
Addressing Barriers to Care Through Strategic, Data-Based Solutions
Optimizing Delivery of Care
Duke Surgery is investing in the future in impactful ways, with innovative infrastructures being implemented within and outside the walls of the hospital with aims of optimizing how—and to whom—we deliver care.
Paramount to that success is an awareness of the way that care is delivered, and the recognition that exceptional surgical care is inconsequential unless patients are able to access it.
Data scientists, health systems operations managers, surgeon–scientists, and a host of experts in other fields are working closer than ever to methodically and systematically identify, address, and rectify issues that limit how patients seek and receive surgical care.
Together, Duke Surgery teams are driving change in health services.
Defining the Approach • CHARTing a New Course • Transformation Through Information • Driving Sustainable Change
Defining the Approach
The first step to meaningfully evaluating any component of surgical care is a foundation of well-defined metrics and methods that allow for clear tracking, interpretation, and sharing of information.
“Using data to quantify complex human behavior is challenging,” explains Nrupen Bhavsar, PhD, Associate Professor in Surgery in the Division of Surgical Sciences, and director of the social informatics program in the Duke Clinical and Translational Science Institute (CTSI).

Associate Professor in Surgery, Division of Surgical Sciences
“And, this is a relatively new field, so you can’t necessarily consult the literature to see how someone else has done it before,” Dr. Bhavsar continues. “But I think that is where the Department of Surgery has been great in providing resources for us to define and develop these methodologies and be a leader in this space.”
In short, while available health data has an immense potential for improving delivery of care, the current methods of collection, analysis, and management are inconsistent and often inefficient. Through collaborative efforts of faculty and staff like Dr. Bhavsar, with his expertise in quantitative epidemiology, informatics, and biostatistics, the goal is to build tools and infrastructures that minimize these challenges and contribute to meaningful applications of data for better assessment of the delivery of patient care.
For example, location of patient residence is often a relevant indicator of the social and environmental factors to which a patient may be exposed, and which contribute to a patient’s health. However, current data collection systems provide limited interpretations of the impact of this variable.
To navigate this complex challenge, Dr. Bhavsar and team collaborate with faculty within the Department of Civil and Environmental Engineering to utilize artificial intelligence and machine learning tools to look deeper into the electronic health record (EHR) to develop novel phenotypes of social, environmental, and climate exposure that inform health risk.
"It’s really lowering the burden and making research more efficient for our investigators within the department and across the university."
– Nrupen Bhavsar, PhD
The data from these approaches are compiled and added to other social and environmental data on a publicly available resource network, called the Social, Environmental, and Equity Drivers (SEED) Health Atlas, that allows other clinician-researchers seeking insights on the impact of social determinants of health —such as impacts of socioenvironmental factors on inflammation, eligibility for transplant, and equity in access to care—to easily visualize, download, and link these curated data with EHR data. The website also allows researchers to collaborate with colleagues at other institutions through access to data on comparable populations and variables across the country.
“We have had an influx of health services research over the past couple of years in the department, and of people who have a keen interest in social determinants of health,” says Dr. Bhavsar. “What we do with the SEED Atlas is consolidate resources into one single location, so it’s really lowering the burden and making research more efficient for our investigators within the department and across the university.”

Assistant Professor of Surgery, Division of Abdominal Transplant Surgery
Further supporting this priority is the recent establishment of the Duke Surgery Collaboratory for Health Services Research (HSR). Led by director of HSR Lisa McElroy, MD, MS, Assistant Professor of Surgery in the Division of Abdominal Transplant Surgery, the Collaboratory aims to bring together and support faculty engaged in health services research, and to facilitate efficient access to the department’s robust research support infrastructure along with a wealth of diverse collaborators across Duke.
“The goal of health services research is to provide data, evidence, and tools to make health care safe, effective, equitable, affordable, and patient-centered,” says Dr. McElroy. “We have a number of faculty within the department who are now here to conduct health services research, and the Collaboratory will support these surgeon–scientists and help them be more successful in their research endeavors.”

CHARTing a New Course
One such endeavor, which is seeking to improve the delivery of transplant care, is the Consortium for the Holistic Assessment of Risk in Transplant (CHART).
Entering its third year of continued growth, CHART is a multi-institutional research consortium of 13 high-volume U.S. transplant centers. The coordinating center team, led by Dr. McElroy, enlists a multidisciplinary team from the Departments of Surgery, Medicine, and Biostatistics & Bioinformatics, along with Duke AI Health and the Pratt School of Engineering, to fulfill its mission to, “impart foundational change to the process of collecting, organizing, and employing data to determine patient eligibility for transplant, leading to improved equity in access to care, increased transparency in the transplant selection process, and improved value for patients and clinicians.”
"A big portion of this project is collecting all of the data through the patient’s entire pipeline."
– Lisa McElroy, MD, MS
“Nationally, there is a database that keeps track of patients who are waiting for an organ, but what is lacking is data for prior to when the patient is placed on the national waitlist,” explains Dr. McElroy, co-principal investigator for CHART.
“So, a big portion of this project is collecting all of the data through the patient’s entire pipeline: from when they are referred, to when they are evaluated and go through testing at their treatment site, to when they are approved to go onto the national waitlist, to when they actually receive their transplant.”
Of particular interest to the members of CHART are variables that contribute to why a patient might not receive the transplant they need, such as social determinants of health or clinical factors.
"We are working with our informatics team to essentially create this map for other centers so that we have all of the same type of data coming back to this common model.”
– Jennifer Gagnon
As a growing program, developing a database to effectively manage these variables of interest across all its institutions requires specialized expertise and the recruitment of instrumental team members like Ursula Rogers, senior informaticist, to pilot its success.
“It’s difficult to understand the landscape of other institutions,” says Jennifer Gagnon, Director of Healthcare Analytics and Information. “So, we are working with our informatics team to essentially create this map for other centers so that we have all of the same type of data coming back to this common model.”
With this “playbook” of a database being employed by all participating centers, researchers will soon be able to investigate nearly any aspect of the transplant candidacy process at a national level.

“The scope of the population of people that we can look at is huge and diverse,” says biostatistician Tyler Schappe, MS. “Having centers all over the country and in different regions, with different cultural practices and different backgrounds, will allow us to capture such a rich set of information with this collective data set.”
Dr. McElroy agrees, adding, “What started out as my interest in access to the transplant waitlist has blossomed into this data asset that can support really meaningful studies along the whole continuum of transplant care.”
Transformation Through Information
Health systems are constantly collecting data across the course of patient care. To find actionable meaning in this wealth of information, Duke clinician¬–scientists work with data scientists and biostatisticians, who expertly translate robust, complex datasets into interpretable outputs used to develop impactful solutions.
With a focus on enhancing Duke Surgery’s operational efficiency and resource allocation, the Laboratory for Transformative Administration (LTA) continues its collaborative mission of using data science methods and models to ensure patients are receiving surgical care when and where they need it.
“One of the first questions asked of the LTA was whether the ORs were being used as efficiently as they could be used,” says Daniel Buckland, MD, Assistant Professor of Emergency Medicine and Medical Director for the LTA.

Assistant Professor in Surgery; Lead Data Scientist, LTA

Assistant Professor of Emergency Medicine; Medical Director, LTA
To find the answer, the LTA team, supported by Bruce Rogers, PhD, Assistant Professor in Surgery and Lead Data Scientist, is analyzing logistical data to identify limitations in OR usage and developing models to test various strategies to resolve them. As a result, the department has implemented several of the LTA's modeled changes to significantly reduce inefficiencies in OR scheduling and staffing.
Where it was once a largely individualized and easily disrupted process, it is now a much more resilient and streamlined operation that is ultimately helping to reduce patient wait times.
“Patients are receiving their surgeries when they need them, regardless of the persistence of a nursing shortage or a pandemic,” says Dr. Buckland.
Recent expansions of the expertise within the LTA have also allowed for greater exploration of artificial intelligence and machine learning (AI/ML) as tools for innovative solutions development.
“If we could take two very different sources of data...and combine them into one model, we think it could find predictive patterns for how long a patient’s post-operative hospital stay will be.”
– Hamed Zaribafzadeh
Hamed Zaribafzadeh, a data scientist who has been a part of the LTA since 2022, for example, is developing machine learning models that can integrate different types of data to predict patient care factors such as length of stay, total procedure time, and resource allocation.
“If we could take two very different sources of data—one being the EHR text and the other being health scan images—and combine them into one model, we think it could find predictive patterns for how long a patient’s post-operative hospital stay will be,” says Zaribafzadeh. Successfully doing so, he says, could help hospital administration better allocate resources to improve efficiency and patient care quality in real time.

Vice Chair of Innovation
Taking AI/ML explorations and data scientist collaborations one step further is the Surgical Artificial Intelligence and Innovation Laboratory (SAIIL) network.
Led by Ozanan Meireles, MD, Vice Chair of Innovation and Associate Professor of Surgery, the multi-institutional SAIIL network, now headquartered at Duke, aims to build a global, information-sharing infrastructure to facilitate the development of trustworthy AI/ML tools to improve surgical care.
Built on principles of education, research, and policy, SAIIL provides opportunities for surgeons to collaborate with and learn from AI/ML experts to foster new ideas and responsible applications of these rapidly expanding technologies.
Driving Sustainable Change
Throughout these advancements, the desire to eliminate barriers and promote change, collaboration, and democratization of data and knowledge is clear, both within and beyond the Duke Department of Surgery.
Methodical development of data infrastructures and solutions made possible by broad, but closely tethered, networks of interdisciplinary collaboration and mentorship are providing researchers with the necessary tools to ask and meaningfully answer important questions surrounding access to and delivery of surgical care.
Through investment in collaborative science and innovative technologies, Duke Surgery is opening doors and building pathways toward improved, equitable, and accessible surgical care for all patients.