Prolonged Elevated Heart Rate and 90-Day Survival in Acutely Ill Patients: Data From the MIMIC-III Database

Prolonged elevated heart rate and 90-Day mortality in acute pancreatitis

Prolonged elevated heart rate (peHR) is recognized as a risk factor for poor prognosis among critically ill patients. However, there is currently a lack of studies investigating the association between peHR and patients with acute pancreatitis. Multiparameter Intelligent Monitoring in Intensive Care IV (MIMIC-IV) database was used to identify patients with acute pancreatitis. PeHR was defined as a heart rate exceeding 100 beats per minute for at least 11 out of 12 consecutive hours. Cox regression analysis was used to assess the association between peHR and the 90-Day mortality. A total of 364 patients (48.9%) experienced a peHR episode. The 90-day mortality was 25%. PeHR is an independent risk factor for 90-day mortality (HR, 1.98; 95% CI 1.53-2.56; P < 0.001). KM survival curves exhibited a significant decrease in the survival rate at 90 days among patients who experienced a peHR episode (P < 0.001, 84.5% vs. 65.1%). We revealed a significant association of peHR with decreased survival in a large cohort of ICU patients with acute pancreatitis.

Predicting the need for intubation within 3 h in the neonatal intensive care unit using a multimodal deep neural network

Respiratory distress is a common chief complaint in neonates admitted to the neonatal intensive care unit. Despite the increasing use of non-invasive ventilation in neonates with respiratory difficulty, some of them require advanced airway support. Delayed intubation is associated with increased morbidity, particularly in urgent unplanned cases. Early and accurate prediction of the need for intubation may provide more time for preparation and increase safety margins by avoiding the late intubation at high-risk infants. This study aimed to predict the need for intubation within 3 h in neonates initially managed with non-invasive ventilation for respiratory distress during the first 48 h of life using a multimodal deep neural network. We developed a multimodal deep neural network model to simultaneously analyze four time-series data collected at 1-h intervals and 19 variables including demographic, physiological and laboratory parameters. Evaluating the dataset of 128 neonates with respiratory distress who underwent non-invasive ventilation, our model achieved an area under the curve of 0.917, sensitivity of 85.2%, and specificity of 89.2%. These findings demonstrate promising results for the multimodal model in predicting neonatal intubation within 3 h.

Effect of Esmolol on Clinical Outcomes in Critically Ill Patients: Data from the MIMIC-IV Database

BACKGROUND AND AIMS

Esmolol is a common short-acting drug to control ventricular rate. This study aimed to evaluate the association between use of esmolol and mortality in critically ill patients.

METHODS

This is a retrospective cohort study from MIMIC-IV database containing adult patients with a heart rate of over 100 beats/min during the intensive care unit (ICU) stay. Multivariable Cox proportional hazard models and logistic regression were used to explore the association between esmolol and mortality and adjust confounders. A 1:1 nearest neighbor propensity score matching (PSM) was performed to minimize potential cofounding bias. The comparison for secondary outcomes was performed at different points of time using an independent t-test.

RESULTS

A total of 30,332 patients were reviewed and identified as critically ill. There was no significant difference in 28-day mortality between two groups before (HR = 0.90, 95% CI = 0.73-1.12, p = 0.343) and after PSM (HR = 0.84, 95% CI = 0.65-1.08, p = 0.167). Similar results were shown in 90-day mortality before (HR = 0.93, 95% CI = 0.75-1.14, p = 0.484) and after PSM (HR = 0.85, 95% CI = 0.67-1.09, p = 0.193). However, esmolol treatment was associated with higher requirement of vasopressor use before (HR = 2.89, 95% CI = 2.18-3.82, p < 0.001) and after PSM (HR = 2.66, 95% CI = 2.06-3.45, p < 0.001). Esmolol treatment statistically reduced diastolic blood pressure (DBP), mean arterial pressure (MAP), and heart rate (all p < 0.001) and increased fluid balance at 24 hours (p < 0.05) but did not significantly lower SBP (p = 0.721). Patients in esmolol group showed no significant difference in lactate levels and daily urine output when compared with those in non-esmolol group when adjusted for confounders (all p > 0.05).

CONCLUSION

Esmolol treatment was associated with reduced heart rate and lowered DBP and MAP, which may increase vasopressor use and fluid balance at the timepoint of 24 hours in critically ill patients during ICU stay. However, after adjusting for confounders, esmolol treatment was not associated with 28-day and 90-day mortality.

Conditional generation of medical time series for extrapolation to underrepresented populations

The widespread adoption of electronic health records (EHRs) and subsequent increased availability of longitudinal healthcare data has led to significant advances in our understanding of health and disease with direct and immediate impact on the development of new diagnostics and therapeutic treatment options. However, access to EHRs is often restricted due to their perceived sensitive nature and associated legal concerns, and the cohorts therein typically are those seen at a specific hospital or network of hospitals and therefore not representative of the wider population of patients. Here, we present HealthGen, a new approach for the conditional generation of synthetic EHRs that maintains an accurate representation of real patient characteristics, temporal information and missingness patterns. We demonstrate experimentally that HealthGen generates synthetic cohorts that are significantly more faithful to real patient EHRs than the current state-of-the-art, and that augmenting real data sets with conditionally generated cohorts of underrepresented subpopulations of patients can significantly enhance the generalisability of models derived from these data sets to different patient populations. Synthetic conditionally generated EHRs could help increase the accessibility of longitudinal healthcare data sets and improve the generalisability of inferences made from these data sets to underrepresented populations.

Predicting ICU Mortality in Rheumatic Heart Disease: Comparison of XGBoost and Logistic Regression

Background

Rheumatic heart disease (RHD) accounts for a large proportion of Intensive Care Unit (ICU) deaths. Early prediction of RHD can help with timely and appropriate treatment to improve survival outcomes, and the XGBoost machine learning technology can be used to identify predictive factors; however, its use has been limited in the past. We compared the performance of logistic regression and XGBoost in predicting hospital mortality among patients with RHD from the Medical Information Mart for Intensive Care IV (MIMIC-IV) database.

Methods

The patients with RHD in the MIMIC-IV database were divided into two groups retrospectively according to the availability of data and its clinical significance based on whether they survived or died. Backward stepwise regression was used to analyze the independent factors influencing patients with RHD, and to compare the differences between the two groups. The XGBoost algorithm and logistic regression were used to establish two prediction models, and the areas under the receiver operating characteristic curves (AUCs) and decision-curve analysis (DCA) were used to test and compare the models. Finally, DCA and the clinical impact curve (CIC) were used to validate the model.

Results

Data on 1,634 patients with RHD were analyzed, comprising 207 who died during hospitalization and 1,427 survived. According to estimated results for the two models using AUCs [0.838 (95% confidence interval = 0.786–0.891) and 0.815 (95% confidence interval = 0.765–0.865)] and DCA, the logistic regression model performed better. DCA and CIC verified that the logistic regression model had convincing predictive value.

Conclusions

We used logistic regression analysis to establish a more meaningful prediction model for the final outcome of patients with RHD. This model might be clinically useful for patients with RHD and help clinicians to provide detailed treatments and precise management.

Unstructured clinical notes within the 24 hours since admission predict short, mid & long-term mortality in adult ICU patients

Mortality prediction for intensive care unit (ICU) patients is crucial for improving outcomes and efficient utilization of resources. Accessibility of electronic health records (EHR) has enabled data-driven predictive modeling using machine learning. However, very few studies rely solely on unstructured clinical notes from the EHR for mortality prediction. In this work, we propose a framework to predict short, mid, and long-term mortality in adult ICU patients using unstructured clinical notes from the MIMIC III database, natural language processing (NLP), and machine learning (ML) models. Depending on the statistical description of the patients' length of stay, we define the short-term as 48-hour and 4-day period, the mid-term as 7-day and 10-day period, and the long-term as 15-day and 30-day period after admission. We found that by only using clinical notes within the 24 hours of admission, our framework can achieve a high area under the receiver operating characteristics (AU-ROC) score for short, mid and long-term mortality prediction tasks. The test AU-ROC scores are 0.87, 0.83, 0.83, 0.82, 0.82, and 0.82 for 48-hour, 4-day, 7-day, 10-day, 15-day, and 30-day period mortality prediction, respectively. We also provide a comparative study among three types of feature extraction techniques from NLP: frequency-based technique, fixed embedding-based technique, and dynamic embedding-based technique. Lastly, we provide an interpretation of the NLP-based predictive models using feature-importance scores.

Association between circadian variation of heart rate and mortality among critically ill patients: a retrospective cohort study

Background

Heart rate (HR) related parameters, such as HR variability, HR turbulence, resting HR, and nighttime mean HR have been recognized as independent predictors of mortality. However, the influence of circadian changes in HR on mortality remains unclear in intensive care units (ICU). The study is designed to evaluate the relationship between the circadian variation in HR and mortality risk among critically ill patients.

Methods

The present study included 4,760 patients extracted from the Multiparameter Intelligent Monitoring in Intensive Care II database. The nighttime mean HR/daytime mean HR ratio was adopted as the circadian variation in HR. According to the median value of the circadian variation in HR, participants were divided into two groups: group A (≤ 1) and group B (> 1). The outcomes included ICU, hospital, 30-day, and 1-year mortalities. The prognostic value of HR circadian variation was investigated by multivariable logistic regression models and Cox proportional hazards models.

Results

Patients in group B (n = 2,471) had higher mortality than those in group A (n = 2,289). Multivariable models revealed that the higher circadian variation in HR was associated with ICU mortality (odds ratio [OR], 1.393; 95% confidence interval [CI], 1.112–1.745; P = 0.004), hospital mortality (OR, 1.393; 95% CI, 1.112–1.745; P = 0.004), 30-day mortality (hazard ratio, 1.260; 95% CI, 1.064–1.491; P = 0.007), and 1-year mortality (hazard ratio, 1.207; 95% CI, 1.057–1.378; P = 0.005), especially in patients with higher SOFA scores.

Conclusions

The circadian variation in HR might aid in the early identification of critically ill patients at high risk of associated with ICU, hospital, 30-day, and 1-year mortalities.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12871-022-01586-9.

Long-term outcomes of adults with acute respiratory failure treated with veno-venous extracorporeal membrane oxygenation

Veno-venous extracorporeal membrane oxygenation is increasingly used for severe but potentially reversible acute respiratory failure in adults; however, there are limited data regarding long-term morbidity. At our institution, most patients requiring veno-venous extracorporeal membrane oxygenation have been followed up by a single physician. Our primary aim was to describe the serial long-term morbidity for respiratory, musculoskeletal and psychological functioning.

A retrospective audit of inpatient and outpatient medical records was conducted. A total of 125 patients treated with veno-venous extracorporeal membrane oxygenation for primary respiratory failure were included. The patients were young (mean (standard deviation) age 43.7 (4.1) years), obese (mean (standard deviation) body mass index 30.8 (10.4) kg/m²), and mostly were male (59%). Most patients (60%) had no comorbidities.

The survival rate to discharge was 70%, with body mass index and the number of comorbidities being independent predictors of survival on multiple logistic regression analysis. Over half (57%) of the Australian survivors had regular outpatient follow-up. They had a median of three reviews (range 1-9) over a median of 11.8 months (range 1.5-79) months. Breathlessness and weakness resolved in most within six months, with lung function abnormalities taking longer to resolve. Over half (60%) returned to employment within six months of discharge. Over a quarter (29%) displayed symptoms of anxiety, depression or post-traumatic stress disorder.

Precision Medicine in Neonates: A Tailored Approach to Neonatal Brain Injury

Despite advances in neonatal care to prevent neonatal brain injury and neurodevelopmental impairment, predicting long-term outcome in neonates at risk for brain injury remains difficult. Early prognosis is currently based on cranial ultrasound (CUS), MRI, EEG, NIRS, and/or general movements assessed at specific ages, and predicting outcome in an individual (precision medicine) is not yet possible. New algorithms based on large databases and machine learning applied to clinical, neuromonitoring, and neuroimaging data and genetic analysis and assays measuring multiple biomarkers (omics) can fulfill the needs of modern neonatology. A synergy of all these techniques and the use of automatic quantitative analysis might give clinicians the possibility to provide patient-targeted decision-making for individualized diagnosis, therapy, and outcome prediction. This review will first focus on common neonatal neurological diseases, associated risk factors, and most common treatments. After that, we will discuss how precision medicine and machine learning (ML) approaches could change the future of prediction and prognosis in this field.

Ten physiological commandments for severe head injury

Advances in multiparametric brain monitoring have allowed us to deepen our knowledge of the physiopathology of head injury and how it can be treated using the therapies available today. It is essential to understand and interpret a series of basic physiological and physiopathological principles that, on the one hand, provide an adequate metabolic environment to prevent worsening of the primary brain injury and favour its recovery, and on the other hand, allow therapeutic resources to be individually adapted to the specific needs of the patient. Based on these notions, this article presents a decalogue of the physiological objectives to be achieved in brain injury, together with a series of diagnostic and therapeutic recommendations for achieving these goals. We emphasise the importance of considering and analysing the physiological variables involved in the transport of oxygen to the brain, such as cardiac output and arterial oxygen content, together with their conditioning factors and possible alterations. Special attention is paid to the basic elements of physiological neuroprotection, and we describe the multiple causes of cerebral hypoxia, how to approach them, and how to correct them. We also examine the increase in intracranial pressure as a physiopathological element, focussing on the significance of thoracic and abdominal pressure in the interpretation of intracranial pressure. Treatment of intracranial pressure should be based on a step-wise model, the first stage of which should be based on a physiopathological reflection combined with information on the tomographic lesions rather than on rigid numerical values.

Effect of heart rate on hospital mortality in critically ill patients may be modified by age: a retrospective observational study from large database

BACKGROUND

Heart rate has been found associated with mortality in critically ill patients. However, whether the association differs between the elderly and non-elderly patients was unknown.

METHODS

We conducted a retrospective observational study of adult patients admitted to the intensive care unit (ICU) in the United States. Demographic, vital signs, laboratory tests, and interventions were extracted and compared between the elderly and non-elderly patients. The main exposure was heart rate, the proportion of time spent in heart rate (PTS-HR) was calculated. The primary outcome was hospital mortality. The multivariable logistic regression model was performed to assess the relationship between PTS-HR and hospital mortality, and interaction between PTS-HR and age categories was explored.

RESULTS

104,276 patients were included, of which 52,378 (50.2%) were elderly patients and 51,898 (49.8%) were non-elderly patients. The median age was 66 (IQR 54-76) years. After adjusting for confounders, PTS-HR < 60 beats per minute (bpm) (OR 0.972, 95% CI [0.945, 0.998], p = 0.031, P_interaction = 0.001) and 60-80 bpm (OR 0.925, 95% CI [0.912, 0.938], p < 0.001, P_interaction = 0.553) were associated with decreased risk of mortality; PTS-HR 80-100 bpm was associated with decreased mortality in the non-elderly patients (OR 0.955, 95% CI [0.941,0.975], p < 0.001) but was associated with increased mortality in the very elderly patients (OR 1.018, 95% CI [1.003,1.029], p = 0.017, P_interaction < 0.001). PTS-HR > 100 bpm (OR 1.093, 95% CI [1.081,1.105], p < 0.001, P_interaction = 0.004) was associated with increased mortality.

CONCLUSIONS

The effect of heart rate on hospital mortality differs between the elderly and non-elderly critically ill patients.

Ten physiological commandments for severe head injury

Advances in multiparametric brain monitoring have allowed us to deepen our knowledge of the physiopathology of head injury and how it can be treated using the therapies available today. It is essential to understand and interpret a series of basic physiological and physiopathological principles that, on the one hand, provide an adequate metabolic environment to prevent worsening of the primary brain injury and favour its recovery, and on the other hand, allow therapeutic resources to be individually adapted to the specific needs of the patient. Based on these notions, this article presents a decalogue of the physiological objectives to be achieved in brain injury, together with a series of diagnostic and therapeutic recommendations for achieving these goals. We emphasise the importance of considering and analysing the physiological variables involved in the transport of oxygen to the brain, such as cardiac output and arterial oxygen content, together with their conditioning factors and possible alterations. Special attention is paid to the basic elements of physiological neuroprotection, and we describe the multiple causes of cerebral hypoxia, how to approach them, and how to correct them. We also examine the increase in intracranial pressure as a physiopathological element, focussing on the significance of thoracic and abdominal pressure in the interpretation of intracranial pressure. Treatment of intracranial pressure should be based on a step-wise model, the first stage of which should be based on a physiopathological reflection combined with information on the tomographic lesions rather than on rigid numerical values.

Prediction of fatal adverse prognosis in patients with fever-related diseases based on machine learning: A retrospective study

Background

Fever is the most common chief complaint of emergency patients. Early identification of patients at an increasing risk of death may avert adverse outcomes. The aim of this study was to establish an early prediction model of fatal adverse prognosis of fever patients by extracting key indicators using big data technology.

Methods

A retrospective study of patients’ data was conducted using the Emergency Rescue Database of Chinese People's Liberation Army General Hospital. Patients were divided into the fatal adverse prognosis group and the good prognosis group. The commonly used clinical indicators were compared. Recursive feature elimination (RFE) method was used to determine the optimal number of the included variables. In the training model, logistic regression, random forest, adaboost and bagging were selected. We also collected the emergency room data from December 2018 to December 2019 with the same inclusion and exclusion criterion. The performance of the model was evaluated by accuracy, F1-score, precision, sensitivity and the areas under receiver operator characteristic curves (ROC-AUC).

Results

The accuracy of logistic regression, decision tree, adaboost and bagging was 0.951, 0.928, 0.924, and 0.924, F1-scores were 0.938, 0.933, 0.930, and 0.930, the precision was 0.943, 0.938, 0.937, and 0.937, ROC-AUC were 0.808, 0.738, 0.736, and 0.885, respectively. ROC-AUC of ten-fold cross-validation in logistic and bagging models were 0.80 and 0.87, respectively. The top six coefficients and odds ratio (OR) values of the variables in the Logistic regression were cardiac troponin T (CTnT) (coefficient=0.346, OR = 1.413), temperature (T) (coefficient=0.235, OR = 1.265), respiratory rate (RR) (coefficient= –0.206,OR = 0.814), serum kalium (K) (coefficient=0.137, OR = 1.146), pulse oxygen saturation (SPO₂) (coefficient= –0.101, OR = 0.904), and albumin (ALB) (coefficient= –0.043, OR = 0.958). The weights of the top six variables in the bagging model were: CTnT, RR, lactate dehydrogenase, serum amylase, heartrate, and systolic blood pressure.

Conclusions

The main clinical indicators of concern included CTnT, RR, SPO₂, T, ALB and K. The bagging model and logistic regression model had better diagnostic performance comprehesively. Those may be conducive to the early identification of critical patients with fever by physicians.