Cognitive computing comes to Asante

This month, inpatient areas at all Asante hospitals will be implementing Epic’s Deterioration Index and Early Detection of Sepsis predictive models for adult patients. The two models are a part of Epic’s cognitive computing platform that monitors patient populations to calculate each patient’s risk for deterioration or sepsis.

The models evaluate patients and their care using criteria that have been found to affect the likelihood of an event, from medications and lab results to pre-existing conditions. The models dynamically respond to information entered into the patient’s chart through any Epic workflow, providing clinicians with the most up-to-date risk assessment based on the information available in Epic.

There are thousands of data points about hospitalized patients — far too much information for the human brain to process quickly. Cognitive computing models helps clinicians make sense of all that data entered and improve patient outcomes by alerting clinicians proactively about high-risk patients, so they can prioritize care and evaluate the patient’s condition earlier. Keep in mind that cognitive computing models are meant to monitor patient populations to identify at-risk patients; they aren’t diagnostic tools or replacements for clinical judgment.

Deterioration Index

Early detection and prevention of patient deterioration is critical to promoting positive patient outcomes. More than 80% of critically ill patients admitted to critical care units, or those who have a cardiac arrest during their hospitalization, show clear and detectable signs of deterioration in the 24 hours preceding these events. The Deterioration Index was devised to assist in the early detection of deteriorating patients. Identifying the deteriorating patient earlier allows for more immediate intervention and reductions in morbidity, mortality, cost, resource utilization, transfers to the ICU and patient length of stay.

Epic’s Deterioration Index is an early warning system that allows care providers to review deterioration scores based on up-to-date patient information from 17 variables, such as vital signs, lab values, age, necessity for supplemental oxygen and cardiac rhythm. An aggregated score is automatically calculated in Epic. When a patient is identified as being a high-risk for deterioration, Epic will display a message to nurses on the Kardex report.

The Deterioration Index improves on traditional early warning systems, such as MEWS, (implemented at Asante in 2016), by including additional data points and using an algorithm that is more accurate than the simple arithmetic one used by other early warning systems. According to Epic, use of the Deterioration Index can help identify deteriorating patients 50% more than early warning systems such as MEWS.

Multiple nursing committees and providers at Asante have reviewed the Deterioration Index and the subsequent policy, Deterioration Index (DI) 400-PCS-NURS-0945. When identified threshold scores are reached, it is recommended the nurse use the Deterioration Index pathway. This pathway outlines recommended nursing actions, ensuring appropriate interventions for patients with the goal of improving patient outcomes by using the Deterioration Index tool. The Deterioration Index predictive model will replace MEWS at Asante.

Early Detection of Sepsis

Sepsis is the body’s over-response to an infection. It is a condition that is often associated with pathogenic microorganisms in the bloodstream and the body’s response to their presence. It progresses quickly and has a high mortality rate, which makes early diagnosis and prompt intervention crucial to saving lives and reducing morbidity.

Researchers estimate that 30 million people worldwide develop sepsis every year, and around 6 million of those affected die from the condition. Of those who develop sepsis, 25% of pediatric patients and 15% of adult patients die in the hospital. Detecting sepsis as early as possible improves patient outcomes and survival. With each hour a patient does not receive treatment, their risk of dying increases by 4%.

Epic’s Early Detection of Sepsis predictive model uses 80 variables from demographics; diagnoses; SIRS criteria; recent lab results; medication orders; and lines, drains and airway information to assist clinicians in detecting patients who are septic or may soon become septic. This enables them to identify and intervene before the patient’s condition has a chance to worsen. The model helps clinicians identify more septic patients and fewer non-septic ones than SIRS criteria alone, freeing up clinicians’ time to care for the patients who need it most.

For patients identified as having a high risk of developing sepsis (score of 7 or greater), the nurse can use the included information in the early warning system report as well as their clinical judgement to see if the patient may need further assessment and provider follow-up.

Jamie Wagner and Desiree Torassa contributed to this report.