Critical, unanticipated and adverse events happen every day in healthcare, resulting in poor outcomes, increased length of stay and unreimbursed cost of care.
It’s imperative to detect patterns or signatures of illness early on to mitigate these events and provide quality care, but there’s a challenge in doing so. Certain indicators of clinical deterioration, such as subtle changes in vital signs like blood pressure and heart rate, aren’t typically spotted by clinicians – until it’s too late.
Enter predictive analytics monitoring.
Predictive analytics monitoring – powered by software that uses artificial intelligence (AI) and data-driven predictive models to estimate a patient’s level of risk in real time – offers clinicians a second set of eyes for early symptom detection, giving them vital information to proactively treat and save lives.
A simple, intuitive visualization of risk score links data from a patient’s electronic health record (EHR) with continuous waveform data from the bedside monitor. With this, clinicians are informed early about possible deterioration, so they can begin a more focused evaluation.
Predictive analytics in action.
In this visualization from Premier’s AI-powered Prediction Assistant CoMET Inside technology, the red, orange and yellow “comets” on the display represent individual patients in a particular hospital unit who are being continuously monitored for the risk of sepsis in the next 8- 12 hours.
Fictional data for demonstration purposes.
The technology calculates a CoMET score based on continuous monitoring analytics and computer algorithms for each of the patients. Larger, darker comets mean patients are at higher risk for deterioration.
In the case of the patient in bed 94, the head of the comet indicates the patient’s high risk of sepsis diagnosis and respiratory instability in the next eight hours. The tail of the comet is the patient’s trajectory over the previous three hours, indicating that deterioration escalated during that time frame. This early warning gives clinicians the opportunity to begin assessments well ahead of a deterioration event.
Patient’s represented with a lighter yellow color and no tail on the comet are stable, below average risk for developing sepsis and may be ready for discharge.
Here are five real-world examples of how predictive analytics monitoring is empowering clinicians to act and prevent poor patient outcomes.
1. Warn of clinical deterioration in pre-symptomatic states.
The article, Nursing and Precision Predictive Analytics Monitoring in the Acute and Intensive Care Setting: An Emerging Role for Responding to COVID-19 and Beyond, outlines key concepts for the intersection of nursing and precision predictive analytics monitoring during the pandemic.
Major Finding: The use of AI-based technologies at the point of care has enabled nurses to detect clinical deterioration early, when COVID-19 patients are not yet showing clinical signs of impending deterioration. As a result, they’re able to intervene sooner and potentially prevent adverse outcomes.
2. Reduce high risk of emergency intubation.
Coronavirus Disease 2019 Calls for Predictive Analytics Monitoring—A New Kind of Illness Scoring System explores how continuous predictive analytics monitoring can provide an early warning of rising risk of endotracheal intubation due to sudden and severe respiratory failure brought on by COVID-19.
Major Finding: A variety of illness severity scoring systems were described in the study and none compared with CoMET’s intuitive and actionable displays for monitoring COVID-19 patients’ trajectories. The inclusion of continuous monitoring data provides more accurate risk monitoring of respiratory distress, allowing clinicians to prepare for intubation and reduce risk of exposure.
3. Reduce the rate of septic shock.
The individual EHRs of more than 4,000 patients admitted to a surgical and a medical ICU in the six months before and after deployment of CoMET technology are reviewed in Impact of Predictive Analytics Based on Continuous Cardiorespiratory Monitoring in a Surgical and Trauma Intensive Care Unit.
Major Finding: The rate of septic shock fell by 50 percent in the surgical ICU where the CoMET display was shown compared to only a 10 percent decrease in the medical ICU where it was not shown.
4. Identify and manage hemorrhage in surgery patients.
Episodes of hemorrhage occurring in a surgical ICU were studied in Hemorrhage Prediction Models in Surgical Intensive Care: Bedside Monitoring Data Adds Information to Lab Values.
Major Finding: Of 3,766 patients, 5.9 percent experienced a hemorrhage. Early warning system models that included both continuous predictive analytics monitoring and laboratory tests had the best performance (appropriate use criteria (AUC) = 92.2 percent), pointing to a combined strategy of continuous monitoring and intermittent lab tests as a the best clinical approach to the early detection of hemorrhage in the surgical ICU.
5. Detect hypoglycemia early.
Patients with low blood glucose requiring immediate therapy have poor outcomes, and earlier detection and therapy might lead to improvements. Pathophysiologic Signature of Impending ICU Hypoglycemia in Bedside Monitoring and Electronic Health Record Data reports on 11,847 intensive care unit patient admissions, and 721 (6.1 percent) had hypoglycemia events.
Major Finding: A signature of early hypoglycemia was identified, paving the way to a new system of monitoring patients, especially diabetics, for this in-hospital complication. Detection is possible even without measuring blood glucose levels.
From reactive to proactive clinical care.
AI is ushering in a new era of healthcare. Reactive care models focused on obvious clinical signs of deterioration to arise before jumping into action are being replaced by proactive care models built around a full picture of a patient’s condition using all the available data.
Predictive analytics monitoring is paving the way for this type of care – continuously looking for patterns in patient data and presenting it to clinicians right in their workflow, enabling them to see what’s up ahead and address risk before it’s too late.
Proactive care enables better quality and delivery at lower costs – a win for patients, providers and payers alike.
- Register for our upcoming Breakthroughs 21 Conference and attend “Using Predictive Analytics to Detect Signatures of Hypoglycemia,” presented by Dr. William Horton, University of Virginia (Premier members only).