Sharing ICU Patient Data Responsibly Under the Society of Critical Care Medicine/European Society of Intensive Care Medicine Joint Data Science Collaboration: The Amsterdam University Medical Centers Database (AmsterdamUMCdb) Example.

OBJECTIVES: Critical care medicine is a natural environment for machine learning approaches to improve outcomes for critically ill patients as admissions to ICUs generate vast amounts of data. However, technical, legal, ethical, and privacy concerns have so far limited the critical care medicine community from making these data readily available. The Society of Critical Care Medicine and the European Society of Intensive Care Medicine have identified ICU patient data sharing as one of the priorities under their Joint Data Science Collaboration. To encourage ICUs worldwide to share their patient data responsibly, we now describe the development and release of Amsterdam University Medical Centers Database (AmsterdamUMCdb), the first freely available critical care database in full compliance with privacy laws from both the United States and Europe, as an example of the feasibility of sharing complex critical care data. SETTING: University hospital ICU. SUBJECTS: Data from ICU patients admitted between 2003 and 2016. INTERVENTIONS: We used a risk-based deidentification strategy to maintain data utility while preserving privacy. In addition, we implemented contractual and governance processes, and a communication strategy. Patient organizations, supporting hospitals, and experts on ethics and privacy audited these processes and the database. MEASUREMENTS AND MAIN RESULTS: AmsterdamUMCdb contains approximately 1 billion clinical data points from 23,106 admissions of 20,109 patients. The privacy audit concluded that reidentification is not reasonably likely, and AmsterdamUMCdb can therefore be considered as anonymous information, both in the context of the U.S. Health Insurance Portability and Accountability Act and the European General Data Protection Regulation. The ethics audit concluded that responsible data sharing imposes minimal burden, whereas the potential benefit is tremendous. CONCLUSIONS: Technical, legal, ethical, and privacy challenges related to responsible data sharing can be addressed using a multidisciplinary approach. A risk-based deidentification strategy, that complies with both U.S. and European privacy regulations, should be the preferred approach to releasing ICU patient data. This supports the shared Society of Critical Care Medicine and European Society of Intensive Care Medicine vision to improve critical care outcomes through scientific inquiry of vast and combined ICU datasets.

Automated peritoneal lavage: an extremely rapid and safe way to induce hypothermia in post-resuscitation patients.

INTRODUCTION: Mild therapeutic hypothermia (MTH) is a worldwide used therapy to improve neurological outcome in patients successfully resuscitated after cardiac arrest (CA). Preclinical data suggest that timing and speed of induction are related to reduction of secondary brain damage and improved outcome. METHODS: Aiming at a rapid induction and stable maintenance phase, MTH induced via continuous peritoneal lavage (PL) using the Velomedix Inc. automated PL system was evaluated and compared to historical controls in which hypothermia was achieved using cooled saline intravenous infusions and cooled blankets. RESULTS: In 16 PL patients, time to reach the core target temperature of 32.5°C was 30 minutes (interquartile range (IQR): 19 to 60), which was significantly faster compare to 150 minutes (IQR: 112 to 240) in controls. The median rate of cooling during the induction phase in the PL group of 4.1°C/h (IQR: 2.2 to 8.2) was significantly faster compared to 0.9°C/h (IQR: 0.5 to 1.3) in controls. During the 24-hour maintenance phase mean core temperature in the PL patients was 32.38 ± 0.18°C (range: 32.03 to 32.69°C) and in control patients 32.46 ± 0.48°C (range: 31.20 to 33.63°C), indicating more steady temperature control in the PL group compared to controls. Furthermore, the coefficient of variation (VC) for temperature during the maintenance phase was lower in the PL group (VC: 0.5%) compared to the control group (VC: 1.5%). In contrast to 23% of the control patients, none of the PL patients showed an overshoot of hypothermia below 31°C during the maintenance phase. Survival and neurological outcome was not different between the two groups. Neither shivering nor complications related to insertion or use of the PL method were observed. CONCLUSIONS: Using PL in post-CA patients results in a rapidly reached target temperature and a very precise maintenance, unprecedented in clinical studies evaluating MTH techniques. This opens the way to investigate the effects on neurological outcome and survival of ultra-rapid cooling compared to standard cooling in controlled trials in various patient groups. TRIAL REGISTRATION: ClinicalTrials.gov: NCT01016236

Explainable Machine Learning on AmsterdamUMCdb for ICU Discharge Decision Support: Uniting Intensivists and Data Scientists.

Unexpected ICU readmission is associated with longer length of stay and increased mortality. To prevent ICU readmission and death after ICU discharge, our team of intensivists and data scientists aimed to use AmsterdamUMCdb to develop an explainable machine learning-based real-time bedside decision support tool. DERIVATION COHORT: Data from patients admitted to a mixed surgical-medical academic medical center ICU from 2004 to 2016. VALIDATION COHORT: Data from 2016 to 2019 from the same center. PREDICTION MODEL: Patient characteristics, clinical observations, physiologic measurements, laboratory studies, and treatment data were considered as model features. Different supervised learning algorithms were trained to predict ICU readmission and/or death, both within 7 days from ICU discharge, using 10-fold cross-validation. Feature importance was determined using SHapley Additive exPlanations, and readmission probability-time curves were constructed to identify subgroups. Explainability was established by presenting individualized risk trends and feature importance. RESULTS: Our final derivation dataset included 14,105 admissions. The combined readmission/mortality rate within 7 days of ICU discharge was 5.3%. Using Gradient Boosting, the model achieved an area under the receiver operating characteristic curve of 0.78 (95% CI, 0.75-0.81) and an area under the precision-recall curve of 0.19 on the validation cohort (n = 3,929). The most predictive features included common physiologic parameters but also less apparent variables like nutritional support. At a 6% risk threshold, the model showed a sensitivity (recall) of 0.72, specificity of 0.70, and a positive predictive value (precision) of 0.15. Impact analysis using probability-time curves and the 6% risk threshold identified specific patient groups at risk and the potential of a change in discharge management to reduce relative risk by 14%. CONCLUSIONS: We developed an explainable machine learning model that may aid in identifying patients at high risk for readmission and mortality after ICU discharge using the first freely available European critical care database, AmsterdamUMCdb. Impact analysis showed that a relative risk reduction of 14% could be achievable, which might have significant impact on patients and society. ICU data sharing facilitates collaboration between intensivists and data scientists to accelerate model development.

Right Dose, Right Now: Development of AutoKinetics for Real Time Model Informed Precision Antibiotic Dosing Decision Support at the Bedside of Critically Ill Patients.

INTRODUCTION: Antibiotic dosing in critically ill patients is challenging because their pharmacokinetics (PK) are altered and may change rapidly with disease progression. Standard dosing frequently leads to inadequate PK exposure. Therapeutic drug monitoring (TDM) offers a potential solution but requires sampling and PK knowledge, which delays decision support. It is our philosophy that antibiotic dosing support should be directly available at the bedside through deep integration into the electronic health record (EHR) system. Therefore we developed AutoKinetics, a clinical decision support system (CDSS) for real time, model informed precision antibiotic dosing. OBJECTIVE: To provide a detailed description of the design, development, validation, testing, and implementation of AutoKinetics. METHODS: We created a development framework and used workflow analysis to facilitate integration into popular EHR systems. We used a development cycle to iteratively adjust and expand AutoKinetics functionalities. Furthermore, we performed a literature review to select and integrate pharmacokinetic models for five frequently prescribed antibiotics for sepsis. Finally, we tackled regulatory challenges, in particular those related to the Medical Device Regulation under the European regulatory framework. RESULTS: We developed a SQL-based relational database as the backend of AutoKinetics. We developed a data loader to retrieve data in real time. We designed a clinical dosing algorithm to find a dose regimen to maintain antibiotic pharmacokinetic exposure within clinically relevant safety constraints. If needed, a loading dose is calculated to minimize the time until steady state is achieved. Finally, adaptive dosing using Bayesian estimation is applied if plasma levels are available. We implemented support for five extensively used antibiotics following model development, calibration, and validation. We integrated AutoKinetics into two popular EHRs (Metavision, Epic) and developed a user interface that provides textual and visual feedback to the physician. CONCLUSION: We successfully developed a CDSS for real time model informed precision antibiotic dosing at the bedside of the critically ill. This holds great promise for improving sepsis outcome. Therefore, we recently started the Right Dose Right Now multi-center randomized control trial to validate this concept in 420 patients with severe sepsis and septic shock.

Normal saline to dilute parenteral drugs and to keep catheters open is a major and preventable source of hypernatremia acquired in the intensive care unit.

PURPOSE: We wanted to identify modifiable risk factors for intensive care unit (ICU)-acquired hypernatremia. MATERIALS AND METHODS: We retrospectively studied sodium and fluid loads and balances up to 7 days prior to the development of hypernatremia (first serum sodium concentration, [Na+], >150 mmol/L; H) vs control (maximum [Na+] ≤150 mmol/L; N), in consecutive patients admitted into the ICU with a normal serum sodium (<145 mmol/L) and without cerebral disease, within a period of 8 months. RESULTS: There were 57 H and 150 N patients. Severity of disease and organ failure was greater, and length of stay and mechanical ventilation in the ICU were longer in H (P<.001), with a mortality rate of 28% vs 16% in N (P=.002). Sodium input was higher in H than in N, particularly from 0.9% saline to dissolve drugs for infusion and to keep catheters open during the week prior to the first day of hypernatremia (P<.001). Fluid balances were positive and did not differ from N on most days in the presence of slightly higher plasma creatinine and more frequent administration of furosemide, at higher doses, in H than in N. CONCLUSIONS: High sodium input by 0.9% saline used to dilute drugs and keep catheters open is a modifiable risk factor for ICU-acquired H. Dissolving drugs in dextrose 5% may partially prevent potentially harmful sodium overloading and H.

Optimal protein and energy nutrition decreases mortality in mechanically ventilated, critically ill patients: a prospective observational cohort study.

BACKGROUND: Optimal nutrition for patients in the intensive care unit has been proposed to be the provision of energy as determined by indirect calorimetry and the provision of protein of at least 1.2 g/kg. METHODS: Prospective observational cohort study in a mixed medical-surgical intensive care unit in an academic hospital. In total, 886 consecutive mechanically ventilated patients were included. Nutrition was guided by indirect calorimetry and protein provision of at least 1.2 g/kg. Cumulative intakes were calculated for the period of mechanical ventilation. Cox regression was used to analyze the effect of protein + energy target achieved or energy target achieved versus neither target achieved on 28-day mortality, with adjustments for sex, age, body mass index, Acute Physiology and Chronic Health Evaluation II, diagnosis, and hyperglycemic index. RESULTS: Patients' mean age was 63 ± 16 years; body mass index, 26 ± 6; and Acute Physiology and Chronic Health Evaluation II, 23 ± 8. For neither target, energy target, and protein + energy target, energy intake was 75% ± 15%, 96% ± 5%, and 99% ± 5% of target, and protein intake was 72% ± 20%, 89% ± 10%, and 112% ± 12% of target, respectively. Hazard ratios (95% confidence interval) for energy target and protein + energy target were 0.83 (0.67-1.01) and 0.47 (0.31-0.73) for 28-day mortality. CONCLUSIONS: Optimal nutritional therapy in mechanically ventilated, critically ill patients, defined as protein and energy targets reached, is associated with a decrease in 28-day mortality by 50%, whereas only reaching energy targets is not associated with a reduction in mortality.