Interpreting a recurrent neural network's predictions of ICU mortality risk

Predicting outcomes using neural networks in the intensive care unit

Patients in intensive care units (ICUs) require rapid critical decision making. Modern ICUs are data rich, where information streams from diverse sources. Machine learning (ML) and neural networks (NN) can leverage the rich data for prognostication and clinical care. They can handle complex nonlinear relationships in medical data and have advantages over traditional predictive methods. A number of models are used: (1) Feedforward networks; and (2) Recurrent NN and convolutional NN to predict key outcomes such as mortality, length of stay in the ICU and the likelihood of complications. Current NN models exist in silos; their integration into clinical workflow requires greater transparency on data that are analyzed. Most models that are accurate enough for use in clinical care operate as ‘black-boxes’ in which the logic behind their decision making is opaque. Advances have occurred to see through the opacity and peer into the processing of the black-box. In the near future ML is positioned to help in clinical decision making far beyond what is currently possible. Transparency is the first step toward validation which is followed by clinical trust and adoption. In summary, NNs have the transformative ability to enhance predictive accuracy and improve patient management in ICUs. The concept should soon be turning into reality.

Predictive analytics in bronchopulmonary dysplasia: past, present, and future

Bronchopulmonary dysplasia (BPD) remains a significant complication of prematurity, impacting approximately 18,000 infants annually in the United States. Advances in neonatal care have not reduced BPD, and its management is challenged by the rising survival of extremely premature infants and the variability in clinical practices. Leveraging statistical and machine learning techniques, predictive analytics can enhance BPD management by utilizing large clinical datasets to predict individual patient outcomes. This review explores the foundations and applications of predictive analytics in the context of BPD, examining commonly used data sources, modeling techniques, and metrics for model evaluation. We also highlight bioinformatics' potential role in understanding BPD's molecular basis and discuss case studies demonstrating the use of machine learning models for risk prediction and prognosis in neonates. Challenges such as data bias, model complexity, and ethical considerations are outlined, along with strategies to address these issues. Future directions for advancing the integration of predictive analytics into clinical practice include improving model interpretability, expanding data sharing and interoperability, and aligning predictive models with precision medicine goals. By overcoming current challenges, predictive analytics holds promise for transforming neonatal care and providing personalized interventions for infants at risk of BPD.

Knowledge enhanced attention aggregation network for medicine recommendation

The combination of deep learning and the medical field has recently achieved great success, particularly in recommending medicine for patients. However, patients' clinical records often contain repeated medical information that can significantly impact their health condition. Most existing methods for modeling longitudinal patient information overlook the impact of individual diagnoses and procedures on the patient's health, resulting in insufficient patient representation and limited accuracy of medicine recommendations. Therefore, we propose a medicine recommendation model called KEAN, which is based on an attention aggregation network and enhanced graph convolution. Specifically, KEAN can aggregate individual diagnoses and procedures in patient visits to capture significant features that affect patients' diseases. We further incorporate medicine knowledge from complex medicine combinations, reduce drug-drug interactions (DDIs), and recommend medicines that are beneficial to patients' health. The experimental results on the MIMIC-III dataset demonstrate that our model outperforms existing advanced methods, which highlights the effectiveness of the proposed method.

Predicting intubation for intensive care units patients: A deep learning approach to improve patient management

OBJECTIVE

For patients in the Intensive Care Unit (ICU), the timing of intubation has a significant association with patients' outcomes. However, accurate prediction of the timing of intubation remains an unsolved challenge due to the noisy, sparse, heterogeneous, and unbalanced nature of ICU data. In this study, our objective is to develop a workflow for pre-processing ICU data and to develop a customized deep learning model to predict the need for intubation.

METHODS

To improve the prediction accuracy, we transform the intubation prediction task into a time series classification task. We carefully design a sequence of data pre-processing steps to handle the multimodal noisy data. Firstly, we discretize the sequential data and address missing data using interpolation. Next, we employ a sampling strategy to address data imbalance and standardize the data to facilitate faster model convergence. Furthermore, we employ the feature selection technique and propose an ensemble model to combine features learned by different deep learning models.

RESULTS

The performance is evaluated on Medical Information Mart for Intensive Care (MIMIC)-III, an ICU dataset. Our proposed Deep Feature Fusion method achieves an area under the curve (AUC) of the receiver operating curve (ROC) of 0.8953, surpassing the performance of other deep learning and traditional machine learning models.

CONCLUSION

Our proposed Deep Feature Fusion method proves to be a viable approach for predicting intubation and outperforms other deep learning and classical machine learning models. The study confirms that high-frequency time-varying indicators, particularly Mean Blood Pressure (MeanBP) and peripheral oxygen saturation (SpO2), are significant risk factors for predicting intubation.

Multi-organ spatiotemporal information aware model for sepsis mortality prediction

BACKGROUND

Sepsis is a syndrome involving multi-organ dysfunction, and the mortality in sepsis patients correlates with the number of lesioned organs. Precise prognosis models play a pivotal role in enabling healthcare practitioners to administer timely and accurate interventions for sepsis, thereby augmenting patient outcomes. Nevertheless, the majority of available models consider the overall physiological attributes of patients, overlooking the asynchronous spatiotemporal interactions among multiple organ systems. These constraints hinder a full application of such models, particularly when dealing with limited clinical data. To surmount these challenges, a comprehensive model, denoted as recurrent Graph Attention Network-multi Gated Recurrent Unit (rGAT-mGRU), was proposed. Taking into account the intricate spatiotemporal interactions among multiple organ systems, the model predicted in-hospital mortality of sepsis using data collected within the 48-hour period post-diagnosis.

MATERIAL AND METHODS

Multiple parallel GRU sub-models were formulated to investigate the temporal physiological variations of single organ systems. Meanwhile, a GAT structure featuring a memory unit was constructed to capture spatiotemporal connections among multi-organ systems. Additionally, an attention-injection mechanism was employed to govern the data flowing within the network pertaining to multi-organ systems. The proposed model underwent training and testing using a dataset of 10,181 sepsis cases extracted from the Medical Information Mart for Intensive Care III (MIMIC-III) database. To evaluate the model's superiority, it was compared with the existing common baseline models. Furthermore, ablation experiments were designed to elucidate the rationale and robustness of the proposed model.

RESULTS

Compared with the baseline models for predicting mortality of sepsis, the rGAT-mGRU model demonstrated the largest area under the receiver operating characteristic curve (AUROC) of 0.8777 ± 0.0039 and the maximum area under the precision-recall curve (AUPRC) of 0.5818 ± 0.0071, with sensitivity of 0.8358 ± 0.0302 and specificity of 0.7727 ± 0.0229, respectively. The proposed model was capable of delineating the varying contribution of the involved organ systems at distinct moments, as specifically illustrated by the attention weights. Furthermore, it exhibited consistent performance even in the face of limited clinical data.

CONCLUSION

The rGAT-mGRU model has the potential to indicate sepsis prognosis by extracting the dynamic spatiotemporal interplay information inherent in multi-organ systems during critical diseases, thereby providing clinicians with auxiliary decision-making support.

MVIRA: A model based on Missing Value Imputation and Reliability Assessment for mortality risk prediction

BACKGROUND

Mortality risk prediction is to predict whether a patient has the risk of death based on relevant diagnosis and treatment data. How to accurately predict patient mortality risk based on electronic health records (EHR) is currently a hot research topic in the healthcare field. In actual medical datasets, there are often many missing values, which can seriously interfere with the effect of model prediction. However, when missing values are interpolated, most existing methods do not take into account the fidelity or confidence of the interpolated values. Misestimation of missing variables can lead to modeling difficulties and performance degradation, while the reliability of the model may be compromised in clinical environments.

MATERIALS AND METHODS

We propose a model based on Missing Value Imputation and Reliability Assessment for mortality risk prediction (MVIRA). The model uses a combination of variational autoencoder and recurrent neural networks to complete the interpolation of missing values and enhance the characterization ability of EHR data, thus improving the performance of mortality risk prediction. In addition, we also introduce the Monte Carlo Dropout method to calculate the uncertainty of the model prediction results and thus achieve the reliability assessment of the model.

RESULTS

We perform performance validation of the model on the public datasets MIMIC-III and MIMIC-IV. The proposed model showed improved performance compared with competitive models in terms of overall specialties.

CONCLUSION

The proposed model can effectively improve the accuracy of mortality risk prediction, and can help medical institutions assess the condition of patients.

Multivariate longitudinal data for survival analysis of cardiovascular event prediction in young adults: insights from a comparative explainable study

BACKGROUND

Multivariate longitudinal data are under-utilized for survival analysis compared to cross-sectional data (CS - data collected once across cohort). Particularly in cardiovascular risk prediction, despite available methods of longitudinal data analysis, the value of longitudinal information has not been established in terms of improved predictive accuracy and clinical applicability.

METHODS

We investigated the value of longitudinal data over and above the use of cross-sectional data via 6 distinct modeling strategies from statistics, machine learning, and deep learning that incorporate repeated measures for survival analysis of the time-to-cardiovascular event in the Coronary Artery Risk Development in Young Adults (CARDIA) cohort. We then examined and compared the use of model-specific interpretability methods (Random Survival Forest Variable Importance) and model-agnostic methods (SHapley Additive exPlanation (SHAP) and Temporal Importance Model Explanation (TIME)) in cardiovascular risk prediction using the top-performing models.

RESULTS

In a cohort of 3539 participants, longitudinal information from 35 variables that were repeatedly collected in 6 exam visits over 15 years improved subsequent long-term (17 years after) risk prediction by up to 8.3% in C-index compared to using baseline data (0.78 vs. 0.72), and up to approximately 4% compared to using the last observed CS data (0.75). Time-varying AUC was also higher in models using longitudinal data (0.86-0.87 at 5 years, 0.79-0.81 at 10 years) than using baseline or last observed CS data (0.80-0.86 at 5 years, 0.73-0.77 at 10 years). Comparative model interpretability analysis revealed the impact of longitudinal variables on model prediction on both the individual and global scales among different modeling strategies, as well as identifying the best time windows and best timing within that window for event prediction. The best strategy to incorporate longitudinal data for accuracy was time series massive feature extraction, and the easiest interpretable strategy was trajectory clustering.

CONCLUSION

Our analysis demonstrates the added value of longitudinal data in predictive accuracy and epidemiological utility in cardiovascular risk survival analysis in young adults via a unified, scalable framework that compares model performance and explainability. The framework can be extended to a larger number of variables and other longitudinal modeling methods.

TRIAL REGISTRATION

ClinicalTrials.gov Identifier: NCT00005130, Registration Date: 26/05/2000.

An explainable knowledge distillation method with XGBoost for ICU mortality prediction

BACKGROUND AND OBJECTIVE

Mortality prediction is an important task in intensive care unit (ICU) for quantifying the severity of patients' physiological condition. Currently, scoring systems are widely applied for mortality prediction, while the performance is unsatisfactory in many clinical conditions due to the non-specificity and linearity characteristics of the used model. As the availability of the large volume of data recorded in electronic health records (EHRs), deep learning models have achieved state-of-art predictive performance. However, deep learning models are hard to meet the requirement of explainability in clinical conditions. Hence, an explainable Knowledge Distillation method with XGBoost (XGB-KD) is proposed to improve the predictive performance of XGBoost while supporting better explainability.

METHODS

In this method, we first use outperformed deep learning teacher models to learn the complex patterns hidden in high-dimensional multivariate time series data. Then, we distill knowledge from soft labels generated by the ensemble of teacher models to guide the training of XGBoost student model, whose inputs are meaningful features obtained from feature engineering. Finally, we conduct model calibration to obtain predicted probabilities reflecting the true posterior probabilities and use SHapley Additive exPlanations (SHAP) to obtain insights about the trained model.

RESULTS

We conduct comprehensive experiments on MIMIC-III dataset to evaluate our method. The results demonstrate that our method achieves better predictive performance than vanilla XGBoost, deep learning models and several state-of-art baselines from related works. Our method can also provide intuitive explanations.

CONCLUSIONS

Our method is useful for improving the predictive performance of XGBoost by distilling knowledge from deep learning models and can provide meaningful explanations for predictions.

Multivariate Sequential Analytics for Cardiovascular Disease Event Prediction

BACKGROUND

Automated clinical decision support for risk assessment is a powerful tool in combating cardiovascular disease (CVD), enabling targeted early intervention that could avoid issues of overtreatment or undertreatment. However, current CVD risk prediction models use observations at baseline without explicitly representing patient history as a time series.

OBJECTIVE

The aim of this study is to examine whether by explicitly modelling the temporal dimension of patient history event prediction may be improved.

METHODS

This study investigates methods for multivariate sequential modelling with a particular emphasis on long short-term memory (LSTM) recurrent neural networks. Data from a CVD decision support tool is linked to routinely collected national datasets including pharmaceutical dispensing, hospitalization, laboratory test results, and deaths. The study uses a 2-year observation and a 5-year prediction window. Selected methods are applied to the linked dataset. The experiments performed focus on CVD event prediction. CVD death or hospitalization in a 5-year interval was predicted for patients with history of lipid-lowering therapy.

RESULTS

The results of the experiments showed temporal models are valuable for CVD event prediction over a 5-year interval. This is especially the case for LSTM, which produced the best predictive performance among all models compared achieving AUROC of 0.801 and average precision of 0.425. The non-temporal model comparator ridge classifier (RC) trained using all quarterly data or by aggregating quarterly data (averaging time-varying features) was highly competitive achieving AUROC of 0.799 and average precision of 0.420 and AUROC of 0.800 and average precision of 0.421, respectively.

CONCLUSION

This study provides evidence that the use of deep temporal models particularly LSTM in clinical decision support for chronic disease would be advantageous with LSTM significantly improving on commonly used regression models such as logistic regression and Cox proportional hazards on the task of CVD event prediction.

Medication adherence prediction through temporal modelling in cardiovascular disease management

Background

Chronic conditions place a considerable burden on modern healthcare systems. Within New Zealand and worldwide cardiovascular disease (CVD) affects a significant proportion of the population and it is the leading cause of death. Like other chronic diseases, the course of cardiovascular disease is usually prolonged and its management necessarily long-term. Despite being highly effective in reducing CVD risk, non-adherence to long-term medication continues to be a longstanding challenge in healthcare delivery. The study investigates the benefits of integrating patient history and assesses the contribution of explicitly temporal models to medication adherence prediction in the context of lipid-lowering therapy.

Methods

Data from a CVD risk assessment tool is linked to routinely collected national and regional data sets including pharmaceutical dispensing, hospitalisation, lab test results and deaths. The study extracts a sub-cohort from 564,180 patients who had primary CVD risk assessment for analysis. Based on community pharmaceutical dispensing record, proportion of days covered (PDC) $$\ge$$ ≥  80 is used as the threshold for adherence. Two years (8 quarters) of patient history before their CVD risk assessment is used as the observation window to predict patient adherence in the subsequent 5 years (20 quarters). The predictive performance of temporal deep learning models long short-term memory (LSTM) and simple recurrent neural networks (Simple RNN) are compared against non-temporal models multilayer perceptron (MLP), ridge classifier (RC) and logistic regression (LR). Further, the study investigates the effect of lengthening the observation window on the task of adherence prediction.

Results

Temporal models that use sequential data outperform non-temporal models, with LSTM producing the best predictive performance achieving a ROC AUC of 0.805. A performance gap is observed between models that can discover non-linear interactions between predictor variables and their linear counter parts, with neural network (NN) based models significantly outperforming linear models. Additionally, the predictive advantage of temporal models become more pronounced when the length of the observation window is increased.

Conclusion

The findings of the study provide evidence that using deep temporal models to integrate patient history in adherence prediction is advantageous. In particular, the RNN architecture LSTM significantly outperforms all other model comparators.

Mortality prediction in ICU Using a Stacked Ensemble Model

Artificial intelligence (AI) technology has huge scope in developing models to predict the survival rate of critically ill patients in the intensive care unit (ICU). The availability of electronic clinical data has led to the widespread use of various machine learning approaches in this field. Innovative algorithms play a crucial role in boosting the performance of models. This study uses a stacked ensemble model to predict mortality in ICU by incorporating the clinical severity scoring results, in which several machine learning algorithms are employed to compare the performance. The experimental results show that the stacked ensemble model achieves good performance compared with the model without integrating the severity scoring results, which has the area under curve (AUC) of 0.879 and 0.862, respectively. To improve the performance of prediction, two feature subsets are obtained based on different feature selection techniques, labeled as SetS and SetT. Evaluation performances show that the SEM based on the SetS achieves a higher AUC value (0.879 and 0.860). Finally, the SHapley Additive exPlanations (SHAP) analysis is employed to interpret the correlation between the risk features and the outcome.

[Long short-term memory and Logistic regression for mortality risk prediction of intensive care unit patients with stroke]

OBJECTIVE

To select variables related to mortality risk of stroke patients in intensive care unit (ICU) through long short-term memory (LSTM) with attention mechanisms and Logistic regression with L1 norm, and to construct mortality risk prediction model based on conventional Logistic regression with important variables selected from the two models and to evaluate the model performance.

METHODS

Medical Information Mart for Intensive Care (MIMIC)-Ⅳ database was retrospectively analyzed and the patients who were primarily diagnosed with stroke were selected as study population. The outcome was defined as whether the patient died in hospital after admission. Candidate predictors included demogra-phic information, complications, laboratory tests and vital signs in the initial 48 h after ICU admission. The data were randomly divided into a training set and a test set for ten times at a ratio of 8 ∶2. In training sets, LSTM with attention mechanisms and Logistic regression with L1 norm were constructed to select important variables. In the test sets, the mean importance of variables of ten times was used as a reference to pick out the top 10 variables in each of the two models, and then these variables were included in conventional Logistic regression to build the final prediction model. Model evaluation was based on the area under the receiver operating characteristic curve (AUC), sensitivity, specificity, and accuracy. And the model performance was compared with the forward Logistic regression model which hadn't conducted variable selection previously.

RESULTS

A total of 2 755 patients with 2 979 ICU admission records were included in the analysis, of which 526 recorded deaths. The AUC of Logistic regression model with L1 norm was statistically better than that of LSTM with attention mechanisms (0.819±0.031 vs. 0.760±0.018, P < 0.001). Age, blood glucose, and blood urea nitrogen were at the top ten important variables in both of the two models. AUC, sensitivity, specificity, and accuracy of Logistic regression models were 0.85, 85.98%, 71.74% and 74.26%, respectively. And the final prediction model was superior to forward Logistic regression model.

CONCLUSION

The variables selected by Logistic regression with L1 norm and LSTM with attention mechanisms had good prediction performance, which showed important implications on the mortality prediction of stroke patients in ICU.

Development of a deep learning model that predicts Bi-level positive airway pressure failure

Delaying intubation for patients failing Bi-Level Positive Airway Pressure (BIPAP) may be associated with harm. The objective of this study was to develop a deep learning model capable of aiding clinical decision making by predicting Bi-Level Positive Airway Pressure (BIPAP) failure. This was a retrospective cohort study in a tertiary pediatric intensive care unit (PICU) between 2010 and 2020. Three machine learning models were developed to predict BIPAP failure: two logistic regression models and one deep learning model, a recurrent neural network with a Long Short-Term Memory (LSTM-RNN) architecture. Model performance was evaluated in a holdout test set. 175 (27.7%) of 630 total BIPAP sessions were BIPAP failures. Patients in the BIPAP failure group were on BIPAP for a median of 32.8 (9.2-91.3) hours prior to intubation. Late BIPAP failure (intubation after using BIPAP > 24 h) patients had fewer 28-day Ventilator Free Days (13.40 [0.68-20.96]), longer ICU length of stay and more post-extubation BIPAP days compared to those who were intubated ≤ 24 h from BIPAP initiation. An AUROC above 0.5 indicates that a model has extracted new information, potentially valuable to the clinical team, about BIPAP failure. Within 6 h of BIPAP initiation, the LSTM-RNN model predicted which patients were likely to fail BIPAP with an AUROC of 0.81 (0.80, 0.82), superior to all other models. Within 6 h of BIPAP initiation, the LSTM-RNN model would identify nearly 80% of BIPAP failures with a 50% false alarm rate, equal to an NNA of 2. In conclusion, a deep learning method using readily available data from the electronic health record can identify which patients on BIPAP are likely to fail with good discrimination, oftentimes days before they are intubated in usual practice.

Dynamic Mortality Risk Predictions for Children in ICUs: Development and Validation of Machine Learning Models

OBJECTIVES

Assess a machine learning method of serially updated mortality risk.

DESIGN

Retrospective analysis of a national database (Health Facts; Cerner Corporation, Kansas City, MO).

SETTING

Hospitals caring for children in ICUs.

PATIENTS

A total of 27,354 admissions cared for in ICUs from 2009 to 2018.

INTERVENTIONS

None.

MAIN OUTCOME

Hospital mortality risk estimates determined at 6-hour time periods during care in the ICU. Models were truncated at 180 hours due to decreased sample size secondary to discharges and deaths.

MEASUREMENTS AND MAIN RESULTS

The Criticality Index, based on physiology, therapy, and care intensity, was computed for each admission for each time period and calibrated to hospital mortality risk (Criticality Index-Mortality [CI-M]) at each of 29 time periods (initial assessment: 6 hr; last assessment: 180 hr). Performance metrics and clinical validity were determined from the held-out test sample (n = 3,453, 13%). Discrimination assessed with the area under the receiver operating characteristic curve was 0.852 (95% CI, 0.843-0.861) overall and greater than or equal to 0.80 for all individual time periods. Calibration assessed by the Hosmer-Lemeshow goodness-of-fit test showed good fit overall (p = 0.196) and was statistically not significant for 28 of the 29 time periods. Calibration plots for all models revealed the intercept ranged from--0.002 to 0.009, the slope ranged from 0.867 to 1.415, and the R2 ranged from 0.862 to 0.989. Clinical validity assessed using population trajectories and changes in the risk status of admissions (clinical volatility) revealed clinical trajectories consistent with clinical expectations and greater clinical volatility in deaths than survivors (p < 0.001).

CONCLUSIONS

Machine learning models incorporating physiology, therapy, and care intensity can track changes in hospital mortality risk during intensive care. The CI-M's framework and modeling method are potentially applicable to monitoring clinical improvement and deterioration in real time.

A Drug Recommendation Model Based on Message Propagation and DDI Gating Mechanism

Drug recommendation task based on deep learning models has been widely studied and applied in the field of health care in recent years. However, the accuracy of drug recommendation models still needs to be improved. In addition, the existing recommendation models either give only one recommendation (however, there may be a variety of treatment options in practice) or can not give the confidence level of the recommendation. To fill these gaps, a Drug Recommendation model based on Message Propagation neural network (denoted as DRMP) is proposed in this paper. Then, the Drug-Drug Interaction (DDI) knowledge is introduced into the proposed model to reduce the DDI rate of recommended results. Finally, we extend the decoding module of the proposed model to Bayesian Neural Network (BNN), which enables the model to output a variety of recommendation results and give the confidence of each recommendation result, so as to provide more information for users (such as doctors) to make decisions. Experimental results on public data sets show that the proposed model is superior to the best existing models.

Continuous Prediction of Mortality in the PICU: A Recurrent Neural Network Model in a Single-Center Dataset

OBJECTIVES

Develop, as a proof of concept, a recurrent neural network model using electronic medical records data capable of continuously assessing an individual child's risk of mortality throughout their ICU stay as a proxy measure of severity of illness.

DESIGN

Retrospective cohort study.

SETTING

PICU in a tertiary care academic children's hospital.

PATIENTS/SUBJECTS

Twelve thousand five hundred sixteen episodes (9,070 children) admitted to the PICU between January 2010 and February 2019, partitioned into training (50%), validation (25%), and test (25%) sets.

INTERVENTIONS

None.

MEASUREMENTS AND MAIN RESULTS

On 2,475 test set episodes lasting greater than or equal to 24 hours in the PICU, the area under the receiver operating characteristic curve of the recurrent neural network's 12th hour predictions was 0.94 (CI, 0.93-0.95), higher than those of Pediatric Index of Mortality 2 (0.88; CI, [0.85-0.91]; p < 0.02), Pediatric Risk of Mortality III (12th hr) (0.89; CI, [0.86-0.92]; p < 0.05), and Pediatric Logistic Organ Dysfunction day 1 (0.85; [0.81-0.89]; p < 0.002). The recurrent neural network's discrimination increased with more acquired data and smaller lead time, achieving a 0.99 area under the receiver operating characteristic curve 24 hours prior to discharge. Despite not having diagnostic information, the recurrent neural network performed well across different primary diagnostic categories, generally achieving higher area under the receiver operating characteristic curve for these groups than the other three scores. On 692 test set episodes lasting greater than or equal to 5 days in the PICU, the recurrent neural network area under the receiver operating characteristic curves significantly outperformed their daily Pediatric Logistic Organ Dysfunction counterparts (p < 0.005).

CONCLUSIONS

The recurrent neural network model can process hundreds of input variables contained in a patient's electronic medical record and integrate them dynamically as measurements become available. Its high discrimination suggests the recurrent neural network's potential to provide an accurate, continuous, and real-time assessment of a child in the ICU.