Machine Learning for Early Identification and Management of Pulmonary Embolism
A machine learning risk stratification model to improve recognition and management of high risk PE while also reducing hospital utilization for low risk patients.
DIHI RFA
Annual Request for Application Projects
The mission of the Duke Institute for Health Innovation (DIHI) is to catalyze transformative innovation in health and healthcare, and to help cultivate a community of innovation and entrepreneurship across Duke University and the health enterprise. Our annual request for innovation applications is open to all faculty, staff, students, and trainees across Duke University and Duke University Health System.
Our next round of applications will open in Summer 2020.
In an environment of value-based care, having an effective platform for innovation will be one way to ensure that we continue to focus on new models of care. Delivering highest-quality care while exceeding quality goals and enhancing operational efficiencies will remain priorities for our clinical enterprise in the foreseeable future."
2020 RFA Innovation Pilots
Read More »
Dermatology Clinical Decision Support in Primary Care
Improving dermatology access, care delivery and cost via machine learning assisted risk stratification.
A Population Health Strategy for NAFLD using an E-Consult Platform and NLP
A population health solution for NAFLD with an ultimate goal to optimize health care resources by improving access for high risk patients and minimizing unnecessary referrals.
Using Xealth to Manage Population Health and Engage Patients with Chronic Kidney Disease
Guiding appropriate specialty consultation and delivering tailored patient educational content.
Transforming Epilepsy Surgery with Augmented Reality
Placing patient-specific 3D images in the Neurosurgeon’s field of view using augmented reality and advanced imaging to increase precision for epilepsy surgery.
A Unified Framework for Predicting Bacteremia and Interpreting Blood Culture Results
Estimates the pre-test probability of bacteremia and post-test probability of blood culture results in hospitalized patients for EHR-based clinical decision support.
Predictive Model for Assisted Triage of Traumatic Brain Injury
A model to aid actionable mild to moderate TBI triage decisions in the ER.
A Virtual Operating Room Hub to Facilitate Peri-Operative Communication
Augment OR case review huddles with a virtual operating room hub to facilitate communication across shifts and within shifts in the OR.
Development of a Maternal Early Warning System Using Machine Learning
Identify obstetric patients at risk for clinical deterioration by using a variety of patient clinical parameters obtained from multiple data modalities and predictive modeling of pregnancy specific, patient related changes.
Development and Implementation of a Hospital at Home Program in Wake County
A Hospital at Home program in Wake County that would allow patients to be treated for hospital-level conditions in their homes.
Using a Mortality Risk Predictor Model for Advance Care Planning via a Transitions of Care Toolkit
Identify high risk mortality inpatients to provide them with a Transition of Care Toolkit to help with advance care planning.
2019 RFA Innovation Pilots
Reducing In-Hospital Mortality
A machine learning model to predict in-hospital mortality, run at the time of admission to the hospital, for all adult patients.
Using Machine Learning in Emergency Department Patient Flow
Develop a clinical dashboard built on existing DIHI databases and predictive modeling to assist in providing more efficient dispositions for emergency department patients by predicting clinical trajectory and need for admission.
Detection and Treatment of Patient Deterioration
Develop and implement a machine learning risk score for patient deterioration.
Duke Virtual Assistant (Chatbot)
A Virtual Assistant to help patients understand whether they should go to the ER immediately rather than sending the email they are about to send to their physician.
Machine-learning algorithm to predict ICU readmissions in the cardiothoracic surgical population
Develop predictive models to identify rapidly and accurately CTS patients at high-risk (and high-cost) for postoperative clinical deterioration necessitating readmission from the SDU back to the ICU.
Machine Learning Models to Recognize Early Clinical Deterioration in DUHS Pediatric Inpatients
Develop machine learning methods that will identify children at risk for clinical deterioration.
High-Sensitivity Cardiac Troponin (hsTn) Decision Aid
Create a chest pain assessment tool which integrates with Epic using FHIR to rule-in or rule-out acute MI.
Understanding Underlying Health Care Costs to Deliver Value-Based Care
Develop a data repository with key clinical and administrative data with the goal of understanding the actual cost of treating ductal carcinoma of the breast within the Duke Health System.
Transforming Cancer Care: Bringing PCPs “Back” into Cancer Care through Onco-Primary Care
Implement a blood pressure management clinical pathway to better incorporate PCPs as active members of the cancer care team.
Telehealth examination versus clinical examination in the detection of rotator cuff tears
Simulated telehealth evaluation of shoulder pain using patient self-examination will be compared to traditional clinical examination of shoulder pain using MRI as the gold standard to determine accuracy in detecting rotator cuff tears.
2018 RFA Innovation Pilots
Optimizing patient-reported outcomes data and workflows in MaestroCare
Demonstrate the feasibility and utility of integrating electronic patient-reported outcomes (ePROs) into the existing Epic electronic health record for use in outpatient cancer care.
Improving racial disparities in unmet palliative care needs among ICU family members with a needs-targeted app intervention
Systematically engaging families and helps ICU clinicians deliver needs-targeted palliative care.
Utilization of machine learning to read transthoracic echocardiograms: Project normal echocardiogram
To develop and validate a computer algorithm to correctly estimate left ventricular ejection fraction (LVEF) as an initial step toward a fully automated echocardiogram evaluation.
Voices of Duke Health Listening Booth
Inviting healthcare providers, staff, students, trainees, patients, and visitors to have conversations about what is meaningful in your lives, work, and relationships.
Early identification of cardiac decompensation and cardiogenic shock
Duke Heart and DIHI are collaborating to implement the phenotype dashboard and model to support real-time identification of patients experiencing or at high risk of experiencing cardiac decompensation.
Improving primary care access for residents and fellows through electronic visits
Developing an innovative GME Trainee Primary Care eVisit Program within Duke Primary Care.
Innovative predictive model to anticipate steroid-induced hyperglycemia and guide insulin regimens
Identify optimal, individualized treatment policies for insulin dosing in the context of systemic glucocorticoids.
RFID surgical instrument tracking and optimization of the operating room
Implement an RFID system that could be integrated into the OR to measure instrument usage autonomously.
Patient-directed acoustic surveillance for LVAD complications
Patients upload cardiac tones from digital stethoscopes to develop acoustic signatures associated with LVAD complications.