This research will develop and validate machine learning enhanced predictive models improving the allocation of critical care pharmacists to intensive care units to reduce adverse drug events.
This research will utilize user-centered design to develop and implement a self-management education and support module called Support-Engage-Empower-Diabetes (SEE-Diabetes) to improve diabetes management in older adults.
This research will test and validate a machine learning predictive analytic intervention to optimize the timely and appropriate use of interpreters for hospitalized patients with language barriers and complex care needs.
This research will develop and test novel approaches for using electronic health record metadata to characterize and evaluate hospital discharge planning practices, with the long-term goal of providing tools to monitor and strengthen specific behaviors within discharge workflows that optimize post-acute transitions of care.
This research developed and evaluated the usability of a shared decision making tool that incorporates risk information and patient and caregiver preferences for hospital-level care in the home and found the tool to be usable and scalable.
This study explored the use of and experiences with sharing inpatient notes during pediatric oncology hospitalization between parents and clinicians, with findings suggesting its potential to positively impact parent engagement and understanding in pediatric inpatient care and providing an opportunity to identify potential patient safety issues.
This study created, trained, and tested machine learning (ML) algorithms to predict emergency department returns and morbidity or mortality among returns for COVID-19 patients, with results showing ML’s potential to inform clinicians on whether admission to the hospital may be needed for the prevention of complications and to prioritize resources for higher risk patients.
Wearable smart glasses linked to a web-based telemedicine system may improve prehospital care coordination, free emergency care professionals’ hands, and enhance patient outcomes.
This research developed and tested a machine learning classifier using natural language processing to identify hospitalized patients at risk for HIV acquisition or transmission due to substance misuse or high-risk sexual behavior. While the model performed moderately well, gaps in how patient information is documented made accurate risk prediction difficult, highlighting the need for better data and tailored interventions in hospital settings.
This study evaluated the feasibility of a mobile personal health record application integrated with electronic health records using Fast Healthcare Interoperability Resources standards for longitudinal care planning, goal tracking, and communication with providers among families of children and youth with special healthcare needs. It found that while technical feasibility and early adoption were high, sustained use was limited due to technical challenges and mixed user experiences.
