Early detection of clinical deterioration in patients promises to reduce mortality and improve outcomes. However, this remains a challenge in both inpatient and outpatient settings.
Stanford Health Care addressed this challenge by integrating validated artificial intelligence and machine learning models into clinical decision support systems. They have also integrated AI into clinical workflows and improved the patient experience, including reducing wait times, improving quality of care and facilitating critical conversations.
Dr. Shreya Shah is a practicing academic internist, certified clinical informatics practitioner, and expert in integrating artificial intelligence into healthcare at Stanford Health Care.
She will speak about the health system’s AI efforts at The HIMSS AI in Healthcare 2023 Forum, scheduled for Dec. 14-15 in San Diego, will feature a case study titled “How Stanford Health Transformed Patient Care by Combining Compassion with AI-Driven Innovations.”
We spoke with Shah to get an overview of his session and a deeper understanding of how Stanford Health Care uses AI and ML.
Q. Why does detecting clinical deterioration in patients remain a challenge?
A. Hospitalized patients present with increasingly complex and serious illness, while lower-acuity care is shifted to the home, outpatient, or subacute care management level. Within an academic medical center, this situation is even more profound among patients at high risk of clinical deterioration.
The first signs can be subtle and vary greatly from patient to patient. Identifying which patients need the most attention is a needle-in-a-haystack activity. Additionally, these patients are cared for by care teams composed of several people and require the evaluation of large amounts of data that evolve over time.
Teams may experience communication gaps, information overload, and cognitive biases leading to unanticipated clinical deterioration with major consequences such as emergency resuscitation efforts and unplanned transfers to intensive care. There may also be varying degrees of alignment among team members on risk perception.
Standardized workflows for care coordination, which empower all members of the healthcare team in decisions regarding patient care, could help overcome these challenges.
Q. How did you decide that AI and ML was the way to address this challenge?
A. We needed to identify patients at increased risk and align the care team around a collaborative, standardized clinical response. We determined that an ML model can identify patients with a high probability of a future clinical deterioration event without additional tasks for our practicing clinicians.
Predictions should be made early enough to allow sufficient time for the healthcare team to respond. Accuracy is always a concern, and clinicians often think the AI system won’t tell them something they don’t already know.
In our implementations, the focus was not on the accuracy of the model predictions. Instead, for any given patient flagged by the model, physician and nonphysician care team members had to implement a structured collaborative workflow to assess risk and response. So a probabilistic model creates a team-based trigger.
Our implementation efforts focused on these priority areas: 1) Designing a system that would integrate the ML model into a complex healthcare system, 2) Creating effective teams and processes to enable collaborative workflows necessary for successful implementation, and 3) Deployment of these AIs. -integrated systems in a way that is both sustainable and scalable for the healthcare enterprise.
The focus was on creating a holistic system that not only integrates cutting-edge technology, but also aligns with clinical, operational and strategic needs.
Q. What is an example of how integrating validated AI and machine learning models into clinical decision support systems has helped Stanford address the challenge of clinical deterioration?
A. Our clinical deterioration model was validated on our data to ensure model performance; Then, the signals were seamlessly integrated into our EHR, including contributing factors and supplemented with a mobile alert to the healthcare team.
The ML model is capable of updating inpatient forecasts every 15 minutes and has been used to act as an objective risk assessor and has helped facilitate alignment and coordination of patient care as a model . AI-integrated system.
The model underwent site-specific validation to ensure its effectiveness in predicting clinical deterioration events such as unplanned ICU transfers within a 6-18 hour window. This workflow led to a significant increase in standardized multidisciplinary patient assessments and a 20% reduction in clinical deterioration events.
Results of the qualitative evaluation identified that the model was useful for aligning mental models and driving necessary workflows for patients flagged by the model with consensus among multidisciplinary team members. Using a reliable, continually updated risk signal, we aligned physicians with the rest of the care team to establish a consistent workflow.
Follow Bill’s HIT coverage on LinkedIn: Bill Siwicki
Email him: bsiwicki@himss.org
Healthcare IT News is a HIMSS Media publication.