Shane Cooke
Leveraging AI in Learning Health Systems to improve patient outcomes, reduce costs
April 07, 2025
By Shane Cooke
It’s no secret that hospitals are facing some of their toughest challenges ever. Workforce shortages and burnout are commonplace and operating budgets are tight. Hospital leaders need to find increasingly creative ways to address these challenges and set themselves up for long-term success.
Enter artificial intelligence (AI) technology. Hospitals are adopting the technology at a breakneck pace, and in seemingly every facet of care. This article will explore one specific area that is seeing overwhelmingly positive results – the power of AI to improve learning health systems in the ICU.
The ICU: An unpredictable care setting
One specific place that hospitals are seeing significant positive impact from implementing AI solutions is in the intensive care unit (ICU). A complex and unpredictable care setting, the ICU is also one of the most expensive areas in healthcare. Adding to workforce challenges seen across the healthcare ecosystem, the ICU environment also often requires quick decision making, with each second a matter of life or death. As with other areas in the hospital, ICU clinicians utilize the electronic health record (EHR) for tracking basic information like medications, inputs and outputs, and key vital signs. However, clinicians in the ICU environment have realized that the EHR leaves significant gaps in data, potentially impacting critical decisions like escalation and de-escalation of care.
A clinical solution that fills in the gaps for a complex environment
There are a variety of IT solutions in healthcare that hospital leaders may consider, but many think first of the EHR. While the EHR is a powerful and important tool, there are limitations when relying exclusively on vital signs documented in the EHR. Relying solely on this data can lead to gaps in a patient’s physiologic picture in critical care environments, which can lead to an increased risk of clinical consequences that may result in an adverse event. However, leveraging proven AI clinical decision support tools to get access to more comprehensive clinical data gives providers a holistic view of a patient in critical care, paired with an uninterrupted flow of data collection. This powerful combination fills in those gaps found in the EHR and may improve risk assessment with clinical decision support. Automating workflows so providers can make the best decisions in stressful moments, compared to relying on EHR data alone, ultimately helps deliver the best possible patient outcomes.
These types of AI-powered clinical solutions can also address the very critical issue of workforce burden and burnout. Implementing critical AI-driven tools can help short-staffed critical care nurses and doctors streamline communication and prioritize where to focus attention, by providing immediate access to aggregated, trended patient data on one screen. This helps alleviate pressure on nurses to clearly articulate issues after witnessing a few events that might not have surfaced during physician rounds. These solutions paint the picture automatically, so nurses can simply show the visualization of what they are observing.
Furthermore, as increasing demand for resources (beds and clinicians) in critical care settings has become the new norm, it’s more important than ever to know when to both escalate and de-escalate care. The EHR also relies on providers to enter data correctly and consistently, which can be challenging in a stressful environment with many competing priorities. There’s also significant cognitive burden required on the part of the provider to locate the data they need and analyze it in a meaningful way. Utilizing FDA-cleared risk algorithms that leverage AI-based physiology-specific deterioration pathways will assess patient data and show clinicians how the patient is trending. This tool allows providers to quickly identify high-risk patients who may need closer monitoring, but also just as importantly, to identify those patients who are doing well and may be ready for de-escalation.
A learning health system
According to the Agency of Healthcare Research and Quality, a Learning Health System is an approach to healthcare that leverages real-world data and evidence to build a culture of continuous learning and improvement. There is another important benefit hospitals are realizing as a result of implementing comprehensive AI platform solutions. These tools are quickly becoming the backbone of their learning health system in the ICU to drive the continuous flow of collecting data and therefore also continuously improving outcomes. Today, in a way like never before, hospital leaders and clinicians can see which clinical practices are effective, which protocols are being adhered to and what improvements need to be made. By leveraging this powerful data and putting it into practice, these insights are leading to improved clinical efficiencies and patient outcomes, such as reducing ICU length of stay, time on ventilation, readmissions, etc., as well as lowering costs.
Conclusion
The American Medical Association conducted an Augmented Intelligence Survey during the time period of August 2023 and November 2024, where they examined how physicians' views on AI in healthcare have evolved. They found that physician adoption of AI tools is rising, with 66% of surveyed physicians in 2024 indicating they use AI in their practice—a significant increase from 38% in 2023. That stat is just one of many that signifies AI is here to stay. While its use in healthcare is likely only in its infancy, the significant impact this technology has already had on improving patient outcomes and reducing costs is without question. The AI solutions that will realize long-term success will be the ones that can be seamlessly integrated into existing workflows, are easy to access and are able to demonstrate proven results and return on investment.
About the author: Shane Cooke joined Etiometry in 2019 as president & CEO, bringing over 20 years of experience in the medical device and pharmaceutical market spaces in a variety of sales, marketing, strategy and portfolio management roles. Before joining Etiometry, Shane spent more than five years as chief strategy officer of Cheetah Medical, which was acquired by Baxter International in 2019. Prior to Cheetah, Shane spent 11 years with Covidien in the patient care, vascular therapies and corporate sectors, with positions such as: corporate strategy, market and competitive intelligence, leading the market development center of excellence and leading strategy efforts for Japan, Europe, Australia and Canada. Shane holds a BA in psychology from the University of Rochester, as well as an MBA from Suffolk University.
It’s no secret that hospitals are facing some of their toughest challenges ever. Workforce shortages and burnout are commonplace and operating budgets are tight. Hospital leaders need to find increasingly creative ways to address these challenges and set themselves up for long-term success.
Enter artificial intelligence (AI) technology. Hospitals are adopting the technology at a breakneck pace, and in seemingly every facet of care. This article will explore one specific area that is seeing overwhelmingly positive results – the power of AI to improve learning health systems in the ICU.
The ICU: An unpredictable care setting
One specific place that hospitals are seeing significant positive impact from implementing AI solutions is in the intensive care unit (ICU). A complex and unpredictable care setting, the ICU is also one of the most expensive areas in healthcare. Adding to workforce challenges seen across the healthcare ecosystem, the ICU environment also often requires quick decision making, with each second a matter of life or death. As with other areas in the hospital, ICU clinicians utilize the electronic health record (EHR) for tracking basic information like medications, inputs and outputs, and key vital signs. However, clinicians in the ICU environment have realized that the EHR leaves significant gaps in data, potentially impacting critical decisions like escalation and de-escalation of care.
A clinical solution that fills in the gaps for a complex environment
There are a variety of IT solutions in healthcare that hospital leaders may consider, but many think first of the EHR. While the EHR is a powerful and important tool, there are limitations when relying exclusively on vital signs documented in the EHR. Relying solely on this data can lead to gaps in a patient’s physiologic picture in critical care environments, which can lead to an increased risk of clinical consequences that may result in an adverse event. However, leveraging proven AI clinical decision support tools to get access to more comprehensive clinical data gives providers a holistic view of a patient in critical care, paired with an uninterrupted flow of data collection. This powerful combination fills in those gaps found in the EHR and may improve risk assessment with clinical decision support. Automating workflows so providers can make the best decisions in stressful moments, compared to relying on EHR data alone, ultimately helps deliver the best possible patient outcomes.
These types of AI-powered clinical solutions can also address the very critical issue of workforce burden and burnout. Implementing critical AI-driven tools can help short-staffed critical care nurses and doctors streamline communication and prioritize where to focus attention, by providing immediate access to aggregated, trended patient data on one screen. This helps alleviate pressure on nurses to clearly articulate issues after witnessing a few events that might not have surfaced during physician rounds. These solutions paint the picture automatically, so nurses can simply show the visualization of what they are observing.
Furthermore, as increasing demand for resources (beds and clinicians) in critical care settings has become the new norm, it’s more important than ever to know when to both escalate and de-escalate care. The EHR also relies on providers to enter data correctly and consistently, which can be challenging in a stressful environment with many competing priorities. There’s also significant cognitive burden required on the part of the provider to locate the data they need and analyze it in a meaningful way. Utilizing FDA-cleared risk algorithms that leverage AI-based physiology-specific deterioration pathways will assess patient data and show clinicians how the patient is trending. This tool allows providers to quickly identify high-risk patients who may need closer monitoring, but also just as importantly, to identify those patients who are doing well and may be ready for de-escalation.
A learning health system
According to the Agency of Healthcare Research and Quality, a Learning Health System is an approach to healthcare that leverages real-world data and evidence to build a culture of continuous learning and improvement. There is another important benefit hospitals are realizing as a result of implementing comprehensive AI platform solutions. These tools are quickly becoming the backbone of their learning health system in the ICU to drive the continuous flow of collecting data and therefore also continuously improving outcomes. Today, in a way like never before, hospital leaders and clinicians can see which clinical practices are effective, which protocols are being adhered to and what improvements need to be made. By leveraging this powerful data and putting it into practice, these insights are leading to improved clinical efficiencies and patient outcomes, such as reducing ICU length of stay, time on ventilation, readmissions, etc., as well as lowering costs.
Conclusion
The American Medical Association conducted an Augmented Intelligence Survey during the time period of August 2023 and November 2024, where they examined how physicians' views on AI in healthcare have evolved. They found that physician adoption of AI tools is rising, with 66% of surveyed physicians in 2024 indicating they use AI in their practice—a significant increase from 38% in 2023. That stat is just one of many that signifies AI is here to stay. While its use in healthcare is likely only in its infancy, the significant impact this technology has already had on improving patient outcomes and reducing costs is without question. The AI solutions that will realize long-term success will be the ones that can be seamlessly integrated into existing workflows, are easy to access and are able to demonstrate proven results and return on investment.
About the author: Shane Cooke joined Etiometry in 2019 as president & CEO, bringing over 20 years of experience in the medical device and pharmaceutical market spaces in a variety of sales, marketing, strategy and portfolio management roles. Before joining Etiometry, Shane spent more than five years as chief strategy officer of Cheetah Medical, which was acquired by Baxter International in 2019. Prior to Cheetah, Shane spent 11 years with Covidien in the patient care, vascular therapies and corporate sectors, with positions such as: corporate strategy, market and competitive intelligence, leading the market development center of excellence and leading strategy efforts for Japan, Europe, Australia and Canada. Shane holds a BA in psychology from the University of Rochester, as well as an MBA from Suffolk University.