Artificial Intelligence-Based Health Information Technology Tools to Optimize Critical Care Pharmacist Resources Through Adverse Drug Event Prediction

Developing artificial intelligence-based health information technology tools to guide critical care pharmacists in preventing medication errors has the potential to optimize resources and improve patient safety by reducing adverse drug events.

Project Details -
Ongoing

Medication regimens for intensive care unit (ICU) patients are complex, leading to these patients having a three times greater risk of adverse drug event (ADEs), with over 1.5 million individuals in ICUs each year experiencing potentially life-threatening ADEs. Although most ADEs are preventable, a patient experiencing a single ADE doubles their chance of mortality and is a significant financial burden on the healthcare system. Working within ICUs, critical care pharmacists (CCPs) make recommendations to the ICU team around medication interventions, such as laboratory monitoring, dose adjustment, identification of drug-drug interactions, order entry error correction, and alternative medications with lower risk. As such, CCPs can prevent ADEs, improve patient-centered outcomes, and reduce healthcare costs. However, CCPs provide care for a high number of ICU patients at one time, resulting in high cognitive loads and issues with effectively allocating their time. Strategies to optimize CCP care hold the potential to make patient care safer, more efficient, and with improved resource allocation. 

To address this need, artificial intelligence (AI) and machine learning (ML) will be used to create algorithms for predicting which patients are at risk by defining intervenable events based on patient features, opportunities for intervention, and those associated with poor outcomes. The researchers will leverage a previously-developed tool called the medication regimen complexity-intensive care unit (MRC-ICU) Scoring Tool, which will be integrated into visualization dashboards for CCPs, called ICView. 

The specific aims of the research are as follows:

  • Create robust prediction models of intervenable events to guide CCP medication interventions.
  • Explore causal relationships among intervenable events, CCP interventions, and outcomes.
  • Design an electronic health record-integrated platform (ICView) to visualize predictions to guide CCP care.

The researchers will apply AI and ML methodology using data from multiple centers to create these integrated prediction tools. The researchers will focus on understanding the best metric for predicting ICU intervenable ADEs and CCP workload, the causal factors of intervenable events that can be prevented by CCPs, and how CCPs are able to efficiently use AI-based predictions at the bedside. Applying user-centered design methods, researchers will create the visualization dashboard called ICView that houses MRC-ICU-based AI-informed prediction models for CCP interventions. It is anticipated that by identifying intervenable events, reducing the cognitive workload of CCPs, and better allocating the time of CCPs, ADEs will be reduced, and patient outcomes will improve.

Use of Artificial Intelligence and Machine Learning to Improve Care by Critical Care Pharmacists

Theme:

Supporting Health Systems in Advancing Care Delivery

Subtheme:

Using Digital Healthcare Tools to Improve Patient Safety

Using machine learning- and artificial intelligence-developed tools in the ICU has the potential to optimize critical care pharmacist resources and improve patient safety by reducing adverse drug events.

Critical care pharmacists have the expertise to handle patients that have complex medical needs

In intensive care units (ICUs), critical care pharmacists (CCPs) are integral members of healthcare teams. CCPs analyze and manage the highly complex medication regimens of ICU patients and work to identify and provide guidance for medication-related problems. Research shows that having a pharmacist on rounds in the ICU has significant benefits, including reduction of medication errors and adverse drug events (ADEs), improved patient outcomes, reduced costs, and most importantly, reduced risk of death by 20 percent.

But access to these pharmacists in the United States is lacking. Not all ICUs have CCP care and even when they do, the CCP is often caring for many patients—up to 50 or more at a time. This leads to high cognitive load and difficulty effectively managing time to provide for the patients most in need of CCP care.

As a CCP herself, Dr. Andrea Sikora sees firsthand how ICU patients benefit from having a CCP on the care team. With two recently awarded AHRQ-funded grants, she is studying the challenges that can be created by these gaps in CCP care and availability and how technology can better support their work to improve patient-centered outcomes.

Implementing a tool can quantify the complexity of a patient’s medication regimen and predict potential ADEs

In the first study, Dr. Sikora and her team at the University of Georgia are using artificial intelligence and machine learning to create algorithms for predicting which patients are at risk for ADEs based on patient features, opportunities for intervention, and those associated with poor outcomes. The tool, called the Medication Regimen Complexity Intensive Care Unit, or MRC-ICU score, quantifies the complexity of a patient’s medication regimen to predict ADEs that could be prevented by timely CCP intervention. The tool will be integrated into visualization dashboards, called ICView, that guide CCPs in preventing ADEs in patients, thus improving patient safety.

Using machine learning to predict when pharmacists are needed for critical care

In the second research study, the team plans to develop and validate machine learning predictive models to optimize the workflow for CCPs by identifying which patients need CCP intervention. The models will be integrated into the MRC-ICU tool and will summarize CCP workload through predicting total CCP interventions. The models will also guide CCP care by predicting ADEs that may be prevented by CCP intervention. Identifying which interventions—and under which conditions—can improve outcomes by helping administrators better determine workload, such as optimal CCP-to-ICU ratios through providing workload insights.

“That tool quantifies the complexity of what the patient is taking in the ICU with the goal that it's going to help predict how much effort you need from a pharmacist. If you knew that, then you could say, ‘Okay, this patient needs an hour of effort, and we have 20 patients in the ICU.’ Well, that's 20 hours, so that is probably more than one 8-hour shift. And so, you'd be able to have those kinds of conversations around workload.”
- Dr. Andrea Sikora

Collectively, Dr. Sikora’s innovative AHRQ research shows the exciting ways that machine learning and artificial intelligence can be used in healthcare. In addition to benefiting patients by facilitating CCPs interventions when medication-related problems are found, the research can benefit the healthcare system at large by providing the justification for the critical role that CCPs play.

This Research is Featured in

Principal Investigator

Sikora, Andrea

Organization

University of Georgia

Research Profile

Machine Learning Validation of Medication Regimen Complexity for Critical Care Pharmacist Resource Prediction

Funded Amount

$2,156,598

Research Type

Emerging

This Research is Featured in

Related Stories