Artificial intelligence-based health IT tools to optimize critical care pharmacist resources through adverse drug event prediction

点击翻译按钮获取中文摘要

Critically ill patients are at heightened risk of adverse drug events (ADEs) that worsen outcomes. Critical care pharmacists (CCPs) prevent ADEs, improve patient-centered outcomes, and reduce healthcare costs through performing medication interventions. However, CCPs are an inequitably distributed and non-optimized healthcare resource due to lack of health information technology (IT)-based predictive tools that can identify CCP-driven medication interventions that prevent ADEs. To identify these preventative medication interventions, delineating the medication outcome causal pathway among patient features, medication interventions, ADEs, and patient-centered outcomes is required. Here, novel causal inference methodologies incorporating artificial intelligence (AI) and machine learning (ML) will be applied for the first time to medication safety questions in the ICU. The strategy will focus on developing a novel scoring tool designed for prediction, the medication regimen complexity-intensive care unit (MRC-ICU) Scoring Tool, to predict intervenable ADEs in this causal pathway that are predictable by patient features, preventable by CCPs, and otherwise associated with poor outcomes. The central hypothesis of this AHRQ Health Services Research Project (R01), based on preliminary data, is that an AI-informed dashboard visualizing the medication outcome causal pathway can optimize CCP care to improve patient-centered outcomes. The objective of this work is to apply AI and ML methodology to multi-center data to create prediction tools for integration into visualization dashboards that answer vital questions including (1) what is the best metric for predicting ICU intervenable events and CCP workload?; (2) what causal factors of intervenable events can be prevented by CCPs?; (3) how can CCPs efficiently use AI-based predictions at the bedside? The long-term goal of the proposed work is to establish validated prediction models that can be embedded into dashboards in the electronic health record (EHR) to help guide CCP delivered care. The rationale for this work is that it will establish the MRC-ICU Scoring Tool as a means of predicting medication safety events and CCP interventions. The central hypothesis will be tested using large, multi-center datasets of ICU patients via these specific aims: (1) Create robust prediction models of intervenable events to guide CCP medication interventions; (2) Explore causal relationships among intervenable events, CCP interventions, and outcomes; (3) Design an EHR-integrated platform (ICView) to visualize predictions to guide CCP care. Applying user centered design methods to create a health IT product will result in a visualization dashboard (ICView) that houses MRC-ICU based, AI-informed prediction models for CCP interventions that can improve patient outcomes. This innovative approach applies state-of-the-art ML methodology to causal outcome predictions using the novel MRC-ICU Scoring Tool. The proposed work is significant because any advances in the understanding of how CCPs improve outcomes would have a profound public health impact due to their established role on the interprofessional healthcare team.