Browse by Medical Category
Research at Mass General
Call for more information
About Healthcare Systems Engineering
Healthcare Systems Engineering (HSE) at Massachusetts General Hospital is leveraging analytics and operations research methodologies to enable system-level changes in a large academic medical center, with the aim to improve patient care and maximize the efficient use of resources. HSE has different components including the Mass General-Massachusetts Institute of Technology (MIT) collaboration, which partners with students and faculty at the MIT Sloan School of Management, and the Mass General Analytics Alliance, which brings together resources from across the hospital to tackle operational challenges using data and analytics.
HSE was created in 2017 to serve as a model that could be scaled to other hospitals to improve their operations and analytics. The group’s operations research has generated quantifiable benefits in areas such as:
The success of HSE has created great demand for new applications of data-driven research at the hospital. Currently, the group is working on projects and collaborates with Mass General's:
Additionally, the group forms the analytic backbone of the hospital’s Capacity Operations Committee, which was initiated by Mass General’s President Peter L. Slavin, MD, to reduce avoidable admissions and readmissions, intraday delays, and emergency department and hospital congestion.
A decade ago, Massachusetts General Hospital and the Massachusetts Institute of Technology (MIT) began a collaboration with a series of small pilot projects to maximize efficient use of hospital resources. This collaboration spawned the creation of Healthcare Systems Engineering (HSE) at Mass General. Since its inception, HSE has created effective solutions to multiple health care challenges, especially in the area of perioperative care (the period beginning with a surgeon’s decision that a patient needs surgery and ending with the patient’s discharge from recovery after surgery).
Over the years, HSE has:
Request for Proposals
Due to hospital wide interest in new system design tools for complex operational problems, Mass General's Peter Dunn, MD, vice president, Perioperative Services and HSE, and MIT’s Retsef Levi, PhD, in 2012 initiated a hospital-wide Request for Proposals (RFP) that led to novel collaborations between the Dunn/Levi team and working groups from many divisions across Mass General.
Below are the criteria for the RFP:
Dr. Peter Dunn currently serves as the vice president of Perioperative Services and HSE at Mass General where he co-leads all aspects of perioperative operations with the associate chief nurse for perioperative services. Prior to this role, he was the clinical director and served as executive vice chair of the Department of Anesthesia, Critical Care and Pain Medicine (DACCPM). Dr. Dunn pursued initial post graduate training in surgery at George Washington University and basic science research at the National Institutes of Health, and then joined Mass General in 1995 where he received his training in both anesthesia and critical care. As a junior staff physician, Dr. Dunn focused on clinical care and teaching medical students, residents and fellows.
More recently, along with colleagues from the MIT Sloan School of Management and a team at Mass General, he has developed tools and methodologies to redesign the provision of patient care in the perioperative environment. This unique collaboration focuses on the utilization of resources throughout the patient care experience, from surgeon’s offices to the post-operative floor care with the aim of improving the quality, efficiency and cost of delivering perioperative care to patients. In addition to his responsibilities at Mass General, he is also the co-chair of the Partners Perioperative Technology Committee, an assistant professor of Anesthesia at Harvard Medical School and the program director of the Perioperative Administration Fellowship at Mass General.
Retsef Levi is the J. Spencer Standish (1945) Professor of Operations Management at the MIT Sloan School of Management. He is a member of the Operations Management Group at MIT Sloan and affiliated with the MIT Operations Research Center. Levi also serves as the faculty co-director of the MIT Leaders for Global Operations (LGO).
Levi's current research is focused on the design of analytical data-driven decision support models and tools addressing complex business and system design decisions under uncertainty in areas such as health and healthcare management, supply chain, procurement and inventory management, revenue management, pricing optimization and logistics. He is interested in the theory underlying these models and algorithms, as well as their computational and organizational applicability in practical settings.
Levi has been leading several industry-based collaborative research efforts with some of the major academic hospitals in the Boston area, such as Mass General, Beth Israel Deaconess Medical Center (BIDMC), Boston Children’s Hospital and across the U.S. (e.g., Memorial Sloan Kettering Cancer Center, NYC Presbyterian Hospital System and the American Association of Medical Colleges).
Levi received the NSF Faculty Early Career Development award, the 2008 INFORMS Optimization Prize for Young Researchers, the 2013 Daniel H. Wagner Prize and the 2016 Harold W. Kuhn Award. Levi teaches regularly courses on operations management, analytics, risk management, system thinking and health care to students from various degree and non-degree programs including master of business administration, executive master of business administration, PhD, master and undergraduate students as well as executive education programs. His health care lab course attracts students from across the MIT campus and engages major industry partners and leaders. Levi has graduated 10 PhD students, 34 master students and six postdoctoral fellows. He was also awarded several prestigious teaching awards.
Bethany Daily is the executive director of Perioperative Services and HSE at Mass General. Her responsibilities include the strategic direction of information systems, statistical and financial reporting, process improvement, human resources management and facilities planning. Daily is also the Mass General program director for the Mass General/MIT collaboration, which uses operations research methods to optimize patient flow in many areas throughout the hospital. Bethany received her bachelor of arts at the University of Michigan-Ann Arbor and her master of healthcare administration at University of North Carolina-Chapel Hill.
Dr. Kyan Safavi is a faculty member of the DACCPM at Mass General. He is the David F Torchiana Fellow in Healthcare Policy and Management for the Mass General Physicians Organization (MGPO). Along with colleagues at Mass General and MIT, he has led projects examining intensive care unit (ICU) transfer delays, emergency department overcrowding and inpatient surgical discharge processes. One of his projects using machine learning to predict surgical discharges has won grant awards from Partners Healthcare and Amazon Web Services. His work at the MGPO includes projects spanning the hospital and its various departments, including efforts to reduce surgical emergency department visits and readmissions post-discharge, building new methods to allocate beds for patients when the hospital has reached critical capacity levels and reducing the burden of prior authorization for Mass General’s numerous clinical practices. In the DACCPM, Dr. Safavi has led the development, implementation and study of a remote monitoring program to reduce failure-to-rescue events at the Mass Eye and Ear for post-operative patients.
Previously, he has served as the DACCPM representative to the Mass General Housestaff Quality and Safety Committee and was the chief quality fellow for the surgical ICU. Dr. Safavi was an inaugural Partners Connect Health Innovation Fellow. He was the innovator in residence for the Mass General Healthcare Transformation Lab. Dr. Safavi is co-founder and chief medical officer of Position Health, a digital health company that seeks to reduce hospital readmissions. He previously received his undergraduate, masters and master of business administration degrees from Yale University. Dr. Safavi completed his residency in anesthesia and critical care fellowship at Mass General.
Cecilia Zenteno is an operations research manager in the Perioperative Services Department at Mass General. She is an integral member of the Mass General-MIT collaboration, a partnership between the hospital and the MIT Sloan School of Management, whose aim is to apply operations research techniques to redesign and improve care processes in large academic medical centers. Zenteno’s responsibilities include data analysis, mentoring the collaboration’s students and fellows and executing initiatives together with all levels of health care providers and administrators. Zenteno received her PhD in operations research from Columbia University and worked as post-doctoral fellow for the Mass General-MIT collaboration for two years before joining Mass General full time in 2014.
Martin Copenhaver is an operations research scientist at Mass General and a lecturer in operations, research and statistics at MIT Sloan School of Management. His research and teaching interests lie in the intersection of optimization and statistics, especially with applications in health care operations management. Copenhaver completed his PhD in operations research at MIT's Operations Research Center (advised by Dimitris Bertsimas) and his bachelor of science in applied mathematics at Georgia Tech.
Mark Seelen is the program manager of Procedural Services and HSE at Mass General. His responsibilities include management of different strategic, operational and financial programs and projects across the hospital and Partners HealthCare. Prior to Mass General, he served as an infantry officer in the United States Army, leading platoons in Iraq and Afghanistan and serving as an instructor at the United States Army Ranger School. Seelen received his bachelor of science from the United States Military Academy at West Point, his master of public health from the Geisel School of Medicine at Dartmouth and his master of business administration from the Tuck School of Business at Dartmouth.
In 2008, Mass General was facing a major problem in its perioperative care environment. The Post-Anesthesia Care Unit (PACU) was often at capacity, forcing patients to wait in the operating room after emerging from anesthesia. This was difficult for patients and prevented surgical teams from moving on to new cases. The problem was particularly severe in the middle of the day on Wednesdays and Thursdays. The congestion was traced back to significant variability in the inpatient-bed occupancy rates over the course of the week: in midweek, patients who were ready to leave the PACU could not do so because there were no beds available on surgical wards, thus patients could not leave the OR to move to the PACU.
To solve this problem, Dr. Dunn and Dr. Levi and their colleagues identified key variables and constructed a large-scale optimization model in close collaboration with the surgical services. The objective was to increase the effective capacity of the surgical wards by smoothing the average bed occupancy throughout the week. Ultimately, about 30 percent of surgeons changed their elective surgery schedules to increase the number of procedures on days when the perioperative facilities had been underutilized (Mondays and Fridays).
Dr. Dunn notes that implementing a new OR schedule is particularly challenging at an academic medical center due to the many other commitments surgeons have beyond the delivery of care in the operating room. At the same time, bed access is such a critical bottleneck at Mass General that change was imperative. Successful implementation of this complex strategy hinged on extensive communication with the surgical services, the use of a data-driven model to accurately explore costs and changing constraints within the system, and the unwavering support of high-level hospital leadership.
Another challenge associated with patient flow in the ORs at Mass General was that approximately 30 percent of patients who needed non-elective surgery were waiting more than the prescribed amount of time (ideal maximum wait times: 24 hours for non-urgent cases, 4 hours for urgent cases, and 45 minutes for emergent cases). This hampered access to care and aggravated the lack of inpatient beds.
After careful analysis, the team of Drs. Dunn and Levi designed and implemented an “open block” strategy; operating room blocks were reserved for nonelective patients during regular working hours (prime time) and management of those blocks was centralized rather than allotted to individual surgical services as in the past. Comparison of metrics before and after implementation of the new strategy showed that the average preoperative wait time for all nonelective surgical patients decreased by 25.5 percent, even with an overall volume increase of nine percent (see Figure). In addition, the number of bed-days occupied by nonurgent patients before surgery declined by 13 percent (saving over 73 bed-days before surgery once the new strategy was implemented).
Internal Medicine Associates (IMA) is a hospital-based primary care practice at Mass General dedicated to the forward movement of health care through research, education and innovation. With 48 staff physicians, 70 primary care and other residents, five nurse practitioners, more than 20 nurses and a full administrative and support staff, it is the largest and most complex Mass General teaching practice.
The Mass General-MIT collaboration set out to help the IMA achieve two key goals emphasized in the 2014 Mass General Strategic Plan: provide exceptional patient-centered care to its primary care patients—featuring a trusted team of providers who communicate and work together effectively—and measure and continually improve care delivery.
The collaboration’s first IMA project has focused on increasing the efficiency, quality and safety of the IMA’s prescription (Rx) management process through centralization. Figure 4 illustrates the effort to reduce variability in the prescription renewal process by 50 percent, reduce time required to process prescriptions from 4.1 to 2.5 FTE’s, eliminate duplication and increase the time health care providers can devote to other activities. This effort also inspired the larger scale work adopted by the MGPO through IMMERSE.
On September 18, 2014, two weeks after launching the centralized prescription refill process, Daniel Horn, MD, of the IMA sent a note to Dr. Dunn and other members of the operations research and management team that read in part:
"We just finished our pod 1 meeting where the bulk of the discussion was spent debriefing on how the first two weeks of the prescription refill module has gone. I have to say I have not experienced that much enthusiasm in a meeting in the IMA since I started here. Nurses were singing the praises of the program, front desk staff were actively engaged in problem solving about prescription related phone calls … This is a change that is going to work and is good for both patients and everyone on the team, and I want to thank you all for championing, supporting, and doing the work" - Dr. Horn
Dr. Dunn notes that the IMA primary care physicians—like the surgeons who agreed to changes in operating room schedules—have had a positive experience with their first exposure to operations research and management and thus will be open to more data-driven process improvements in the future.
Hospital bed assignment is a challenging task: there are random patient arrivals from multiple sources, patients must be matched to a bed that fits their clinical needs, and there is high supply uncertainty: it is not clear when beds will become available as patients are discharged throughout the day. Furthermore, Mass General consistently operates at 90% of its total bed capacity. As bed managers are left with little room for maneuver, high occupancy levels often result in severe congestion in upstream units such as the Emergency Department (ED) and the Post Anesthesia Care Unit (PACU).
Upon mapping the bed assignment process, the Mass General-MIT collaboration noted that Mass General had a very conservative bed assignment policy: in an effort to be proactive, all patients were assigned as early as possible, often even before they underwent surgery. This frequently resulted in beds that would remain empty for hours “waiting for patients”, and vice versa.
To make the bed assignment process more efficient, the collaboration used discrete event simulation to model and evaluate online ‘just-in-time’ bed assignment policy. Its underlying principle: to assign ready patients to ready (empty) beds. After working with the floor, ED, PACU staff and the bed managers, the collaboration successfully implemented this policy across all the surgical and neuroscience units.
The impact of the implementation has been significant. More than six months after the last roll-out wave, the average patient wait time to surgical units from the ED decreased by over 20% and by almost 40% from the PACU (Figure). Most notably, there have been significant reductions in the upper percentiles of the wait time distribution.
As hospitals continue to experience high demand for their services, efficient capacity management is critical to fulfilling their mission to serve patients. Real-time assignment of patients to beds has significant benefits, particularly in highly utilized systems.
Many large outpatient infusion centers in the U.S. have distinct periods of overcrowding but remains widely underutilized otherwise. The Mass General Cancer Center was not the exception. The uneven use of resources results in highly strained staff and physical resources during congested hours, causing delays, staff and patient dissatisfaction, a higher risk for quality and safety problems and, most pervasively, the perception of insufficient capacity.
After extensive data analysis, shadowing and detailed process mapping, the Mass General-MIT collaboration determined that a major root cause for the mid-day congestion, was the use of sub-optimal scheduling practices. To address this problem, HSE developed, in collaboration with the Cancer Center, a real-time scheduling algorithm that provides patients with appointment start times that balance resource utilization throughout the day as visits are scheduled.
The scheduling algorithm was constructed by deriving insights from running an offline retrospective integer program repeatedly and validated using prospective simulation modeling. Further analysis showed that the proposed state could be achieved with minimal adjustments to staffing in the infusion unit and pharmacy. The algorithm also respects the existing primary nursing model in which patients are matched with the same infusion nurse for treatment as much as possible and considers treatment-specific schedule limitations.
To implement the algorithm, the collaboration partnered with an internal Mass General software-developing group that created and deployed a tailored web-based application. The program, which we call Opt-In, interfaces with the existing medical record and scheduling systems to extract all the patient’s and treatment information.
Since its roll-out in April 2017, the average peak occupancy of the infusion unit has decreased by three chairs, while the average occupancy in the evening hours has increased by as much as four chairs (Figure).
At the same time, the number average patients seen per day increased by five, the number of average scheduled chair hours increased by a similar amount and, most notably, the average number of actual utilized infusion chair hours per day decreased by 10 chairs. This indicates that the infusion unit is now able to see more patients, while providing more efficient care due to the reduced congestion in peak hours, without having to make any changes in its existent capacity.
The use of highly effective infused specialty drugs has increased significantly over the past two decades as they have led to dramatic improvements in patients’ quality-of-life. This transformation introduced a variety of operational challenges to the 10 clinics that started administering these treatments at Mass General. As they began to grow infusion capacity organically, most of them began experiencing a supply and demand imbalance. Furthermore, many of the infused medications are being used in a manner not yet specified by the FDA’s approved packaging label or insert, creating the need for obtaining prior authorizations (PAs) from the insurance companies to guarantee reimbursement. This approval may take anywhere from hours to several weeks, which adds a considerable layer of complexity to the scheduling process and the efficient utilization of the infusion chairs. Additional challenges include keeping adequate staffing levels for fluctuating patient volumes and the financing of these drugs, which account for a sizable proportion of the growth in the pharmacy spend.
In sum, patient access to these critical treatments was severely curtailed mainly due to high infusion chair utilization, poor scheduling and administrative practices and inadequate staffing. At the same time, further investigation revealed that these clinics maintained vastly similar operating procedures, yet in separate physical environments.
The Mass General-MIT collaboration proposed the creation of a Medical Infusion Center (MIC) that would centralize all the administrative tasks such as: scheduling, PA processing, and nursing staff management (Figure). The center would be divided in two locations to accommodate an increasing patient population. Using an extension of the infusion scheduling algorithm above, the center would be able to use its capacity as effectively as possible. Thus, the center would increase access to care, offer emergent appointments, generate financial savings from efficient handling of PAs and, most importantly, avoid unnecessary hospital infusion admissions. The MIC opened its doors in October of 2017 and is already serving an average of 44 patients per day in a 15-chair unit in the Mass General Main Campus.
Schoenmeyr T, Dunn PF, Gamarnik D, Levi R, Berger DL, Daily BJ, Levine WC, Sandberg WS. A model for understanding the impacts of demand and capacity on waiting time to enter a congested recovery room. Anesthesiology. 2009 Jun; 110(6): 1293-304. (link)
Seger RF, Dunn PF, Prestipino AL, McDougal WS. Multidisciplinary team streamlines hospital schedules, patient care. MGMA Connex. 2010 Aug; 10(7): 46-9. (link)
Segev D, Levi R, Dunn PF, Sandberg W. Modeling the Impact of Changing Patient Transportation System on Perioperative Process Performance in a Large Hospital: Insights from a Computer Simulation Study. Healthcare Management. 2012; 15 (2): 155-169. (link)
Zenteno AC, Carnes T, Levi R, Daily BJ, Price D, Moss SC, Dunn PF. Pooled Open Blocks Shorten Wait Times for Non-Elective Surgical Cases. Annals of Surgery. 2015; 262(1): 60-67. (link)
Zenteno AC, Carnes T, Levi R, Daily BJ, Dunn PF. Systematic OR Block Allocation in Large Academic Medical Centers. Annals of Surgery. 2016; 264(6): 973-981 (link)
Ghobadhi K, Zenteno AC, Marshall AR, Dunn PF, Levi R, Stone JH. Translating a Biologic Revolution into an Organizational Overhaul. NEJM Catalyst. 2017. ePub. (link)
Safavi KC, Furtado J, Zenteno AC, Scheinker D, Schmidt U, Levi R, Dunn PF. Non-clinical delays in transfer out of the surgical ICU are associated with increased hospital length of stay and delayed progress of care. Journal of Critical Care. 2018; 50: 126-31. (link)
Healthcare Systems EngineeringMassachusetts General Hospital, White 4-40055 Fruit St.Boston MA, 02114
Back to Top