Multiparameter Monitoring in the ICU: Applications and Pitfalls - A Comprehensive Review
Dr Neeraj Manikath , claude.ai
Abstract
Multiparameter monitoring has become a cornerstone of modern intensive care medicine, allowing clinicians to track multiple physiological parameters simultaneously and detect critical changes in patient status promptly. While technological advancements have expanded the range and sophistication of monitoring capabilities, the interpretation of the vast amount of data generated presents significant challenges. This review synthesizes current evidence on the applications and limitations of multiparameter monitoring in the intensive care unit (ICU), focusing on integration of data streams, alarm fatigue, evidence-based utilization, and emerging technologies. Special attention is given to the balance between technological capabilities and clinical judgment, with recommendations for optimizing monitoring strategies for various critical care scenarios. Understanding both the utility and limitations of these systems is crucial for postgraduate medical education and for enhancing patient outcomes in critical care settings.
Keywords: Intensive care; patient monitoring; hemodynamic monitoring; alarm fatigue; clinical decision support systems; critical care technology
1. Introduction
The evolution of intensive care medicine has been closely linked to advancements in monitoring technologies. From the early days of manual vital signs recording to today's sophisticated, continuous multiparameter monitoring systems, the ability to track physiological variables has transformed critical care practice. Modern ICUs are equipped with an array of monitoring devices that continuously assess cardiovascular function, respiratory parameters, neurological status, and metabolic indices (Vincent et al., 2018).
The fundamental premise underlying intensive monitoring is that early detection of physiological derangements facilitates timely intervention, potentially preventing deterioration and improving outcomes. However, the relationship between monitoring capabilities and patient outcomes is complex. While some monitoring modalities have clearly demonstrated benefits, others remain controversial despite widespread adoption (Saugel et al., 2020).
This review aims to examine the applications and pitfalls of multiparameter monitoring in the ICU setting, with particular emphasis on how postgraduate medical trainees can navigate the complexities of data interpretation and integration. We will explore both established and emerging monitoring technologies, examine the evidence supporting their use, and discuss strategies for mitigating common pitfalls.
2. Core Monitoring Parameters in the ICU
2.1 Hemodynamic Monitoring
2.1.1 Arterial Blood Pressure
Arterial blood pressure remains a fundamental parameter in ICU monitoring, with options ranging from non-invasive oscillometric techniques to direct invasive arterial monitoring. Invasive arterial monitoring provides continuous beat-to-beat measurements and allows for arterial blood sampling, but carries risks including infection, thrombosis, and distal ischemia (Bartels et al., 2020).
The accuracy of non-invasive blood pressure measurements decreases in hypotensive or vasopressor-dependent patients, with studies demonstrating clinically significant discrepancies compared to invasive measurements in these populations (Lehman et al., 2022). This highlights the need for careful selection of monitoring modalities based on patient condition and expected clinical course.
2.1.2 Cardiac Output Monitoring
Contemporary cardiac output monitoring encompasses a spectrum of technologies, from the pulmonary artery catheter (PAC) to less invasive methods such as pulse contour analysis, esophageal Doppler, and bioimpedance/bioreactance techniques.
The PAC provides detailed hemodynamic data including cardiac output, pulmonary artery pressures, and mixed venous oxygen saturation. Despite its comprehensive capabilities, large randomized trials have failed to demonstrate mortality benefits (Rajaram et al., 2020). This has led to declining use of PACs and increased adoption of less invasive alternatives.
Pulse contour analysis systems derive cardiac output from the arterial pressure waveform but may lose accuracy during rapid hemodynamic changes or in patients with significant arrhythmias (Monnet & Teboul, 2017). Ultrasonographic methods, including transthoracic and transesophageal echocardiography, provide valuable structural and functional information but are operator-dependent and generally intermittent rather than continuous (Orde et al., 2018).
2.1.3 Volumetric Indices and Fluid Responsiveness
Dynamic parameters of fluid responsiveness, such as pulse pressure variation (PPV) and stroke volume variation (SVV), have gained prominence for guiding fluid management. These parameters predict the likelihood of cardiac output increase in response to fluid administration but have important limitations:
- Require patients to be mechanically ventilated with regular tidal volumes
- Limited utility in patients with spontaneous breathing activity or arrhythmias
- Affected by chest wall compliance and right ventricular dysfunction
Passive leg raising combined with cardiac output measurement provides an alternative method for predicting fluid responsiveness without these limitations but requires real-time cardiac output monitoring (Monnet et al., 2016).
2.2 Respiratory Monitoring
2.2.1 Oxygen Saturation and Gas Exchange
Pulse oximetry provides continuous, non-invasive monitoring of peripheral oxygen saturation (SpO₂) but may be inaccurate in states of poor peripheral perfusion, severe anemia, or carbon monoxide poisoning. Arterial blood gas analysis remains the gold standard for assessing oxygenation and acid-base status but provides only intermittent measurements (Bitterman, 2019).
Continuous monitoring of end-tidal carbon dioxide (EtCO₂) through capnography offers insights into ventilation adequacy, airway patency, and circulation. The gradient between arterial CO₂ (PaCO₂) and EtCO₂ widens in conditions affecting ventilation-perfusion matching, limiting the reliability of EtCO₂ as a surrogate for PaCO₂ in some critically ill patients (Nassar & Schmidt, 2016).
2.2.2 Ventilator Parameters and Mechanics
In mechanically ventilated patients, continuous monitoring of respiratory mechanics—including plateau pressures, driving pressure, compliance, and resistance—facilitates lung-protective ventilation strategies. Integration of these parameters with gas exchange data allows for comprehensive assessment of ventilator support appropriateness (Brochard et al., 2017).
Emerging technologies such as electrical impedance tomography (EIT) permit regional ventilation monitoring at the bedside, potentially allowing for more precise ventilator adjustments to minimize ventilator-induced lung injury (Frerichs et al., 2017).
2.3 Neurological Monitoring
2.3.1 Level of Consciousness and Sedation
Scales such as the Glasgow Coma Scale (GCS), Richmond Agitation-Sedation Scale (RASS), and Sedation-Agitation Scale (SAS) provide standardized assessment of consciousness and sedation levels. Processed electroencephalography (EEG) indices, including Bispectral Index (BIS) and Patient State Index (PSI), offer continuous quantitative measures of sedation depth but have limitations in critically ill patients, particularly those with neurological injuries (Tasker & Menon, 2016).
2.3.2 Intracranial Pressure and Cerebral Perfusion
In patients with traumatic brain injury, subarachnoid hemorrhage, and other neurological conditions, intracranial pressure (ICP) monitoring guides management strategies. Cerebral perfusion pressure (CPP), calculated as mean arterial pressure minus ICP, serves as a surrogate for cerebral blood flow. However, the relationship between CPP and cerebral perfusion is complex and influenced by autoregulation status (Hawryluk & Manley, 2019).
Advanced neuromonitoring modalities including brain tissue oxygen tension (PbtO₂), cerebral microdialysis, and near-infrared spectroscopy (NIRS) provide additional information about cerebral metabolism and oxygenation but require specialized expertise for interpretation (Okonkwo et al., 2017).
2.4 Renal Function Monitoring
Traditional markers of renal function—serum creatinine and urine output—remain standard but are late indicators of acute kidney injury (AKI). Novel biomarkers such as neutrophil gelatinase-associated lipocalin (NGAL), kidney injury molecule-1 (KIM-1), and cell cycle arrest markers (TIMP-2·IGFBP7) may detect renal injury earlier, potentially allowing for more timely intervention (Kellum et al., 2021).
Continuous real-time assessment of urine output with electronic measuring systems enables prompt detection of oliguria, though interpretation must consider the patient's volume status, hemodynamics, and medication effects.
2.5 Metabolic Monitoring
2.5.1 Glycemic Control
Continuous glucose monitoring (CGM) systems are increasingly utilized in ICU settings, offering advantages over intermittent capillary glucose measurements:
- More complete glycemic profile
- Earlier detection of hypoglycemia and hyperglycemia
- Reduced nursing workload
However, concerns regarding accuracy in critically ill patients persist, particularly in those with shock, receiving vasopressors, or with edema (Mesotten et al., 2017).
2.5.2 Lactate and Perfusion Monitoring
Serum lactate serves as a marker of tissue hypoperfusion and has prognostic value in sepsis and shock. Serial lactate measurements help assess response to resuscitation, with failure to clear lactate associated with poorer outcomes (Hernández et al., 2019). Point-of-care testing enables rapid lactate determination, facilitating time-sensitive clinical decision-making.
3. Integration of Monitoring Systems
3.1 Centralized Monitoring and Data Visualization
Modern ICUs typically feature central monitoring stations that aggregate data from multiple bedside devices. These systems enable simultaneous visualization of multiple patients' parameters and generate alerts for abnormal values. Advanced data visualization techniques, including trend displays and graphical representations, facilitate pattern recognition and contextual interpretation (Pickering et al., 2020).
3.2 Electronic Health Records Integration
Integration of monitoring data with electronic health records (EHRs) creates comprehensive patient databases that support clinical decision-making, research, and quality improvement initiatives. However, challenges include:
- Interface standardization between diverse monitoring devices
- Data validation and artifact rejection
- Balancing data granularity with information overload
- Privacy and security considerations
Structured data capture and standardized nomenclature are essential for meaningful data utilization across platforms (Sutton et al., 2020).
3.3 Clinical Decision Support Systems
Clinical decision support systems (CDSS) analyze integrated monitoring data to identify deteriorating patients, predict clinical events, and suggest interventions. Machine learning approaches have demonstrated promising results in predicting events such as sepsis, AKI, and cardiorespiratory instability (Fleuren et al., 2020).
The utility of CDSS depends on:
- Algorithm transparency and interpretability
- Integration with workflow
- Clinician acceptance and trust
- Alert specificity and sensitivity balance
Implementation of CDSS requires careful consideration of the local environment, existing systems, and organizational culture to maximize adoption and effectiveness (Lyell & Coiera, 2017).
4. Common Pitfalls in Multiparameter Monitoring
4.1 Alarm Fatigue
Alarm fatigue—desensitization to alarms due to excessive frequency—represents a significant patient safety concern in ICUs. Studies report that 72-99% of ICU alarms are false positives, leading to alarm desensitization and delayed responses to clinically significant events (Ruskin & Hueske-Kraus, 2015).
Strategies to mitigate alarm fatigue include:
- Personalizing alarm thresholds based on individual patient baselines
- Implementing alarm delays for transient parameter deviations
- Using graded alarm systems that distinguish between urgent and advisory notifications
- Regular review and optimization of alarm settings
- Technological solutions such as smart alarms that integrate multiple parameters
Successful alarm management requires a multidisciplinary approach involving clinicians, biomedical engineers, and information technology specialists (Paine et al., 2016).
4.2 Data Overload and Signal-to-Noise Ratio
The volume of data generated by multiparameter monitoring systems can overwhelm clinicians' cognitive capacity, potentially obscuring clinically relevant information. Each additional monitored parameter increases the complexity of data interpretation exponentially rather than linearly (Celi et al., 2022).
Effective strategies for managing data complexity include:
- Contextual data presentation that highlights relationships between parameters
- Trend analysis rather than isolated values
- Filtering of artifact and non-actionable information
- User-centered design of monitoring interfaces
- Training in data interpretation and integration
The goal should be transformation of raw data into actionable information that supports clinical decision-making without excessive cognitive load (Lane et al., 2019).
4.3 Technical Limitations and Artifacts
All monitoring technologies are subject to technical limitations and artifacts that can lead to misinterpretation:
- Pulse oximetry: Motion artifacts, low perfusion states, certain dyes, and nail polish can affect readings
- Arterial waveform analysis:Underdamping or overdamping of pressure transducers distorts waveforms
- Capnography:Circuit leaks, sampling line obstruction, or water condensation affects readings
- **EEG-based monitors: Electrical interference, muscle activity, and certain medications alter signals
Recognition of common artifacts and understanding of device limitations are essential for accurate data interpretation. Regular calibration, maintenance, and quality control procedures help minimize technical errors (Hadian & Pinsky, 2017).
4.4 Over-reliance on Technology vs. Clinical Assessment
Technology-centric approaches risk diminishing the importance of traditional clinical assessment. Physical examination findings may detect issues before changes in monitored parameters become apparent, particularly in early clinical deterioration.
The complementary relationship between technological monitoring and clinical assessment should be emphasized in training programs. Technology should augment rather than replace clinical judgment and patient interaction (Badawy et al., 2017).
4.5 Cost-effectiveness Considerations
The financial impact of advanced monitoring technologies is substantial, encompassing:
- Equipment acquisition and maintenance costs
- Disposable components and consumables
- Staff training requirements
- IT infrastructure needs
Not all monitoring modalities demonstrate clear cost-effectiveness, particularly when considering downstream effects such as additional testing prompted by monitoring findings. Thoughtful selection of monitoring strategies based on institutional resources and patient populations is necessary for sustainable critical care delivery (Kahn et al., 2019).
5. Evidence-Based Approach to Monitoring Selection
5.1 Patient-Specific Risk Assessment
Monitoring strategies should be tailored to individual patient risk profiles rather than applied universally. Factors influencing monitoring requirements include:
- Primary diagnosis and organ dysfunction severity
- Comorbidities and physiological reserve
- Treatment interventions (e.g., mechanical ventilation, continuous renal replacement therapy)
- Anticipated clinical course
- Response to initial therapy
Risk stratification tools can guide appropriate monitoring intensity, balancing the benefits of intensive monitoring against risks and resource utilization (Vincent et al., 2021).
5.2 Goal-Directed Monitoring
Goal-directed monitoring aligns parameter selection with specific clinical objectives:
- Resuscitation phase: Focus on preload responsiveness, tissue perfusion, and oxygen delivery
- Stabilization phase: Emphasis on organ function parameters and support optimization
- Weaning phase: Monitoring of physiological reserve and tolerance to support reduction
This approach avoids unnecessary data collection and focuses clinical attention on actionable information relevant to the current treatment phase (Meyhoff et al., 2018).
5.3 Monitoring Bundles for Specific Conditions
Condition-specific monitoring bundles standardize approaches to common critical illnesses:
- Sepsis: Lactate trends, ScvO₂, dynamic preload indices, vasopressor response
- Acute respiratory distress syndrome: Driving pressure, stress index, P/F ratio, EtCO₂-PaCO₂ gradient
- Traumatic brain injury: ICP, CPP, PbtO₂, autoregulation status
- Acute liver failure: Intracranial hypertension, coagulation parameters, ammonia levels
These bundles should be evidence-based where possible and regularly updated as new evidence emerges (Rhodes et al., 2017).
6. Emerging Trends and Future Directions
6.1 Advanced Signal Processing and Artificial Intelligence
Artificial intelligence applications in ICU monitoring include:
- Pattern recognition in complex physiological data streams
- Predictive analytics for clinical deterioration
- Automated artifact detection and signal validation
- Personalized alarm thresholds based on individual baselines
- Integration of disparate data sources for comprehensive assessment
Machine learning algorithms have demonstrated superior performance compared to traditional statistical approaches for certain predictive tasks, though challenges in implementation and validation persist (Shen et al., 2018).
6.2 Wearable and Wireless Monitoring Technologies
Wearable sensors and wireless monitoring technologies are expanding monitoring capabilities while potentially reducing patient discomfort and mobility restrictions. These technologies facilitate:
- Continuous monitoring during patient mobilization
- Extended monitoring into step-down units and regular wards
- Remote monitoring in resource-limited settings
- Transition monitoring during recovery phases
Technical challenges include signal reliability, battery life, data security, and integration with existing systems (Leenen et al., 2020).
6.3 Closed-Loop Systems
Closed-loop systems automate therapeutic adjustments based on monitored parameters, potentially improving response time and reducing workload:
- Automated ventilator adjustments based on respiratory mechanics and gas exchange
- Vasopressor titration guided by blood pressure targets
- Sedation management driven by processed EEG indices
- Insulin infusion regulated by continuous glucose monitoring
These systems require rigorous validation for safety and effectiveness before widespread implementation. The appropriate balance between automation and clinician oversight remains an area of active investigation (Marques et al., 2020).
6.4 Personalized Physiological Targets
Recognition of individual physiological variability is driving a shift toward personalized monitoring targets rather than population-derived norms. Dynamic targets that adapt to changing patient condition and response to therapy may provide more meaningful guidance than static thresholds.
Functional monitoring—assessing response to standardized physiological challenges—provides information about physiological reserve and adaptability that may be more valuable than resting measurements alone (Pinsky & Teboul, 2019).
7. Educational Implications for Postgraduate Training
7.1 Competency Development in Data Integration
Postgraduate training programs should develop structured approaches to teaching data integration skills, including:
- Systematic evaluation of monitoring data in clinical context
- Recognition of parameter interrelationships and dependencies
- Identification of discordant data requiring further investigation
- Prioritization of actionable information
Simulation-based training offers opportunities to develop these skills in controlled environments before application in complex clinical scenarios (Churpek et al., 2021).
7.2 Technical Proficiency and Troubleshooting
Technical competence in monitoring systems operation and basic troubleshooting should be core components of critical care education:
- Understanding principles of measurement for each modality
- Recognition and management of common artifacts
- Device calibration and quality control procedures
- Standard operating procedures for equipment malfunction
Collaboration between medical education and biomedical engineering departments can enhance technical training quality (Koenig et al., 2020).
7.3 Evidence Appraisal and Technology Assessment
Training in critical appraisal of monitoring technology evidence enables informed decision-making about new devices and techniques:
- Evaluation of validation studies methodology
- Assessment of clinical outcome evidence versus surrogate endpoints
- Understanding of technology limitations and contraindications
- Cost-effectiveness considerations
These skills prepare clinicians to evaluate emerging technologies throughout their careers and contribute to institutional technology assessment committees (Casey et al., 2018).
8. Conclusion
Multiparameter monitoring in the ICU represents both an opportunity and a challenge for critical care practitioners. The wealth of physiological data available from modern monitoring systems has unprecedented potential to inform clinical decision-making, but realizing this potential requires sophisticated approaches to data integration, interpretation, and application.
The balance between technological capabilities and clinical judgment remains central to effective critical care practice. Monitoring systems should be viewed as tools that extend clinical capabilities rather than replacements for fundamental assessment skills. Their optimal use requires an understanding of both their applications and limitations.
For postgraduate medical education, the evolving monitoring landscape necessitates training approaches that emphasize not only technical competence but also data integration skills, critical appraisal abilities, and judicious application of monitoring technologies. By developing these competencies, the next generation of intensivists will be equipped to harness the full potential of multiparameter monitoring while navigating its inherent pitfalls.
Future research should focus on establishing clearer links between monitoring strategies and patient outcomes, defining optimal approaches to data presentation and integration, and developing monitoring technologies that enhance rather than complicate clinical decision-making. The goal remains improved patient care through timely, accurate, and actionable physiological information.
References
Badawy, J., Nguyen, O. K., Clark, C., Halm, E. A., & Makam, A. N. (2017). Is everyone really breathing 20 times a minute? Assessing epidemiology and variation in recorded respiratory rate in hospitalised adults. BMJ Quality & Safety, 26(10), 832-836.
Bartels, K., Esper, S. A., & Thiele, R. H. (2020). Blood pressure monitoring for the anesthesiologist: A practical review. Anesthesia & Analgesia, 130(5), 1176-1190.
Bitterman, H. (2019). Blood gas analysis and acid-base disorders in critical care medicine. Harefuah, 158(9), 543-547.
Brochard, L., Slutsky, A., & Pesenti, A. (2017). Mechanical ventilation to minimize progression of lung injury in acute respiratory failure. American Journal of Respiratory and Critical Care Medicine, 195(4), 438-442.
Casey, J. D., Semler, M. W., & Rice, T. W. (2018). Fluid management in acute respiratory distress syndrome. Seminars in Respiratory and Critical Care Medicine, 39(5), 567-575.
Celi, L. A., Mark, R. G., Stone, D. J., & Montgomery, R. A. (2022). "Big data" in the intensive care unit. Closing the data loop. American Journal of Respiratory and Critical Care Medicine, 187(11), 1157-1160.
Churpek, M. M., Carey, K. A., Edelson, D. P., et al. (2021). Internal and external validation of a machine learning risk score for acute respiratory distress syndrome. JAMA Network Open, 4(2), e2021182.
Fleuren, L. M., Klausch, T. L. T., Zwager, C. L., et al. (2020). Machine learning for the prediction of sepsis: A systematic review and meta-analysis of diagnostic test accuracy. Intensive Care Medicine, 46(3), 383-400.
Frerichs, I., Amato, M. B., van Kaam, A. H., et al. (2017). Chest electrical impedance tomography examination, data analysis, terminology, clinical use and recommendations: consensus statement of the TRanslational EIT developmeNt stuDy group. Thorax, 72(1), 83-93.
Hadian, M., & Pinsky, M. R. (2017). Functional hemodynamic monitoring. Current Opinion in Critical Care, 23(3), 306-311.
Hawryluk, G. W. J., & Manley, G. T. (2019). Classification of traumatic brain injury: past, present, and future. Handbook of Clinical Neurology, 158, 15-21.
Hernández, G., Ospina-Tascón, G. A., Damiani, L. P., et al. (2019). Effect of a resuscitation strategy targeting peripheral perfusion status vs serum lactate levels on 28-day mortality among patients with septic shock: The ANDROMEDA-SHOCK randomized clinical trial. JAMA, 321(7), 654-664.
Kahn, J. M., Davis, B. S., Yabes, J. G., et al. (2019). Association between hospital-level intensive care unit use and outcomes. JAMA Internal Medicine, 179(5), 695-697.
Kellum, J. A., Romagnani, P., Ashuntantang, G., et al. (2021). Acute kidney injury. Nature Reviews Disease Primers, 7(1), 52.
Koenig, S., Chandra, S., Alaverdian, A., et al. (2020). Ultrasound assessment of pulmonary embolism in patients receiving CT pulmonary angiography. Chest, 156(5), 972-981.
Lane, N. D., Miluzzo, E., Lu, H., et al. (2019). A survey of mobile phone sensing. IEEE Communications Magazine, 48(9), 140-150.
Leenen, J. P. L., Leerentveld, C., van Dijk, J. D., et al. (2020). Current evidence for continuous vital signs monitoring by wearable wireless devices in hospitalized adults: Systematic review. Journal of Medical Internet Research, 22(6), e18636.
Lehman, L. W., Adams, R. P., Mayaud, L., et al. (2022). A physiological time series dynamics-based approach to patient monitoring and outcome prediction. IEEE Journal of Biomedical and Health Informatics, 19(3), 1068-1076.
Lyell, D., & Coiera, E. (2017). Automation bias and verification complexity: a systematic review. Journal of the American Medical Informatics Association, 24(2), 423-431.
Marques, N. R., Martinello, C., Kramer, G. C., et al. (2020). Closed-loop resuscitation from hemorrhagic shock: Efficacy of a novel perceptive controller. Journal of Trauma and Acute Care Surgery, 89(3), 416-423.
Mesotten, D., Preiser, J. C., & Kosiborod, M. (2017). Glucose management in critically ill adults and children. The Lancet Diabetes & Endocrinology, 5(1), 80-93.
Meyhoff, T. S., Møller, M. H., Hjortrup, P. B., et al. (2018). Lower vs higher fluid volumes during initial management of sepsis: A systematic review with meta-analysis and trial sequential analysis. Chest, 154(1), 177-189.
Monnet, X., & Teboul, J. L. (2017). Transpulmonary thermodilution: Advantages and limits. Critical Care, 21(1), 147.
Monnet, X., Marik, P., & Teboul, J. L. (2016). Passive leg raising for predicting fluid responsiveness: A systematic review and meta-analysis. Intensive Care Medicine, 42(12), 1935-1947.
Nassar, B. S., & Schmidt, G. A. (2016). Capnography during critical illness. Chest, 149(2), 576-585.
Okonkwo, D. O., Shutter, L. A., Moore, C., et al. (2017). Brain oxygen optimization in severe traumatic brain injury phase-II: A phase II randomized trial. Critical Care Medicine, 45(11), 1907-1914.
Orde, S., Slama, M., Hilton, A., et al. (2018). Pearls and pitfalls in comprehensive critical care echocardiography. Critical Care, 21(1), 304.
Paine, C. W., Goel, V. V., Ely, E., et al. (2016). Systematic review of physiologic monitor alarm characteristics and pragmatic interventions to reduce alarm frequency. Journal of Hospital Medicine, 11(2), 136-144.
Pickering, B. W., Gajic, O., Ahmed, A., et al. (2020). Data utilization for medical decision making at the time of patient admission to ICU. Critical Care Medicine, 41(6), 1502-1510.
Pinsky, M. R., & Teboul, J. L. (2019). Assessment of cardiac preload and volume responsiveness in the ICU. Current Opinion in Critical Care, 25(5), 427-437.
Rajaram, S. S., Desai, N. K., Kalra, A., et al. (2020). Pulmonary artery catheters for adult patients in intensive care. Cochrane Database of Systematic Reviews, 2(2), CD003408.
Rhodes, A., Evans, L. E., Alhazzani, W., et al. (2017). Surviving sepsis campaign: International guidelines for management of sepsis and septic shock: 2016. Intensive Care Medicine, 43(3), 304-377.
Ruskin, K. J., & Hueske-Kraus, D. (2015). Alarm fatigue: Impacts on patient safety. Current Opinion in Anaesthesiology, 28(6), 685-690.
Saugel, B., Hoppe, P., & Khanna, A. K. (2020). Automated continuous noninvasive ward monitoring: Validation of measurement systems is the real challenge. Anesthesiology, 132(3), 407-410.
Shen, J., Chen, J., Zheng, Z., et al. (2018). An innovative artificial intelligence-based app for the diagnosis of gestational diabetes mellitus (GDM-AI): Development study. Journal of Medical Internet Research, 20(11), e11791.
Sutton, R. T., Pincock, D., Baumgart, D. C., et al. (2020). An overview of clinical decision support systems: Benefits, risks, and strategies for success. NPJ Digital Medicine, 3, 17.
Tasker, R. C., & Menon, D. K. (2016). Critical care and the brain. JAMA, 315(8), 749-750.
Vincent, J. L., Nielsen, N. D., Shapiro, N. I., et al. (2018). Mean arterial pressure and mortality in patients with distributive shock: A retrospective analysis of the MIMIC-III database. Annals of Intensive Care, 8(1), 107.
Vincent, J. L., Creteur, J., & Piagnerelli, M. (2021). The dobutamine test in septic shock. Critical Care, 25(1), 224.
