New AI Models Predict Mortality and Safety Risks
Researchers are piloting new AI models in acute care to improve patient outcomes. One system automates the analysis of patient safety incidents to extract actionable insights, while another model aims to predict medium-term mortality risk in heart failure patients. The push is toward explainable AI that integrates with clinical workflows to reduce cognitive load in high-acuity settings like the ICU.
- Epic's AI, including its Deterioration Index, analyzes real-time patient data from monitors to flag early signs of decline, with one health system reporting a 17% decrease in mortality after implementation. Another Epic predictive model for inpatient fall risk saved one health network 1,000 nursing hours and eliminated 6.54 million clicks annually by removing manual fall-risk assessments. - To become a board-certified informatics nurse (NI-BC), you generally need an active RN license, a BSN, two years of full-time RN experience, and 30 hours of continuing education in informatics within the last three years. You must also meet practice hour requirements, such as 2,000 hours in informatics nursing over the last three years. - ICU nurses frequently cite EHR usability issues as a major frustration, including slow system response and login times, excessive clicking to complete tasks, and a lack of intuitive design that forces more time on the computer than with the patient. These usability problems are not just an inconvenience; research suggests that EHRs which are difficult to use are also less likely to catch medical errors that could harm patients. - Fast Healthcare Interoperability Resources (FHIR) is an HL7 standard that enables data exchange between different health IT systems using modern web technologies. For an ICU informaticist, this is crucial for integrating data from bedside monitors, ventilators, and infusion pumps directly into the EHR, providing a more complete and real-time view of the patient's condition. - Epic is developing and deploying hundreds of AI features, including "Art for Clinicians," an AI assistant that helps draft clinical notes from conversations, and "Cosmos AI," which uses a dataset from over 300 million patients to build predictive models for things like readmission risk. These tools are designed to be embedded directly into clinical workflows to reduce administrative burden. - When positioning your ICU experience for an informatics role, emphasize your expertise in managing high-acuity patient data, understanding critical care workflows, and identifying the technological needs of frontline staff. Employers value this clinical experience as it provides essential context for designing and implementing IT systems that are practical and effective in a real-world setting. - Common complaints from nurses about EHRs include the feeling that the system was not designed with clinical input, leading to counter-intuitive workflows and an inability to accurately chart specific patient needs due to rigid, pre-determined fields. A survey showed that 65% of nurses did not feel they had a voice in EHR changes, and 39% did not feel they could even ask for fixes.
