~ - 2) must account for ≥ 70% of the study drug administration time and without other rescue treatments. Secondary outcomes will include the average time to reach the sedation goal, study drug usage, rescue medication given per unit weight, extubation time, recovery time to full consciousness and nursing scores. Safety endpoints will include adverse events (AEs), drug related AEs and serious AEs.

PLANNED OUTCOMES

The results of this study will provide crucial information on the use of ciprofol for sedation of patients in ICUs.

TRIAL REGISTRATION

ClinicalTrials.gov identifier, NCT04620031.

Development and Validation of a Prediction Model for Elevated Arterial Stiffness in Chinese Patients With Diabetes Using Machine Learning

Background

Arterial stiffness assessed by pulse wave velocity is a major risk factor for cardiovascular diseases. The incidence of cardiovascular events remains high in diabetics. However, a clinical prediction model for elevated arterial stiffness using machine learning to identify subjects consequently at higher risk remains to be developed.

Methods

Least absolute shrinkage and selection operator and support vector machine-recursive feature elimination were used for feature selection. Four machine learning algorithms were used to construct a prediction model, and their performance was compared based on the area under the receiver operating characteristic curve metric in a discovery dataset (n = 760). The model with the best performance was selected and validated in an independent dataset (n = 912) from the Dryad Digital Repository (https://doi.org/10.5061/dryad.m484p). To apply our model to clinical practice, we built a free and user-friendly web online tool.

Results

The predictive model includes the predictors: age, systolic blood pressure, diastolic blood pressure, and body mass index. In the discovery cohort, the gradient boosting-based model outperformed other methods in the elevated arterial stiffness prediction. In the validation cohort, the gradient boosting model showed a good discrimination capacity. A cutoff value of 0.46 for the elevated arterial stiffness risk score in the gradient boosting model resulted in a good specificity (0.813 in the discovery data and 0.761 in the validation data) and sensitivity (0.875 and 0.738, respectively) trade-off points.

Conclusion

The gradient boosting-based prediction system presents a good classification in elevated arterial stiffness prediction. The web online tool makes our gradient boosting-based model easily accessible for further clinical studies and utilization.

Using Machine Learning Algorithms to Predict Candidaemia in ICU Patients With New-Onset Systemic Inflammatory Response Syndrome

Background: Distinguishing ICU patients with candidaemia can help with the precise prescription of antifungal drugs to create personalized guidelines. Previous prediction models of candidaemia have primarily used traditional logistic models and had some limitations. In this study, we developed a machine learning algorithm trained to predict candidaemia in patients with new-onset systemic inflammatory response syndrome (SIRS).

Methods: This retrospective, observational study used clinical information collected between January 2013 and December 2017 from three hospitals. The ICU patient data were used to train 4 machine learning algorithms–XGBoost, Support Vector Machine (SVM), Random Forest (RF), ExtraTrees (ET)–and a logistic regression (LR) model to predict patients with candidaemia.

Results: Of the 8,002 cases of new-onset SIRS (in 7,932 patients) included in the analysis, 137 new-onset SIRS cases (in 137 patients) were blood culture positive for candidaemia. Risk factors, such as fungal colonization, diabetes, acute kidney injury, the total number of parenteral nutrition days and renal replacement therapy, were important predictors of candidaemia. The XGBoost machine learning model outperformed the other models in distinguishing patients with candidaemia [XGBoost vs. SVM vs. RF vs. ET vs. LR; area under the curve (AUC): 0.92 vs. 0.86 vs. 0.91 vs. 0.90 vs. 0.52, respectively]. The XGBoost model had a sensitivity of 84%, specificity of 89% and negative predictive value of 99.6% at the best cut-off value.

Conclusions: Machine learning algorithms can potentially predict candidaemia in the ICU and have better efficiency than previous models. These prediction models can be used to guide antifungal treatment for ICU patients when SIRS occurs.

Ability of a Machine Learning Algorithm to Predict the Need for Perioperative Red Blood Cells Transfusion in Pelvic Fracture Patients: A Multicenter Cohort Study in China

Background: Predicting the perioperative requirement for red blood cells (RBCs) transfusion in patients with the pelvic fracture may be challenging. In this study, we constructed a perioperative RBCs transfusion predictive model (ternary classifications) based on a machine learning algorithm.

Materials and Methods: This study included perioperative adult patients with pelvic trauma hospitalized across six Chinese centers between September 2012 and June 2019. An extreme gradient boosting (XGBoost) algorithm was used to predict the need for perioperative RBCs transfusion, with data being split into training test (80%), which was subjected to 5-fold cross-validation, and test set (20%). The ability of the predictive transfusion model was compared with blood preparation based on surgeons' experience and other predictive models, including random forest, gradient boosting decision tree, K-nearest neighbor, logistic regression, and Gaussian naïve Bayes classifier models. Data of 33 patients from one of the hospitals were prospectively collected for model validation.

Results: Among 510 patients, 192 (37.65%) have not received any perioperative RBCs transfusion, 127 (24.90%) received less-transfusion (RBCs < 4U), and 191 (37.45%) received more-transfusion (RBCs ≥ 4U). Machine learning-based transfusion predictive model produced the best performance with the accuracy of 83.34%, and Kappa coefficient of 0.7967 compared with other methods (blood preparation based on surgeons' experience with the accuracy of 65.94%, and Kappa coefficient of 0.5704; the random forest method with an accuracy of 82.35%, and Kappa coefficient of 0.7858; the gradient boosting decision tree with an accuracy of 79.41%, and Kappa coefficient of 0.7742; the K-nearest neighbor with an accuracy of 53.92%, and Kappa coefficient of 0.3341). In the prospective dataset, it also had a food performance with accuracy 81.82%.

Conclusion: This multicenter retrospective cohort study described the construction of an accurate model that could predict perioperative RBCs transfusion in patients with pelvic fractures.

Risk Factors for Maternal and Fetal Mortality in Acute Fatty Liver of Pregnancy and New Predictive Models

Acute fatty liver of pregnancy (AFLP) is a rare but potentially life-threatening hepatic disorder that leads to considerable maternal and fetal mortality. To explore the risk factors for maternal and fetal mortality in AFLP and develop new predictive models, through this retrospective study, we analyzed the demographic characteristics, clinical symptoms, and laboratory findings of 106 patients with AFLP who were admitted to Shandong Provincial Hospital. Risk factors for maternal and fetal mortality were analyzed by univariate and multivariate logistic regression analysis. The new models based on the multivariate logistic regression analysis and the model for end-stage liver disease (MELD) were tested in AFLP. The receiver operating characteristic curve (ROC) was applied to compare the predictive efficiency, sensitivity, and specificity of the two models. Prenatal nausea (p = 0.037), prolonged prothrombin time (p = 0.003), and elevated serum creatinine (p = 0.003) were independent risk factors for maternal mortality. The ROC curve showed that the area under the curve (AUC) of the MELD was 0.948, with a sensitivity of 100% and a specificity of 83.3%. The AUC of the new model for maternal mortality was 0.926, with a sensitivity of 90% and a specificity of 94.8%. Hepatic encephalopathy (p = 0.016) and thrombocytopenia (p = 0.001) were independent risk factors for fetal mortality. Using the ROC curve, the AUC of the MELD was 0.694, yielding a sensitivity of 68.8% and a specificity of 64.4%. The AUC of the new model for fetal mortality was 0.893, yielding a sensitivity of 100% and a specificity of 73.3%. Both the new predictive model for maternal mortality and the MELD showed good predictive efficacy for maternal mortality in patients with AFLP (AUC = 0.926 and 0.948, respectively), and the new predictive model for fetal mortality was superior to the MELD in predicting fetal mortality (AUC = 0.893 and 0.694, respectively). The two new predictive models were more readily available, less expensive, and easier to implement clinically, especially in low-income countries.

Development and Validation of a Prediction Model for Elevated Arterial Stiffness in Chinese Patients With Diabetes Using Machine Learning

Background

Arterial stiffness assessed by pulse wave velocity is a major risk factor for cardiovascular diseases. The incidence of cardiovascular events remains high in diabetics. However, a clinical prediction model for elevated arterial stiffness using machine learning to identify subjects consequently at higher risk remains to be developed.

Methods

Least absolute shrinkage and selection operator and support vector machine-recursive feature elimination were used for feature selection. Four machine learning algorithms were used to construct a prediction model, and their performance was compared based on the area under the receiver operating characteristic curve metric in a discovery dataset (n = 760). The model with the best performance was selected and validated in an independent dataset (n = 912) from the Dryad Digital Repository (https://doi.org/10.5061/dryad.m484p). To apply our model to clinical practice, we built a free and user-friendly web online tool.

Results

The predictive model includes the predictors: age, systolic blood pressure, diastolic blood pressure, and body mass index. In the discovery cohort, the gradient boosting-based model outperformed other methods in the elevated arterial stiffness prediction. In the validation cohort, the gradient boosting model showed a good discrimination capacity. A cutoff value of 0.46 for the elevated arterial stiffness risk score in the gradient boosting model resulted in a good specificity (0.813 in the discovery data and 0.761 in the validation data) and sensitivity (0.875 and 0.738, respectively) trade-off points.

Conclusion

The gradient boosting-based prediction system presents a good classification in elevated arterial stiffness prediction. The web online tool makes our gradient boosting-based model easily accessible for further clinical studies and utilization.