key: cord-1020957-kzh0qk3i authors: Keim-Malpass, Jessica; Moorman, Liza P. title: Nursing and precision predictive analytics monitoring in the acute and intensive care setting: An emerging role for responding to COVID-19 and beyond date: 2021-01-05 journal: Int J Nurs Stud Adv DOI: 10.1016/j.ijnsa.2021.100019 sha: 28628f8a48ce1e3a32a79250aef9e0858d98ea9a doc_id: 1020957 cord_uid: kzh0qk3i As the global response to COVID-19 continues, nurses will be tasked with appropriately triaging patients, responding to events of clinical deterioration, and developing family-centered plans of care within a healthcare system exceeding capacity. Predictive analytics monitoring, an artificial intelligence (AI)-based tool that translates streaming clinical data into a real-time visual estimation of patient risks, allows for evolving acuity assessments and detection of clinical deterioration while the patient is in pre-symptomatic states. While nurses are on the frontline for the COVID-19 pandemic, the use of AI-based predictive analytics monitoring may help cognitively complex clinical decision-making tasks and pave a pathway for early detection of patients at risk for decompensation. We must develop strategies and techniques to study the impact of AI-based technologies on patient care outcomes and the clinical workflow. This paper outlines key concepts for the intersection of nursing and precision predictive analytics monitoring.  The advent of predictive analytics allows for many applications to nursing practice and science including the use of precision predictive analytics monitoring to allow for early detection of clinical deterioration when patients are not yet showing clinical signs of impending deterioration.  Nurses are critical to move from artificial intelligence (AI) creation to implementation in a complex learning health system. What does this paper add?  This paper introduces the concept of precision predictive analytics monitoring, or AIbased tool that translates streaming clinical data into a real-time estimation of patient risks, allows for evolving acuity assessments and detection of clinical deterioration while the patient is in pre-symptomatic states.  Here, we explore the intersections of nursing practice, nursing science, and precision predictive analytics monitoring with unique considerations during the COVID-19 pandemic. The World Health Organization (WHO) has announced the year 2020 as Year of the Nurse and Midwife as recognition of the bicentenary celebration of Florence Nightingale"s birth. 1-3 The WHO designation recognizes the unique contribution of nurses in health care delivery and overall population health, culminating in a report highlighting the state of the global nursing workforce. 2 There has been an ongoing recognition of the fragility of the global nursing workforce and the simultaneous rising acuity of many patients being cared for in the hospital setting. 4, 5 Concurrently, 2020 is also the year that the novel Severe Acute Respiratory Nursing roles and responsibilities will continue to evolve during the COVID-19 pandemic. Even when on the acute care hospital floor and intensive care unit under routine circumstances, many patients will have complications or events of rapid clinical deterioration that prolong their stay and increase their overall morbidity and mortality. [9] [10] [11] One way to optimize appropriate acuity assessment, accurately determine patients who are at risk for clinical deterioration and determine appropriate clinical actions is through precision predictive analytics monitoring; by this, we mean the use of quantitative tools and statistical models to detect clinical deterioration before overtly catastrophic presentations. 12 Patients infected with COVID-19 have demonstrated rapid and unpredictable events of clinical deterioration leading to sepsis, intensive care unit utilization, and death, and use of precision predictive analytic systems can allow for early detection for appropriate clinical action or upgrade in care. 13 As such, these technologies have uses in a variety of settings including the acute care wards, the emergency department, and the intensive care unit (ICU). Advances in continuous bedside monitoring technology capabilities make a wealth of data available to healthcare providers for use in patient evaluation. 12, 14, 15 These data form the foundation for computational algorithms that integrate real-time bedside physiologic monitoring data to provide early warning of potentially catastrophic clinical events, including sepsis, respiratory distress, cardiac instability, and the need to upgrade to ICU level care. 11, [16] [17] [18] [19] [20] [21] Precision predictive analytics monitoring involves an advanced mathematical analysis of data from a variety of inputs (including cardiorespiratory monitoring using streaming ECG data to perform additional waveform calculations such as measurement of RR intervals, measures of entropy and heart rate variability; laboratory data from the electronic medical record; and nurse entered vital signs and assessments) to derive an estimate of fold-increase in the risk of clinical deterioration that clinicians can observe in real-time in a streaming environment. 12, 21 In the neonatal setting, Moorman and colleagues detected abnormal heart rate characteristics in the hours preceding a clinical diagnosis of sepsis and developed a computational model that yielded the fold-increase in the risk of sepsis in the next 24 hours. 17, 19 When 3000 very low birth weight neonates were randomized to display, or not, of the risk of imminent sepsis, there was a 20 percent decrease in mortality among those whose display was shown. 22, 23 The mathematical approach of precision predictive analytics monitoring has since been extended to acute and critically ill adults and critically ill children in the pediatric intensive care unit by first determining physiologic signatures of illness. 11 Novel precision predictive analytics present a way for nurses, physicians, respiratory therapists, and clinicians at all points of care to be able to visualize changes in patients" risks over time particularly with ever-changing clinical illness trajectories. Another unique aspect of the precision approach is that it allows for continuous predictive analytics (derived from continuous ECG waveform computations collected every 2 seconds with a risk score updated every 15 minutes) to estimate a patient"s changing risk, which in turn informs clinical reasoning and decision-making. 15 For example, a patient can have a high baseline acuity or risk score from underlying conditions, but access to novel precision predictive analytics monitoring allows clinicians to employ tailored alert or assessment strategies governed by a continual restratification of risk. While other forms of early warning systems also allow you to follow trends, visual analytics such as CoMET present the trend over the past three hours within the visual display of the presented (specifically the CoMET "tail), the size of the CoMET score displayed corresponds with greater risk, and the clinician can access trends over the past 72 hours within on a large LCD display and touchpad. Additionally, clinicians can assess response to therapies or response to escalation in care in real-time. Precision predictive analytics monitoring offers a visual display of the patient"s risk trajectory and can allow for personalized trend assessment instead of using arbitrary cut-points or alert thresholds for action. 25 lead the nurse to assume the patient is at a higher risk for decompensation and early warning is of less clinical value.) Further, most of these canonical early warning scores were trained only on one clinical deterioration outcome, a one-size-fits-all modeling approach that clinicians might question. 35, 36, [38] [39] [40] Finally, they generally use cut-offs to award points for illness severity, and lose fine degrees of changes within ranges of normal and abnormal. 25, 41 Clinicians at the point of care during a global pandemic must engage with the risk estimates in the context of a learning health system (LHS), or an integrated health system in which informatics and care culture align for optimal patient outcomes ( Figure 2 ). [42] [43] [44] As the Institute of Medicine report entitled "Digital Infrastructure for the Learning Health System" suggests, digital health infrastructures (including predictive analytics, risk predictions, and use of AI in healthcare) will be central to the healthcare of the future and necessary for continued improvement in patient outcomes. 42 It is estimated that upwards of 90 percent of clinical decisions will soon be supported by accurate, timely, and up-to-date digital clinical information in the near future. 42 In order to keep pace, we must simultaneously study the clinical processes necessary in responding to emerging precision predictive analytics monitoring. We must also develop innovative approaches for health systems and individual clinicians to understand the benefits and consequences of implementing and acting on risk predictions. Finally, we must be aware of the inherent pressures of an already extended clinical workforce during the global COVID-19 pandemic and work alongside nurse stakeholders to utilize models that allow them to use the correct risk scores at the right time to make the right clinical actions in a way that is synergistic to their current practices. 45 Beyond clinical use features, a LHS allows precision predictive analytics monitoring to re-train underlying statistical models and be updated while more events of COVID-19 are treated within a system. Optimized use of precision continuous predictive analytics monitoring relies on theoretical underpinnings of situational analysis, as first conceptualized by Endsley 46 and applied to the field of engineering, human factors, and ergonomics research. Endsley explored the relationship between situational awareness and environmental and individual factors that impact dynamic decision making and performance of subsequent action(s). 46 Among these factors, limited attention and working memory have been identified as critical components preventing "operators" from acquiring and interpreting information from the environment in realtime. 46 These elements have been demonstrated in a variety of settings including aviation/air traffic control, military command, and among clinicians in complex healthcare environments. 46 Through our preliminary qualitative research among over 40 adult surgical and neonatal clinicians, we have found that elements of situational awareness are a critical antecedent necessary prior to initiating clinical action in response to a change in risk score presented by precision continuous predictive analytics monitoring. 45, 47 A balanced approach between the development of meaningful analytics and engaged situational awareness among nurses on the acute care floor or ICU is critical, so attention and working memory can be properly devoted to dynamic decision making and translation to relevant clinical actions. 48 In the fight against COVID-19 investment in clinical decision support infrastructure and systems can allow nurses to better optimize care delivery and assessments of risk in ways that allow them to be proactive in care delivery. Figure 3 depicts how "Level 3" of situational awareness, or the projection and action based on a future status (as opposed to perception or comprehension current situation) is the most challenging level to achieve particularly in the context of dynamic acute and intensive care settings with numerous other clinical elements competing for attention. Precision predictive analytics monitoring is attempting to intervene on Level 3, and in order to be successful must add additional value to clinical decision-making while not becoming another form of technology that contributes to unnecessary alert fatigue. 10, 14 Paradoxically, the alarm and alert systems that are created to enhance patient safety have become an emerging patient safety concern. In the ICU, the bedside cardiorespiratory monitor alone generates, on average, 187 audible alarms per day averaging 1 per every 7.7 minutes with upwards of 90 percent of them being non-actionable. [49] [50] [51] [52] [53] Clinicians become accustomed to ignoring non-actionable or false alarms and in doing so may overlook alarms in a true emergent situation, otherwise known as "alarm fatigue". 54, 55 Precision predictive analytics monitoring that rely on alert strategies must be designed to assess and account for alarm fatigue within the design and implementation in order to offset unintended consequences. Further, there is a need to examine whether some existing alarm systems can be removed when predictive analytics are added into the acute care floor ICU environment. Nurses already are and will continue to be first responders in the context of the global COVID-19 pandemic and pandemics of the future. In the acute and intensive care setting, clinical bedside nurses are with the patient continuously and already tasked with monitoring a diverse array of vital signs and data inputs. [56] [57] [58] Nurses use this information to initiate patientcentered interventions and communicate subtle changes in patient status to their physician colleagues with the ultimate goal of (1) reducing catastrophic events and complications for their patients and (2) determining overall clinical stability to promote nursing care that is beneficial to overall clinical outcomes. Precision predictive analytics monitoring demonstrates the potential to synthesize and compute diverse data inputs through visual risk estimates that nurses can interface with and transform care from reactive to proactive. In doing so, nurses can aid in the dynamic decision-making processes to pinpoint physiological acuity assessment and help to (1) refine the correct intensity of care delivery (home, acute care, intensive care); (2) determine early warning for patients at risk of clinical deterioration and refine clinical interventions; (3) define when an upgrade in intervention is needed, for instance defining the need for extracorporeal membrane oxygenation among COVID-19 patients or when a patient should be transferred to a higher level of care; (4) add additional information to inform a nurses" worry or "gut feeling" that a patient is at risk for clinical deterioration. 59 The primary goal of precision predictive analytics monitoring is to draw attention to a patient who may have a change in risk, but is not exhibiting overt clinical signs that are indicative of imminent deterioration. These technologies do not operate in isolation and are only successful when paired with clinicians" careful assessment. Perhaps a change in heart rate variability, undetected clinically but calculated computationally, leads an increase in risk score, garners the attention of the nurse to re-assess the patient, and communicate this finding with the physician team. Perhaps they were undecided about sending a blood culture earlier in the day, but this new finding pushes the decision-making towards obtaining a needed blood culture earlier than would have been obtained otherwise. In a different patient there may be a few subtle changes in vital sign characteristics and laboratory results that when assessed independently are not outside of normal limits, but when that risk is combined cumulatively represents a physiological change for that patient. The impact of precision predictive analytics monitoring technologies has the potential to change the clinical paradigm from reactive to proactive. The utility depends on situational awareness: perceiving and comprehending data, and projecting that comprehension into the future in order to alter the patient"s trajectory. The implementation centers on assigning specific responsibilities and tasks: Who watches the data? (i.e., a routine bedside nurse or a clinician designated to monitor the technology from a distance, such as an eICU?) 62 Nursing in 2020: a call for papers Introduction to the WHO Year of the Nurse and Midwife: The impact of critical care nurses and meet the new editors 2020 year of the nurse and midwife: Meeting new challenges Relevance and Effectiveness of the WHO Global Code Practice on the International Recruitment of Health Personnel--Ethical and Systems Perspectives Nurse-patient assignment models considering patient acuity metrics and nurses' perceived workload Covid-19: WHO declares pandemic because of "alarming levels" of spread, severity, and inaction Clinical course and risk factors for mortality of adult inpatients with COVID-19 in Wuhan, China: a retrospective cohort study Early in-hospital clinical deterioration is not predicted by severity of illness, functional status, or comorbidity Towards development of alert thresholds for clinical deterioration using continuous predictive analytics monitoring Signatures of subacute potentially catastrophic illness in the ICU: model development and validation Cardiorespiratory dynamics measured from continuous ECG monitoring improves detection of deterioration in acute care patients: A retrospective cohort study Clinical predictors of mortality due to COVID-19 based on an analysis of data of 150 patients from Wuhan, China Advancing continuous predictive analytics monitoring: Moving from implementation to clincial action in a learning health system Impact of predictive analytics based on continuous cardiorespiratory monitoring in a surgical and trauma intensive care unit Stochastic modeling of central apnea events in preterm infants Local dynamics of heart rate: detection and prognostic implications Septicemia mortality reduction in neonates in a heart rate characteristics monitoring trial Sample entropy analysis of neonatal heart rate variability Accurate estimation of entropy in very short physiological time series: the problem of atrial fibrillation detection in implanted ventricular devices Complex signals bioinformatics: evaluation of heart rate characteristics monitoring as a novel risk marker for neonatal sepsis HeRO monitoring to reduce mortality in NICU patients Mortality reduction by heart rate characteristic monitoring in very low birth weight neonates: a randomized trial Predictive analytics in the pediatric intensive care unit for early identification of sepsis: capturing the context of age Early Detection of In-Patient Deterioration Dynamic data in the ED predict requirement for ICU transfer following acute care admission External validation in an intermediate unit of a respiratory decompensation model trained in an intensive care unit Predicting the need for urgent intubation in a surgical/trauma intensive care unit Heart rate dynamics preceding hemorrhage in the intensive care unit Identifying the low risk patient in surgical intensive and intermediate care units using continuous monitoring The Royal College of Physicians. The National Early Warning Score (NEWS) thresholds and triggers National early warning score (NEWS): Standardising the assessment of acute-illness severity in the NHS. London: The Royal College of Physicians Early detection, prevention, and mitigation of critical illness outside intensive care settings Early detection of impending physiologic deterioration among patients who are not in intensive care: development of predictive models using data from an automated electronic medical record Development and validation of a continuous measure of patient condition using the Electronic Medical Record Association between intensive care unit transfer delay and hospital mortality: A multicenter investigation Prognostic Accuracy of the SOFA Score, SIRS Criteria, and qSOFA Score for In-Hospital Mortality Among Adults With Suspected Infection Admitted to the Intensive Care Unit Development and validation of a continuously ageadjusted measure of patient condition for hospitalized children using the electronic medical record National Early Warning Score (NEWS) -Standardising the Assessment of Acute-Illness Severity in the NHS Comparison of the Between the Flags calling criteria to the MEWS, NEWS and the electronic Cardiac Arrest Risk Triage (eCART) score for the identification of deteriorating ward patients Do optimal prognostic thresholds in continuous physiological variables really exist? Analysis of origin of apparent thresholds, with systematic review for peak oxygen consumption, ejection fraction and BNP Digital Infrastructure for the Learning Health System: The Foundation for Continuous Improvement in Health and Health Care The science of Learning Health Systems: Foundations for a new journal Achieving a nationwide learning health system Advancing Continuous Predictive Analytics Monitoring: Moving from Implementation to Clinical Action in a Learning Health System Toward a theory of situation awareness in dynamic systems Diffusing an innovation: clinician perceptions of continuous predictive analytics monitoring in intensive care Nurses' use of situation awareness in decision-making: an integrative review Insights into the problem of alarm fatigue with physiologic monitor devices: a comprehensive observational study of consecutive intensive care unit patients Reducing false alarm rates for critical arrhythmias using the arterial blood pressure waveform The heuristics of nurse responsiveness to critical patient monitor and ventilator alarms in a private room neonatal intensive care unit Nurse responses to physiologic monitor alarms on a general pediatric unit Alarms, oxygen saturations, and SpO2 averaging time in the NICU Technological distractions (part 2): A summary of approaches to manage clinical alarms with intent to reduce alarm fatigue Addressing vital sign alarm fatigue using personalized alarm thresholds Technologic distractions (part 1): summary of approaches to manage alert quantity with intent to reduce alert fatigue and suggestions for alert fatigue metrics Automated monitoring compared to standard care for the early detection of sepsis in critically ill patients Monitoring cardiorespiratory instability: Current approaches and implications for nursing practice Nurses' worry or concern and early recognition of deteriorating patients on general wards in acute care hospitals: a systematic review The fifth vital sign? Nurse worry predicts inpatient deterioration within 24 hours Heart rate characteristics and clinical signs in neonatal sepsis Innovative interdisciplinary strategies to address the intensivist shortage Acknowledgement: The authors would like to acknowledge and thank (1) J. Randall Moorman, MD, for his brilliant review of this manuscript -"great stuff", and (2) Doug E. Lake, PhD, for providing the inspiration to one day use precision predictive analytics to be a top 5 finisher in the sepsis challenge.Contribution JKM conceptualized the manuscript. JKM and LPM wrote and edited the manuscript. LPM declares a conflict of interest as she is Chief Implementation Officer and shareholder in AMP3D, Charlottesville, VA, USA.Funding JKM was funded through the Gordon and Betty Moore Research Fund as a Betty Irene Nurse Fellow that supported this paper.