A targeted real-time early warning score (TREWScore) for septic shock

Unlocking stress and forecasting its consequences with digital technology

Chronic stress is a major underlying origin of the top leading causes of death, globally. Yet, the mechanistic explanation of the association between stress and disease is poorly understood. This stems from the inability to adequately measure stress in its naturally occurring state and the extreme heterogeneity by inter and intraindividual characteristics. The growth and availability of digital technologies involving wearable devices and mobile phone apps afford the opportunity to dramatically improve measurement of the biological stress response in real time. In parallel, the advancement and capabilities of artificial intelligence (AI) and machine learning could discern heterogeneous, multidimensional information from individual signs of stress, and possibly inform how these signs forecast the downstream consequences of stress in the form of end-organ damage. The marriage of these tools could dramatically enhance the field of stress research contributing to impactful and empowering interventions for individuals bridging knowledge to practice, and intervention to real-world use. Here we discuss this potential, anticipated challenges, and emerging opportunities.

Multitask learning and benchmarking with clinical time series data

Health care is one of the most exciting frontiers in data mining and machine learning. Successful adoption of electronic health records (EHRs) created an explosion in digital clinical data available for analysis, but progress in machine learning for healthcare research has been difficult to measure because of the absence of publicly available benchmark data sets. To address this problem, we propose four clinical prediction benchmarks using data derived from the publicly available Medical Information Mart for Intensive Care (MIMIC-III) database. These tasks cover a range of clinical problems including modeling risk of mortality, forecasting length of stay, detecting physiologic decline, and phenotype classification. We propose strong linear and neural baselines for all four tasks and evaluate the effect of deep supervision, multitask training and data-specific architectural modifications on the performance of neural models.

Clinical decision support system to assess the risk of sepsis using Tree Augmented Bayesian networks and electronic medical record data

Early and accurate diagnoses of sepsis enable practitioners to take timely preventive actions. The existing diagnostic criteria suffer from deficiencies, such as triggering false alarms or leaving conditions undiagnosed. This study aims to develop a clinical decision support system to predict the risk of sepsis using tree augmented naive Bayesian network by identifying the optimal set of biomarkers. The key feature of our approach is that we captured the dynamics among biomarkers. With an area under receiver operating characteristic of 0.84, the proposed model outperformed the competing diagnostic criteria (systemic inflammatory response syndrome = 0.59, quick sepsis-related organ failure assessment = 0.65, modified early warning system = 0.75, sepsis-related organ failure assessment = 0.80). The richness of our proposed model is measured not only by achieving high accuracy, but also by utilizing fewer biomarkers. We also propose a left-center-right imputation method suitable for electronic medical record data. This method uses the individual patient’s visit, instead of aggregated (mean or median) value, to impute the missing data.

Challenges and Opportunities for Using Big Health Care Data to Advance Medical Science and Public Health

Methodological advancements in epidemiology, biostatistics, and data science have strengthened the research world's ability to use data captured from electronic health records (EHRs) to address pressing medical questions, but gaps remain. We describe methods investments that are needed to curate EHR data toward research quality and to integrate complementary data sources when EHR data alone are insufficient for research goals. We highlight new methods and directions for improving the integrity of medical evidence generated from pragmatic trials, observational studies, and predictive modeling. We also discuss needed methods contributions to further ease data sharing across multisite EHR data networks. Throughout, we identify opportunities for training and for bolstering collaboration among subject matter experts, methodologists, practicing clinicians, and health system leaders to help ensure that methods problems are identified and resulting advances are translated into mainstream research practice more quickly.

Evaluation of a machine learning algorithm for up to 48-hour advance prediction of sepsis using six vital signs

OBJECTIVE

Sepsis remains a costly and prevalent syndrome in hospitals; however, machine learning systems can increase timely sepsis detection using electronic health records. This study validates a gradient boosted ensemble machine learning tool for sepsis detection and prediction, and compares its performance to existing methods.

MATERIALS AND METHODS

Retrospective data was drawn from databases at the University of California, San Francisco (UCSF) Medical Center and the Beth Israel Deaconess Medical Center (BIDMC). Adult patient encounters without sepsis on admission, and with at least one recording of each of six vital signs (SpO₂, heart rate, respiratory rate, temperature, systolic and diastolic blood pressure) were included. We compared the performance of the machine learning algorithm (MLA) to that of commonly used scoring systems. Area under the receiver operating characteristic (AUROC) curve was our primary measure of accuracy. MLA performance was measured at sepsis onset, and at 24 and 48 h prior to sepsis onset.

RESULTS

The MLA achieved an AUROC of 0.88, 0.84, and 0.83 for sepsis onset and 24 and 48 h prior to onset, respectively. These values were superior to those of SIRS (0.66), MEWS (0.61), SOFA (0.72), and qSOFA (0.60) at time of onset. When trained on UCSF data and tested on BIDMC data, sepsis onset AUROC was 0.89.

DISCUSSION AND CONCLUSION

The MLA predicts sepsis up to 48 h in advance and identifies sepsis onset more accurately than commonly used tools, maintaining high performance for sepsis detection when trained and tested on separate datasets.

Predicting childhood obesity using electronic health records and publicly available data

BACKGROUND

Because of the strong link between childhood obesity and adulthood obesity comorbidities, and the difficulty in decreasing body mass index (BMI) later in life, effective strategies are needed to address this condition in early childhood. The ability to predict obesity before age five could be a useful tool, allowing prevention strategies to focus on high risk children. The few existing prediction models for obesity in childhood have primarily employed data from longitudinal cohort studies, relying on difficult to collect data that are not readily available to all practitioners. Instead, we utilized real-world unaugmented electronic health record (EHR) data from the first two years of life to predict obesity status at age five, an approach not yet taken in pediatric obesity research.

METHODS AND FINDINGS

We trained a variety of machine learning algorithms to perform both binary classification and regression. Following previous studies demonstrating different obesity determinants for boys and girls, we similarly developed separate models for both groups. In each of the separate models for boys and girls we found that weight for length z-score, BMI between 19 and 24 months, and the last BMI measure recorded before age two were the most important features for prediction. The best performing models were able to predict obesity with an Area Under the Receiver Operator Characteristic Curve (AUC) of 81.7% for girls and 76.1% for boys.

CONCLUSIONS

We were able to predict obesity at age five using EHR data with an AUC comparable to cohort-based studies, reducing the need for investment in additional data collection. Our results suggest that machine learning approaches for predicting future childhood obesity using EHR data could improve the ability of clinicians and researchers to drive future policy, intervention design, and the decision-making process in a clinical setting.

Predicting adverse hemodynamic events in critically ill patients

PURPOSE OF REVIEW

The art of predicting future hemodynamic instability in the critically ill has rapidly become a science with the advent of advanced analytical processed based on computer-driven machine learning techniques. How these methods have progressed beyond severity scoring systems to interface with decision-support is summarized.

RECENT FINDINGS

Data mining of large multidimensional clinical time-series databases using a variety of machine learning tools has led to our ability to identify alert artifact and filter it from bedside alarms, display real-time risk stratification at the bedside to aid in clinical decision-making and predict the subsequent development of cardiorespiratory insufficiency hours before these events occur. This fast evolving filed is primarily limited by linkage of high-quality granular to physiologic rationale across heterogeneous clinical care domains.

SUMMARY

Using advanced analytic tools to glean knowledge from clinical data streams is rapidly becoming a reality whose clinical impact potential is great.

Data-driven discovery of a novel sepsis pre-shock state predicts impending septic shock in the ICU

Septic shock is a life-threatening condition in which timely treatment substantially reduces mortality. Reliable identification of patients with sepsis who are at elevated risk of developing septic shock therefore has the potential to save lives by opening an early window of intervention. We hypothesize the existence of a novel clinical state of sepsis referred to as the "pre-shock" state, and that patients with sepsis who enter this state are highly likely to develop septic shock at some future time. We apply three different machine learning techniques to the electronic health record data of 15,930 patients in the MIMIC-III database to test this hypothesis. This novel paradigm yields improved performance in identifying patients with sepsis who will progress to septic shock, as defined by Sepsis- 3 criteria, with the best method achieving a 0.93 area under the receiver operating curve, 88% sensitivity, 84% specificity, and median early warning time of 7 hours. Additionally, we introduce the notion of patient-specific positive predictive value, assigning confidence to individual predictions, and achieving values as high as 91%. This study demonstrates that early prediction of impending septic shock, and thus early intervention, is possible many hours in advance.

The Promise of Machine Learning: When Will it be Delivered?

BACKGROUND

The real-life applications of machine learning clinical decision making is currently lagging behind its promise. One of the critics on machine learning is that it doesn't outperform more traditional statistical approaches in every problem.

METHODS AND RESULTS

Authors of "Predictive Abilities of Machine Learning Techniques May Be Limited by Dataset Characteristics: Insights From the UNOS Database" presented in the current issue of the Journal of Cardiac Failure that machine learning approaches do not provide significantly higher performance when compared to more traditional statistical approaches in predicting mortality following heart transplant. In this brief report, we provide an insight on the possible reasons for why machine learning methods do not outperform more traditional approaches for every problem and every dataset.

CONCLUSIONS

Most of the performance-focused critics on machine learning are because the bar is set unfairly too high for machine learning. In most cases, machine learning methods provides at least as good results as traditional statistical methods do. It is normal for machine learning models to provide similar performance with linear models if the actual underlying input-outcome relationship is linear. Moreover, machine learning methods outperforms linear statistical models when the underlying input-output relationship is not linear and if the dataset is large enough and include predictors capturing that nonlinear relationship.

Ultrasensitive magnetic resonance imaging of systemic reactive oxygen species in vivo for early diagnosis of sepsis using activatable nanoprobes

Current diagnostic methods for sepsis lack required speed or precision, often failing to make timely accurate diagnosis for early medical treatment. The systemic excess generation of reactive oxygen species (ROS) during sepsis has been considered as an early indicator of sepsis. Herein, we present the rational design of novel activatable nanoprobes (ROS CAs) composed of a clinically approved iron oxide core, Gd-DTPA, and hyaluronic acid (HA) that can image ROS down to sub-micromolar concentrations via magnetic resonance imaging (MRI), and use them as sensitive contrast agents for sepsis evaluation. Such a well-defined nanostructure allows them to undergo ROS-triggered degradation and release Gd-DTPA in the presence of ROS, leading to the recovery of the quenched T 1-weighted MRI signal with fast response. With outstanding sensitivity and unlimited tissue penetration depth, ROS CAs are capable of imaging systemic ROS overproduction in mice with early sepsis. Moreover, by using these well-prepared ROS CAs, the severity of the sepsis can be rapidly evaluated by monitoring the systemic ROS levels in vivo. Overall, the present study will not only provide a new strategy to aid in the early diagnosis and risk assessment of sepsis, but also offer valuable insight for the study of sepsis and ROS biology.

Predicting need for advanced illness or palliative care in a primary care population using electronic health record data

Timely outreach to individuals in an advanced stage of illness offers opportunities to exercise decision control over health care. Predictive models built using Electronic health record (EHR) data are being explored as a way to anticipate such need with enough lead time for patient engagement. Prior studies have focused on hospitalized patients, who typically have more data available for predicting care needs. It is unclear if prediction driven outreach is feasible in the primary care setting. In this study, we apply predictive modeling to the primary care population of a large, regional health system and systematically examine the impact of technical choices, such as requiring a minimum number of health care encounters (data density requirements) and aggregating diagnosis codes using Clinical Classifications Software (CCS) groupings to reduce dimensionality, on model performance in terms of discrimination and positive predictive value. We assembled a cohort of 349,667 primary care patients between 65 and 90 years of age who sought care from Sutter Health between July 1, 2011 and June 30, 2014, of whom 2.1% died during the study period. EHR data comprising demographics, encounters, orders, and diagnoses for each patient from a 12 month observation window prior to the point when a prediction is made were extracted. L1 regularized logistic regression and gradient boosted tree models were fit to training data and tuned by cross validation. Model performance in predicting one year mortality was assessed using held-out test patients. Our experiments systematically varied three factors: model type, diagnosis coding, and data density requirements. We found substantial, consistent benefit from using gradient boosting vs logistic regression (mean AUROC over all other technical choices of 84.8% vs 80.7% respectively). There was no benefit from aggregation of ICD codes into CCS code groups (mean AUROC over all other technical choices of 82.9% vs 82.6% respectively). Likewise increasing data density requirements did not affect discrimination (mean AUROC over other technical choices ranged from 82.5% to 83%). We also examine model performance as a function of lead time, which is the interval between death and when a prediction was made. In subgroup analysis by lead time, mean AUROC over all other choices ranged from 87.9% for patients who died within 0 to 3 months to 83.6% for those who died 9 to 12 months after prediction time.

The Uniformed Services University's Surgical Critical Care Initiative (SC2i): Bringing Precision Medicine to the Critically Ill

Precision medicine endeavors to leverage all available medical data in pursuit of individualized diagnostic and therapeutic plans to improve patient outcomes in a cost-effective manner. Its promise in the field of critical care remains incompletely realized. The Department of Defense has a vested interest in advancing precision medicine for those sent into harm's way and specifically seeks means of individualizing care in the context of complex and highly dynamic combat clinical decision environments. Building on legacy research efforts conducted during the Afghanistan and Iraq conflicts, the Uniformed Service University (USU) launched the Surgical Critical Care Initiative (SC2i) in 2013 to develop clinical- and biomarker-driven Clinical Decision Support Systems (CDSS), with the goals of improving both patient-specific outcomes and resource utilization for conditions with a high risk of morbidity or mortality. Despite technical and regulatory challenges, this military-civilian partnership is beginning to deliver on the promise of personalized care, organizing and analyzing sizable, real-time medical data sets to support complex clinical decision-making across critical and surgical care disciplines. We present the SC2i experience as a generalizable template for the national integration of federal and non-federal research databanks to foster critical and surgical care precision medicine.

Diffusing an Innovation: Clinician Perceptions of Continuous Predictive Analytics Monitoring in Intensive Care

BACKGROUND

The purpose of this article is to describe neonatal intensive care unit clinician perceptions of a continuous predictive analytics technology and how those perceptions influenced clinician adoption. Adopting and integrating new technology into care is notoriously slow and difficult; realizing expected gains remain a challenge.

METHODS

Semistructured interviews from a cross-section of neonatal physicians (n = 14) and nurses (n = 8) from a single U.S. medical center were collected 18 months following the conclusion of the predictive monitoring technology randomized control trial. Following qualitative descriptive analysis, innovation attributes from Diffusion of Innovation Theory-guided thematic development.

RESULTS

Results suggest that the combination of physical location as well as lack of integration into work flow or methods of using data in care decisionmaking may have delayed clinicians from routinely paying attention to the data. Once data were routinely collected, documented, and reported during patient rounds and patient handoffs, clinicians came to view data as another vital sign. Through clinicians' observation of senior physicians and nurses, and ongoing dialogue about data trends and patient status, clinicians learned how to integrate these data in care decision making (e.g., differential diagnosis) and came to value the technology as beneficial to care delivery.

DISCUSSION

The use of newly created predictive technologies that provide early warning of illness may require implementation strategies that acknowledge the risk-benefit of treatment clinicians must balance and take advantage of existing clinician training methods.

The Classification of Minor Gait Alterations Using Wearable Sensors and Deep Learning

OBJECTIVE

This paper describes how non-invasive wearable sensors can be used in combination with deep learning to classify artificially induced gait alterations without the requirement for a medical professional or gait analyst to be present. This approach is motivated by the goal of diagnosing gait abnormalities on a symptom-by-symptom basis, irrespective of other neuromuscular movement disorders the patients may be affected by. This could lead to improvements in treatment and offer a greater insight into movement disorders.

METHODS

In-shoe pressure was measured for 12 able-bodied participants, each subject to eight artificially induced gait alterations, achieved by modifying the underside of the shoe. The data were recorded at 100 Hz over 2520 data channels and were analyzed using the deep learning architecture and the long term short term memory networks. Additionally, the rationale for the decision-making process of these networks was investigated.

CONCLUSION

Long term short term memory networks are applicable to the classification of the gait function. The classifications can be made using only 2 s of sparse data (82.0% accuracy over 96 000 instances of test data) from participants who were not a part of the training set.

SIGNIFICANCE

This paper provides potential for the gait function to be accurately classified using non-invasive techniques, and at more regular intervals, outside of a clinical setting, without the need for healthcare professionals to be present.

Machine learning models for early sepsis recognition in the neonatal intensive care unit using readily available electronic health record data

BACKGROUND

Rapid antibiotic administration is known to improve sepsis outcomes, however early diagnosis remains challenging due to complex presentation. Our objective was to develop a model using readily available electronic health record (EHR) data capable of recognizing infant sepsis at least 4 hours prior to clinical recognition.

METHODS AND FINDINGS

We performed a retrospective case control study of infants hospitalized ≥48 hours in the Neonatal Intensive Care Unit (NICU) at the Children's Hospital of Philadelphia between September 2014 and November 2017 who received at least one sepsis evaluation before 12 months of age. We considered two evaluation outcomes as cases: culture positive-positive blood culture for a known pathogen (110 evaluations); and clinically positive-negative cultures but antibiotics administered for ≥120 hours (265 evaluations). Case data was taken from the 44-hour window ending 4 hours prior to evaluation. We randomly sampled 1,100 44-hour windows of control data from all times ≥10 days removed from any evaluation. Model inputs consisted of up to 36 features derived from routine EHR data. Using 10-fold nested cross-validation, 8 machine learning models were trained to classify inputs as sepsis positive or negative. When tasked with discriminating culture positive cases from controls, 6 models achieved a mean area under the receiver operating characteristic (AUC) between 0.80-0.82 with no significant differences between them. Including both culture and clinically positive cases, the same 6 models achieved an AUC between 0.85-0.87, again with no significant differences.

CONCLUSIONS

Machine learning models can identify infants with sepsis in the NICU hours prior to clinical recognition. Learning curves indicate model improvement may be achieved with additional training examples. Additional input features may also improve performance. Further research is warranted to assess potential performance improvements and clinical efficacy in a prospective trial.

An attention based deep learning model of clinical events in the intensive care unit

This study trained long short-term memory (LSTM) recurrent neural networks (RNNs) incorporating an attention mechanism to predict daily sepsis, myocardial infarction (MI), and vancomycin antibiotic administration over two week patient ICU courses in the MIMIC-III dataset. These models achieved next-day predictive AUC of 0.876 for sepsis, 0.823 for MI, and 0.833 for vancomycin administration. Attention maps built from these models highlighted those times when input variables most influenced predictions and could provide a degree of interpretability to clinicians. These models appeared to attend to variables that were proxies for clinician decision-making, demonstrating a challenge of using flexible deep learning approaches trained with EHR data to build clinical decision support. While continued development and refinement is needed, we believe that such models could one day prove useful in reducing information overload for ICU physicians by providing needed clinical decision support for a variety of clinically important tasks.

Machine-Learning-Based Laboratory Developed Test for the Diagnosis of Sepsis in High-Risk Patients

Sepsis, a dysregulated host response to infection, is a major health burden in terms of both mortality and cost. The difficulties clinicians face in diagnosing sepsis, alongside the insufficiencies of diagnostic biomarkers, motivate the present study. This work develops a machine-learning-based sepsis diagnostic for a high-risk patient group, using a geographically and institutionally diverse collection of nearly 500,000 patient health records. Using only a minimal set of clinical variables, our diagnostics outperform common severity scoring systems and sepsis biomarkers and benefit from being available immediately upon ordering.

Heart rate variability based machine learning models for risk prediction of suspected sepsis patients in the emergency department

Early identification of high-risk septic patients in the emergency department (ED) may guide appropriate management and disposition, thereby improving outcomes. We compared the performance of machine learning models against conventional risk stratification tools, namely the Quick Sequential Organ Failure Assessment (qSOFA), National Early Warning Score (NEWS), Modified Early Warning Score (MEWS), and our previously described Singapore ED Sepsis (SEDS) model, in the prediction of 30-day in-hospital mortality (IHM) among suspected sepsis patients in the ED.Adult patients who presented to Singapore General Hospital (SGH) ED between September 2014 and April 2016, and who met ≥2 of the 4 Systemic Inflammatory Response Syndrome (SIRS) criteria were included. Patient demographics, vital signs and heart rate variability (HRV) measures obtained at triage were used as predictors. Baseline models were created using qSOFA, NEWS, MEWS, and SEDS scores. Candidate models were trained using k-nearest neighbors, random forest, adaptive boosting, gradient boosting and support vector machine. Models were evaluated on F1 score and area under the precision-recall curve (AUPRC).A total of 214 patients were included, of whom 40 (18.7%) met the outcome. Gradient boosting was the best model with a F1 score of 0.50 and AUPRC of 0.35, and performed better than all the baseline comparators (SEDS, F1 0.40, AUPRC 0.22; qSOFA, F1 0.32, AUPRC 0.21; NEWS, F1 0.38, AUPRC 0.28; MEWS, F1 0.30, AUPRC 0.25).A machine learning model can be used to improve prediction of 30-day IHM among suspected sepsis patients in the ED compared to traditional risk stratification tools.

Improving Prediction Performance Using Hierarchical Analysis of Real-Time Data: A Sepsis Case Study

This paper presents a novel method for hierarchical analysis of machine learning algorithms to improve predictions of at risk patients, thus further enabling prompt therapy. Specifically, we develop a multi-layer machine learning approach to analyze continuous, high-frequency data. We illustrate the capabilities of this approach for early identification of patients at risk of sepsis, a potentially life-threatening complication of an infection, using high-frequency (minute-by-minute) physiological data collected from bedside monitors. In our analysis of a cohort of 586 patients, the model obtained from analyzing the output of a previously developed sepsis prediction model resulted in improved outcomes. Specifically, the original model failed to predict 11.76 ± 4.26% of sepsis patients earlier than Systemic Inflammatory Response Syndrome (SIRS) criteria, commonly used to identify patients at risk for rapid physiological deterioration resulting from sepsis. In contrast, the multi-layer model only failed to predict 3.21 ± 3.11% of sepsis patients earlier than SIRS. In addition, sepsis patients were predicted on average 204.87 ± 7.90 minutes earlier than SIRS criteria using the multi-layer model, which can potentially help reduce mortality and morbidity if implemented in the ICU.

Development and Evaluation of a Machine Learning Model for the Early Identification of Patients at Risk for Sepsis

STUDY OBJECTIVE

The Third International Consensus Definitions (Sepsis-3) Task Force recommended the use of the quick Sequential [Sepsis-related] Organ Failure Assessment (qSOFA) score to screen patients for sepsis outside of the ICU. However, subsequent studies raise concerns about the sensitivity of qSOFA as a screening tool. We aim to use machine learning to develop a new sepsis screening tool, the Risk of Sepsis (RoS) score, and compare it with a slate of benchmark sepsis-screening tools, including the Systemic Inflammatory Response Syndrome, Sequential Organ Failure Assessment (SOFA), qSOFA, Modified Early Warning Score, and National Early Warning Score.

METHODS

We used retrospective electronic health record data from adult patients who presented to 49 urban community hospital emergency departments during a 22-month period (N=2,759,529). We used the Rhee clinical surveillance criteria as our standard definition of sepsis and as the primary target for developing our model. The data were randomly split into training and test cohorts to derive and then evaluate the model. A feature selection process was carried out in 3 stages: first, we reviewed existing models for sepsis screening; second, we consulted with local subject matter experts; and third, we used a supervised machine learning called gradient boosting. Key metrics of performance included alert rate, area under the receiver operating characteristic curve, sensitivity, specificity, and precision. Performance was assessed at 1, 3, 6, 12, and 24 hours after an index time.

RESULTS

The RoS score was the most discriminant screening tool at all time thresholds (area under the receiver operating characteristic curve 0.93 to 0.97). Compared with the next most discriminant benchmark (Sequential Organ Failure Assessment), RoS was significantly more sensitive (67.7% versus 49.2% at 1 hour and 84.6% versus 80.4% at 24 hours) and precise (27.6% versus 12.2% at 1 hour and 28.8% versus 11.4% at 24 hours). The sensitivity of qSOFA was relatively low (3.7% at 1 hour and 23.5% at 24 hours).

CONCLUSION

In this retrospective study, RoS was more timely and discriminant than benchmark screening tools, including those recommend by the Sepsis-3 Task Force. Further study is needed to validate the RoS score at independent sites.

High-performance medicine: the convergence of human and artificial intelligence

The use of artificial intelligence, and the deep-learning subtype in particular, has been enabled by the use of labeled big data, along with markedly enhanced computing power and cloud storage, across all sectors. In medicine, this is beginning to have an impact at three levels: for clinicians, predominantly via rapid, accurate image interpretation; for health systems, by improving workflow and the potential for reducing medical errors; and for patients, by enabling them to process their own data to promote health. The current limitations, including bias, privacy and security, and lack of transparency, along with the future directions of these applications will be discussed in this article. Over time, marked improvements in accuracy, productivity, and workflow will likely be actualized, but whether that will be used to improve the patient-doctor relationship or facilitate its erosion remains to be seen.

Pediatric Severe Sepsis Prediction Using Machine Learning

Background: Early detection of pediatric severe sepsis is necessary in order to optimize effective treatment, and new methods are needed to facilitate this early detection. Objective: Can a machine-learning based prediction algorithm using electronic healthcare record (EHR) data predict severe sepsis onset in pediatric populations? Methods: EHR data were collected from a retrospective set of de-identified pediatric inpatient and emergency encounters for patients between 2-17 years of age, drawn from the University of California San Francisco (UCSF) Medical Center, with encounter dates between June 2011 and March 2016. Results: Pediatric patients (n = 9,486) were identified and 101 (1.06%) were labeled with severe sepsis following the pediatric severe sepsis definition of Goldstein et al. (1). In 4-fold cross-validation evaluations, the machine learning algorithm achieved an AUROC of 0.916 for discrimination between severe sepsis and control pediatric patients at the time of onset and AUROC of 0.718 at 4 h before onset. The prediction algorithm significantly outperformed the Pediatric Logistic Organ Dysfunction score (PELOD-2) (p < 0.05) and pediatric Systemic Inflammatory Response Syndrome (SIRS) (p < 0.05) in the prediction of severe sepsis 4 h before onset using cross-validation and pairwise t-tests. Conclusion: This machine learning algorithm has the potential to deliver high-performance severe sepsis detection and prediction through automated monitoring of EHR data for pediatric inpatients, which may enable earlier sepsis recognition and treatment initiation.

Big Data Analysis and Machine Learning in Intensive Care Units

Intensive care is an ideal environment for the use of Big Data Analysis (BDA) and Machine Learning (ML), due to the huge amount of information processed and stored in electronic format in relation to such care. These tools can improve our clinical research capabilities and clinical decision making in the future. The present study reviews the foundations of BDA and ML, and explores possible applications in our field from a clinical viewpoint. We also suggest potential strategies to optimize these new technologies and describe a new kind of hybrid healthcare-data science professional with a linking role between clinicians and data.

A minimal set of physiomarkers in continuous high frequency data streams predict adult sepsis onset earlier

PURPOSE

Sepsis is a life-threatening condition with high mortality rates and expensive treatment costs. To improve short- and long-term outcomes, it is critical to detect at-risk sepsis patients at an early stage.

METHODS

A data-set consisting of high-frequency physiological data from 1161 critically ill patients was analyzed. 377 patients had developed sepsis, and had data at least 3 h prior to the onset of sepsis. A random forest classifier was trained to discriminate between sepsis and non-sepsis patients in real-time using a total of 132 features extracted from a moving time-window. The model was trained on 80% of the patients and was tested on the remaining 20% of the patients, for two observational periods of lengths 3 and 6 h prior to onset.

RESULTS

The model that used continuous physiological data alone resulted in sensitivity and F1 score of up to 80% and 67% one hour before sepsis onset. On average, these models were able to predict sepsis 294.19 ± 6.50 min (5 h) before the onset.

CONCLUSIONS

The use of machine learning algorithms on continuous streams of physiological data can allow for early identification of at-risk patients in real-time with high accuracy.

Impact of Electronic Physician Order-Set on Antibiotic Ordering Time in Septic Patients in the Emergency Department

BACKGROUND

Sepsis is a serious medical condition that can lead to organ dysfunction and death. Research shows that each hour delay in antibiotic administration increases mortality. The Surviving Sepsis Campaign Bundles created standards to assist in the timely treatment of patients with suspected sepsis to improve outcomes and reduce mortality.

OBJECTIVE

This article determines if the use of an electronic physician order-set decreases time to antibiotic ordering for patients with sepsis in the emergency department (ED).

METHODS

A retrospective chart review was performed on adult patients who presented to the ED of four community hospitals from May to July 2016. Patients with severe sepsis and/or septic shock were included. Primary outcome was the difference in time to antibiotic ordering in patients whose physicians utilized the order-set versus those whose physicians did not. Secondary outcomes included differences in time to antibiotic administration, time to lactate test, hospital length of stay, and posthospitalization disposition. The institution's Quality Improvement Committee approved the project.

RESULTS

Forty-five of 123 patients (36.6%) with sepsis had physicians who used the order-set. Order-set utilization reduced the mean time to ordering antibiotics by 20 minutes (99 minutes, 95% confidence interval [CI]: 69-128 vs. 119 minutes, 95% CI: 91-147), but this finding was not statistically significant. Mean time to antibiotic administration (145 minutes, 95% CI: 108-181 vs. 182 minutes, 95% CI: 125-239) and median time to lactate tests (12 minutes, 95% CI: 0-20 vs. 19 minutes, 95% CI: 8-34), although in the direction of the hypotheses, were not significantly different.

CONCLUSION

Utilization of the order-set was associated with a potentially clinically significant, but not statistically significant, reduced time to antibiotic ordering in patients with sepsis. Electronic order-sets are a promising tool to assist hospitals with meeting the Centers for Medicare and Medicaid Services core measure.

Better medicine through machine learning: What's real, and what's artificial?

Machine Learning Special Issue Guest Editors Suchi Saria, Atul Butte, and Aziz Sheikh cut through the hyperbole with an accessible and accurate portrayal of the forefront of machine learning in clinical translation.

Accurate prediction of blood culture outcome in the intensive care unit using long short-term memory neural networks

INTRODUCTION

Blood cultures are often performed in the intensive care unit (ICU) to detect bloodstream infections and identify pathogen type, further guiding treatment. Early detection is essential, as a bloodstream infection can give cause to sepsis, a severe immune response associated with an increased risk of organ failure and death.

PROBLEM STATEMENT

The early clinical detection of a bloodstream infection is challenging but rapid targeted treatment, within the first place antimicrobials, substantially increases survival chances. As blood cultures require time to incubate, early clinical detection using physiological signals combined with indicative lab values is pivotal.

OBJECTIVE

In this work, a novel method is constructed and explored for the potential prediction of the outcome of a blood culture test. The approach is based on a temporal computational model which uses nine clinical parameters measured over time.

METHODOLOGY

We use a bidirectional long short-term memory neural network, a type of recurrent neural network well suited for tasks where the time lag between a predictive event and outcome is unknown. Evaluation is performed using a novel high-quality database consisting of 2177 ICU admissions at the Ghent University Hospital located in Belgium.

RESULTS

The network achieves, on average, an area under the receiver operating characteristic curve of 0.99 and an area under the precision-recall curve of 0.82. In addition, our results show that predicting several hours upfront is possible with only a small decrease in predictive power. In this setting, it outperforms traditional non-temporal, machine learning models.

CONCLUSION

Our proposed computational model accurately predicts the outcome of blood culture tests using nine clinical parameters. Moreover, it can be used in the ICU as an early warning system to detect patients at risk of blood stream infection.

3,3'-Diindolylmethane protects cardiomyocytes from LPS-induced inflammatory response and apoptosis

BACKGROUND

3,3'-Diindolylmethane (DIM) has been extensively studied as a potential therapeutic drug with free radical scavenging, antioxidant and anti-angiogenic effects. However, whether DIM has similar effects on cardiomyocytes remains unknown. Here we evaluated DIM's influence on inflammation and apoptosis of H9C2 cardiomyocytes induced by LPS and to explore the possible mechanism of the effects.

METHODS

H9C2 cells were incubated with DIM (10, 20 and 30 μM) with or without LPS for 24 h. The cytotoxicity of DIM was detected by CCK-8. The levels of tumour necrosis factor (TNF)-α and interleukin (IL)-6 were then measured using RT-qPCR and ELISA. Cell apoptosis rate and reactive oxygen species (ROS) content after DIM treatment were measured by flow cytometry. Expressions of NFκB, P-NFκB, IκBa, P-IκBa, Bax and Bcl-2 after DIM treatment were detected by western blot. The rate of NFκB nuclear translocation after DIM treatment was determined by immunocytochemical analysis.

RESULTS

LPS stimulation promoted TNF-α and IL-6 mRNA expression. After treatment with various concentrations of DIM (10, 20 and 30 μM), TNF-α and IL-6 mRNA expression was clearly impaired, especially in the LPS + DIM30(μM) group. ELISA was used to measure TNF-α and IL-6 concentrations in cellular supernatant, and the result was verified to be consistent with RT-qPCR. Additionally, DIM treatment significantly blocked LPS-induced oxidative stress and inhibited LPS-induced apoptosis in H9C2 cardiomyocytes according to the results detected by flow cytometry. Moreover, compared with LPS alone, DIM significantly inhibited the LPS-induced phosphorylation of NFκB (p-NFκB) and Bax expression and increased Bcl-2 expression.

CONCLUSIONS

DIM may have a protective effect for H9C2 cardiomyocytes against LPS-induced inflammatory response and apoptosis. DIM may be a new insight into the treatment of septic cardiomyopathy.

Applying Artificial Intelligence to Identify Physiomarkers Predicting Severe Sepsis in the PICU

OBJECTIVES

We used artificial intelligence to develop a novel algorithm using physiomarkers to predict the onset of severe sepsis in critically ill children.

DESIGN

Observational cohort study.

SETTING

PICU.

PATIENTS

Children age between 6 and 18 years old.

INTERVENTIONS

None.

MEASUREMENTS AND MAIN RESULTS

Continuous minute-by-minute physiologic data were available for a total of 493 critically ill children admitted to a tertiary care PICU over an 8-month period, 20 of whom developed severe sepsis. Using an alert time stamp generated by an electronic screening algorithm as a reference point, we studied up to 24 prior hours of continuous physiologic data. We identified physiomarkers, including SD of heart rate, systolic and diastolic blood pressure, and symbolic transitions probabilities of those variables that discriminated severe sepsis patients from controls (all other patients admitted to the PICU who did not meet severe sepsis criteria). We used logistic regression, random forests, and deep Convolutional Neural Network methods to derive our models. Analysis was performed using data generated in two windows prior to the firing of the electronic screening algorithm, namely, 2-8 and 8-24 hours. When analyzing the physiomarkers present in the 2-8 hours analysis window, logistic regression performed with specificity of 87.4% and sensitivity of 55.0%, random forest performed with 79.6% specificity and 80.0% sensitivity, and the Convolutional Neural Network performed with 83.0% specificity and 75.0% sensitivity. When analyzing physiomarkers from the 8-24 hours window, logistic regression resulted in 77.1% specificity and 39.3% sensitivity, random forest performed with 82.3% specificity and 61.1% sensitivity, whereas the Convolutional Neural Network method achieved 81% specificity and 76% sensitivity.

CONCLUSIONS

Artificial intelligence can be used to predict the onset of severe sepsis using physiomarkers in critically ill children. Further, it may detect severe sepsis as early as 8 hours prior to a real-time electronic severe sepsis screening algorithm.

Explainable machine-learning predictions for the prevention of hypoxaemia during surgery

Although anaesthesiologists strive to avoid hypoxemia during surgery, reliably predicting future intraoperative hypoxemia is not currently possible. Here, we report the development and testing of a machine-learning-based system that, in real time during general anaesthesia, predicts the risk of hypoxemia and provides explanations of the risk factors. The system, which was trained on minute-by-minute data from the electronic medical records of over fifty thousand surgeries, improved the performance of anaesthesiologists when providing interpretable hypoxemia risks and contributing factors. The explanations for the predictions are broadly consistent with the literature and with prior knowledge from anaesthesiologists. Our results suggest that if anaesthesiologists currently anticipate 15% of hypoxemia events, with this system's assistance they would anticipate 30% of them, a large portion of which may benefit from early intervention because they are associated with modifiable factors. The system can help improve the clinical understanding of hypoxemia risk during anaesthesia care by providing general insights into the exact changes in risk induced by certain patient or procedure characteristics.

Using survival analysis to predict septic shock onset in ICU patients

PURPOSE

To determine the efficacy of survival analysis for predicting septic shock onset in ICU patients.

MATERIALS AND METHODS

We performed a retrospective analysis on ICU cases from Mercy Hospital St. Louis from 2012 to 2016. As part of the procedure for inclusion in the Apache Outcomes database, each case is reviewed by critical care clinicians to identify septic shock patients as well as the time of septic shock onset. We used survival analysis to predict septic shock onset in these cases and employed lagging to compensate for uncertainties in septic shock onset time.

RESULTS

Survival analysis was highly effective at predicting septic shock onset, producing AUC values of >0.87. The methodology was robust to lag times as well as the specific method of survival analysis used.

CONCLUSIONS

This methodology has the potential to be implemented in the ICU for real time prediction and can be used as a building block to expand the approach to other hospital wards or care environments.

Assessing patient risk of central line-associated bacteremia via machine learning

BACKGROUND

Central line-associated bloodstream infections (CLABSIs) contribute to increased morbidity, length of hospital stay, and cost. Despite progress in understanding the risk factors, there remains a need to accurately predict the risk of CLABSIs and, in real time, prevent them from occurring.

METHODS

A predictive model was developed using retrospective data from a large academic healthcare system. Models were developed with machine learning via construction of random forests using validated input variables.

RESULTS

Fifteen variables accounted for the most significant effect on CLABSI prediction based on a retrospective study of 70,218 unique patient encounters between January 1, 2013, and May 31, 2016. The area under the receiver operating characteristic curve for the best-performing model was 0.82 in production.

DISCUSSION

This model has multiple applications for resource allocation for CLABSI prevention, including serving as a tool to target patients at highest risk for potentially cost-effective but otherwise time-limited interventions.

CONCLUSIONS

Machine learning can be used to develop accurate models to predict the risk of CLABSI in real time prior to the development of infection.

Innovation at the Intersection of Clinical Trials and Real-World Data Science to Advance Patient Care

While efficacy and safety data collected from randomized clinical trials are the evidentiary standard for determining market authorization, this alone may no longer be sufficient to address the needs of key stakeholders (regulators, providers, and payers) and guarantee long-term success of pharmaceutical products. There is a heightened interest from stakeholders on understanding the use of real-world evidence (RWE) to substantiate benefit-risk assessment and support the value of a new drug. This review provides an overview of real-world data (RWD) and related advances in the regulatory framework, and discusses their impact on clinical research and development. A framework for linking drug development decisions with the value proposition of the drug, utilizing pharmacokinetic-pharmacodynamic-pharmacoeconomic models, is introduced. The summary presented here is based on the presentations and discussion at the symposium entitled Innovation at the Intersection of Clinical Trials and Real-World Data to Advance Patient Care at the American Society for Clinical Pharmacology and Therapeutics (ASCPT) 2017 Annual Meeting.

Impact of predictive analytics based on continuous cardiorespiratory monitoring in a surgical and trauma intensive care unit

Predictive analytics monitoring, the use of patient data to provide continuous risk estimation of deterioration, is a promising new application of big data analytical techniques to the care of individual patients. We tested the hypothesis that continuous display of novel electronic risk visualization of respiratory and cardiovascular events would impact intensive care unit (ICU) patient outcomes. In an adult tertiary care surgical trauma ICU, we displayed risk estimation visualizations on a large monitor, but in the medical ICU in the same institution we did not. The risk estimates were based solely on analysis of continuous cardiorespiratory monitoring. We examined 4275 individual patient records within a 7 month time period preceding and following data display. We determined cases of septic shock, emergency intubation, hemorrhage, and death to compare rates per patient care pre-and post-implementation. Following implementation, the incidence of septic shock fell by half (p < 0.01 in a multivariate model that included age and APACHE) in the surgical trauma ICU, where the data were continuously on display, but by only 10% (p = NS) in the control Medical ICU. There were no significant changes in the other outcomes. Display of a predictive analytics monitor based on continuous cardiorespiratory monitoring was followed by a reduction in the rate of septic shock, even when controlling for age and APACHE score.

Scalable Joint Models for Reliable Uncertainty-Aware Event Prediction

Missing data and noisy observations pose significant challenges for reliably predicting events from irregularly sampled multivariate time series (longitudinal) data. Imputation methods, which are typically used for completing the data prior to event prediction, lack a principled mechanism to account for the uncertainty due to missingness. Alternatively, state-of-the-art joint modeling techniques can be used for jointly modeling the longitudinal and event data and compute event probabilities conditioned on the longitudinal observations. These approaches, however, make strong parametric assumptions and do not easily scale to multivariate signals with many observations. Our proposed approach consists of several key innovations. First, we develop a flexible and scalable joint model based upon sparse multiple-output Gaussian processes. Unlike state-of-the-art joint models, the proposed model can explain highly challenging structure including non-Gaussian noise while scaling to large data. Second, we derive an optimal policy for predicting events using the distribution of the event occurrence estimated by the joint model. The derived policy trades-off the cost of a delayed detection versus incorrect assessments and abstains from making decisions when the estimated event probability does not satisfy the derived confidence criteria. Experiments on a large dataset show that the proposed framework significantly outperforms state-of-the-art techniques in event prediction.

Smart Medical Information Technology for Healthcare (SMITH)

INTRODUCTION

This article is part of the Focus Theme of Methods of Information in Medicine on the German Medical Informatics Initiative. "Smart Medical Information Technology for Healthcare (SMITH)" is one of four consortia funded by the German Medical Informatics Initiative (MI-I) to create an alliance of universities, university hospitals, research institutions and IT companies. SMITH's goals are to establish Data Integration Centers (DICs) at each SMITH partner hospital and to implement use cases which demonstrate the usefulness of the approach.

OBJECTIVES

To give insight into architectural design issues underlying SMITH data integration and to introduce the use cases to be implemented.

GOVERNANCE AND POLICIES

SMITH implements a federated approach as well for its governance structure as for its information system architecture. SMITH has designed a generic concept for its data integration centers. They share identical services and functionalities to take best advantage of the interoperability architectures and of the data use and access process planned. The DICs provide access to the local hospitals' Electronic Medical Records (EMR). This is based on data trustee and privacy management services. DIC staff will curate and amend EMR data in the Health Data Storage.

METHODOLOGY AND ARCHITECTURAL FRAMEWORK

To share medical and research data, SMITH's information system is based on communication and storage standards. We use the Reference Model of the Open Archival Information System and will consistently implement profiles of Integrating the Health Care Enterprise (IHE) and Health Level Seven (HL7) standards. Standard terminologies will be applied. The SMITH Market Place will be used for devising agreements on data access and distribution. 3LGM² for enterprise architecture modeling supports a consistent development process.The DIC reference architecture determines the services, applications and the standardsbased communication links needed for efficiently supporting the ingesting, data nourishing, trustee, privacy management and data transfer tasks of the SMITH DICs. The reference architecture is adopted at the local sites. Data sharing services and the market place enable interoperability.

USE CASES

The methodological use case "Phenotype Pipeline" (PheP) constructs algorithms for annotations and analyses of patient-related phenotypes according to classification rules or statistical models based on structured data. Unstructured textual data will be subject to natural language processing to permit integration into the phenotyping algorithms. The clinical use case "Algorithmic Surveillance of ICU Patients" (ASIC) focusses on patients in Intensive Care Units (ICU) with the acute respiratory distress syndrome (ARDS). A model-based decision-support system will give advice for mechanical ventilation. The clinical use case HELP develops a "hospital-wide electronic medical record-based computerized decision support system to improve outcomes of patients with blood-stream infections" (HELP). ASIC and HELP use the PheP. The clinical benefit of the use cases ASIC and HELP will be demonstrated in a change of care clinical trial based on a step wedge design.

DISCUSSION

SMITH's strength is the modular, reusable IT architecture based on interoperability standards, the integration of the hospitals' information management departments and the public-private partnership. The project aims at sustainability beyond the first 4-year funding period.

Surviving sepsis campaign: research priorities for sepsis and septic shock

Objective

To identify research priorities in the management, epidemiology, outcome and underlying causes of sepsis and septic shock.

Design

A consensus committee of 16 international experts representing the European Society of Intensive Care Medicine and Society of Critical Care Medicine was convened at the annual meetings of both societies. Subgroups had teleconference and electronic-based discussion. The entire committee iteratively developed the entire document and recommendations.

Methods

Each committee member independently gave their top five priorities for sepsis research. A total of 88 suggestions (ESM 1 - supplemental table 1) were grouped into categories by the committee co-chairs, leading to the formation of seven subgroups: infection, fluids and vasoactive agents, adjunctive therapy, administration/epidemiology, scoring/identification, post-intensive care unit, and basic/translational science. Each subgroup had teleconferences to go over each priority followed by formal voting within each subgroup. The entire committee also voted on top priorities across all subgroups except for basic/translational science.

Results

The Surviving Sepsis Research Committee provides 26 priorities for sepsis and septic shock. Of these, the top six clinical priorities were identified and include the following questions: (1) can targeted/personalized/precision medicine approaches determine which therapies will work for which patients at which times?; (2) what are ideal endpoints for volume resuscitation and how should volume resuscitation be titrated?; (3) should rapid diagnostic tests be implemented in clinical practice?; (4) should empiric antibiotic combination therapy be used in sepsis or septic shock?; (5) what are the predictors of sepsis long-term morbidity and mortality?; and (6) what information identifies organ dysfunction?

Conclusions

While the Surviving Sepsis Campaign guidelines give multiple recommendations on the treatment of sepsis, significant knowledge gaps remain, both in bedside issues directly applicable to clinicians, as well as understanding the fundamental mechanisms underlying the development and progression of sepsis. The priorities identified represent a roadmap for research in sepsis and septic shock.

Electronic supplementary material

The online version of this article (10.1007/s00134-018-5175-z) contains supplementary material, which is available to authorized users.

Big Data and Data Science in Critical Care

The digitalization of the health-care system has resulted in a deluge of clinical big data and has prompted the rapid growth of data science in medicine. Data science, which is the field of study dedicated to the principled extraction of knowledge from complex data, is particularly relevant in the critical care setting. The availability of large amounts of data in the ICU, the need for better evidence-based care, and the complexity of critical illness makes the use of data science techniques and data-driven research particularly appealing to intensivists. Despite the increasing number of studies and publications in the field, thus far there have been few examples of data science projects that have resulted in successful implementations of data-driven systems in the ICU. However, given the expected growth in the field, intensivists should be familiar with the opportunities and challenges of big data and data science. The present article reviews the definitions, types of algorithms, applications, challenges, and future of big data and data science in critical care.

MicroRNA-326 aggravates acute lung injury in septic shock by mediating the NF-κB signaling pathway

Our previous studies have demonstrated that the activation of the nuclear factor-kappa B (NF-κB) signaling pathway contributes to the development of lipopolysaccharide (LPS)-induced acute lung injury (ALI) as well as an inflammatory reaction, and its inhibition may provide future therapeutic values. Thereby, this study aims to explore the effects of miR-326 on inflammatory response and ALI in mice with septic shock via the NF-κB signaling pathway. The study included normal mice and LPS-induced mouse models of septic shock with ALI. Modeled mice were transfected with the blank plasmid, miR-326 mimic, miR-326 inhibitor, si-BCL2A1 and miR-326 inhibitor + si-BCL2A1. Mean arterial pressure (MAP), airway pressure (AP), heart rate (HR) and lung wet dry (W/D) ratio were determined. Serum levels of interleukin (IL)-6, IL-10, IL-1β, and tumor necrosis factor-α (TNF-α) were detected using ELISA. Reverse transcription quantitative polymerase chain reaction (RT-qPCR) and Western blot analysis were performed to detect the miR-326 expression and expression levels of BCL2A1, related genes of inflammatory response and the NF-κB signaling pathway in lung tissues. Cell viability and apoptosis were measured using the CCK-8 assay and flow cytometry, respectively. Compared to the ALI models and those transfected with blank plasmid, the up-regulated miR-326 expression and silenced BCL2A1 lead to decreased levels of MAP, increased AP, HR and lung W/D, increased serum levels of IL-6, IL-10, IL-1β and TNF-α, increased expressions of IL-6, IL-1β, TNF-α, NF-κB p65 (p-NF-κB p65), and iNOS with decreased expressions of BCL2A1s as well as inhibition of cell viability and enhanced cell apoptosis; the down-regulated miR-326 expression reversed the aforementioned situation. MiR-326 targeting the BCL2A1 gene activated the NF-κB signaling pathway, resulting in aggravated inflammatory response and lung injury of septic shock with ALI in mice.

Emerging Technologies for Molecular Diagnosis of Sepsis

Rapid and accurate profiling of infection-causing pathogens remains a significant challenge in modern health care. Despite advances in molecular diagnostic techniques, blood culture analysis remains the gold standard for diagnosing sepsis. However, this method is too slow and cumbersome to significantly influence the initial management of patients. The swift initiation of precise and targeted antibiotic therapies depends on the ability of a sepsis diagnostic test to capture clinically relevant organisms along with antimicrobial resistance within 1 to 3 h. The administration of appropriate, narrow-spectrum antibiotics demands that such a test be extremely sensitive with a high negative predictive value. In addition, it should utilize small sample volumes and detect polymicrobial infections and contaminants. All of this must be accomplished with a platform that is easily integrated into the clinical workflow. In this review, we outline the limitations of routine blood culture testing and discuss how emerging sepsis technologies are converging on the characteristics of the ideal sepsis diagnostic test. We include seven molecular technologies that have been validated on clinical blood specimens or mock samples using human blood. In addition, we discuss advances in machine learning technologies that use electronic medical record data to provide contextual evaluation support for clinical decision-making.

Development and validation of a prediction model for insulin-associated hypoglycemia in non-critically ill hospitalized adults

OBJECTIVE

To develop and validate a multivariable prediction model for insulin-associated hypoglycemia in non-critically ill hospitalized adults.

RESEARCH DESIGN AND METHODS

We collected pharmacologic, demographic, laboratory, and diagnostic data from 128 657 inpatient days in which at least 1 unit of subcutaneous insulin was administered in the absence of intravenous insulin, total parenteral nutrition, or insulin pump use (index days). These data were used to develop multivariable prediction models for biochemical and clinically significant hypoglycemia (blood glucose (BG) of ≤70 mg/dL and <54 mg/dL, respectively) occurring within 24 hours of the index day. Split-sample internal validation was performed, with 70% and 30% of index days used for model development and validation, respectively.

RESULTS

Using predictors of age, weight, admitting service, insulin doses, mean BG, nadir BG, BG coefficient of variation (CVBG), diet status, type 1 diabetes, type 2 diabetes, acute kidney injury, chronic kidney disease (CKD), liver disease, and digestive disease, our model achieved a c-statistic of 0.77 (95% CI 0.75 to 0.78), positive likelihood ratio (+LR) of 3.5 (95% CI 3.4 to 3.6) and negative likelihood ratio (-LR) of 0.32 (95% CI 0.30 to 0.35) for prediction of biochemical hypoglycemia. Using predictors of sex, weight, insulin doses, mean BG, nadir BG, CVBG, diet status, type 1 diabetes, type 2 diabetes, CKD stage, and steroid use, our model achieved a c-statistic of 0.80 (95% CI 0.78 to 0.82), +LR of 3.8 (95% CI 3.7 to 4.0) and -LR of 0.2 (95% CI 0.2 to 0.3) for prediction of clinically significant hypoglycemia.

CONCLUSIONS

Hospitalized patients at risk of insulin-associated hypoglycemia can be identified using validated prediction models, which may support the development of real-time preventive interventions.

Multicentre validation of a sepsis prediction algorithm using only vital sign data in the emergency department, general ward and ICU

OBJECTIVES

We validate a machine learning-based sepsis-prediction algorithm (InSight) for the detection and prediction of three sepsis-related gold standards, using only six vital signs. We evaluate robustness to missing data, customisation to site-specific data using transfer learning and generalisability to new settings.

DESIGN

A machine-learning algorithm with gradient tree boosting. Features for prediction were created from combinations of six vital sign measurements and their changes over time.

SETTING

A mixed-ward retrospective dataset from the University of California, San Francisco (UCSF) Medical Center (San Francisco, California, USA) as the primary source, an intensive care unit dataset from the Beth Israel Deaconess Medical Center (Boston, Massachusetts, USA) as a transfer-learning source and four additional institutions' datasets to evaluate generalisability.

PARTICIPANTS

684 443 total encounters, with 90 353 encounters from June 2011 to March 2016 at UCSF.

INTERVENTIONS

None.

PRIMARY AND SECONDARY OUTCOME MEASURES

Area under the receiver operating characteristic (AUROC) curve for detection and prediction of sepsis, severe sepsis and septic shock.

RESULTS

For detection of sepsis and severe sepsis, InSight achieves an AUROC curve of 0.92 (95% CI 0.90 to 0.93) and 0.87 (95% CI 0.86 to 0.88), respectively. Four hours before onset, InSight predicts septic shock with an AUROC of 0.96 (95% CI 0.94 to 0.98) and severe sepsis with an AUROC of 0.85 (95% CI 0.79 to 0.91).

CONCLUSIONS

InSight outperforms existing sepsis scoring systems in identifying and predicting sepsis, severe sepsis and septic shock. This is the first sepsis screening system to exceed an AUROC of 0.90 using only vital sign inputs. InSight is robust to missing data, can be customised to novel hospital data using a small fraction of site data and retains strong discrimination across all institutions.