Sharpening healthcare for the next wave

The way the novel coronavirus tore through the U.S. healthcare system and threatened to overwhelm it in places has brought into sharp relief the need to apply data-driven analytics and optimization in order to hone healthcare practices and service delivery for the next wave of challenges. Healthcare in the U.S. is characterized by innumerable variables that make it difficult to fully understand, much less improve, the system. That’s where data and modeling come in.

Our Biomedical Analytics and Systems Optimization (BASO) research lab is developing healthcare analytics models, techniques, and tools, which are critical to smart and connected health. These innovations can be used to address problems in healthcare service, clinical practice and public health via technology-based solutions and community-engaged deployment.

Smart, connected health aims to develop groundbreaking approaches to help transform healthcare from reactive and hospital-centered to proactive, preventive, evidence-based and person-centered — that is, focused on well-being, rather than disease. These inventive technologies will provide the next-generation solutions and breakthrough ideas in healthcare.

Medical drone base location map for opioid overdose rescue with two different budget levels. Under limited budget for base construction and drone procurement, Purdue’s Biomedical Analytics and Systems Optimization research lab has developed a facility location-and-sizing joint optimization model that minimizes mean response time while ensuring a specified level of care access.

In the past, there’s been a reluctance to tackle healthcare renovation because it is so complex. The key to success is developing computational modeling and optimization techniques for complex, uncertain, fragmented systems like those in healthcare, for which we lack the in-depth, detailed knowledge and the accurate, ample measurements needed for data-driven modeling. In the absence of confidence in the data, we model and optimize probabilities for various outcomes, assigning a value to each probability to reflect its likelihood of occurring.

Healthcare systems analytics is an emerging area that holds great potential to harness big data for advances in personalized, precision healthcare. It includes descriptive, predictive and prescriptive analytics to sift through huge, often poorly structured datasets to discover complex patterns, and then use those patterns to forecast and manage future events.

Our research team focuses on healthcare operations management and clinical decision-making, biomedicine, and healthcare analytics. These areas address such needs as trauma/emergency care, infectious disease control, rehabilitative/long-term care, cancer screening, the opioid crisis, and mental health.

For example, care network design and capacity management drive operational decisions involving staff scheduling, referral coordination, and emergency logistics — all crucial during the COVID-19 surge. With these approaches, we’ve uncovered insights to enable care organizations to streamline patient flow and manage staff workload more efficiently for the inpatient discharge process. We also have developed novel stochastic (probabilistic) programming models to reduce preventable hospital readmissions.

Additionally, we’ve investigated system design and operations management challenges around access to care in care organizations with a tiered-care delivery system — central hospitals and many more satellite clinics — in which there is a geographic mismatch between care need and timely provision.

Moreover, an important area ripe for innovation is staffing planning at nursing homes and other long-term care facilities. For these environments, we’ve developed novel stochastic programming models and methods for analyzing and optimizing staffing decisions that balance trade-offs between a resident-centered approach, the employee experience, and operating cost. Another area requiring an upgrade is network design for emergency care. To address this issue, our team has created novel bi-level integer programming models and methods to simultaneously boost health and well-being and hospital network profitability.

Our smart, connected healthcare engineering projects are well-placed at the interface of computational modeling, data science, and clinical engineering, with equal emphasis on basic model/methodology development, research translation, and community engagement. Lab members come from disciplines including biomedical, industrial and chemical engineering, as well as computer science, mathematics, and statistics. We actively collaborate with researchers from medicine, nursing, pharmacy and management.

It’s all hands on deck to find ways to transform healthcare practices and delivery so we can be better prepared for the next wave, whatever it is.

Overview of BASO lab’s research portfolio. We undertake research endeavors in the following three areas interfacing with biomedicine and healthcare: 1) delivery operations engineering (the area highlighted in this article); 2) big data analytics; and 3) system dynamics modeling. In the first area, we focus on healthcare capacity management. In the second area, the main data source is claims data, supplemented by ubiquitous sensing and clinical assessment. The clinical areas include aging, mental health, and opioid use disorder. In the third area, our aim is to advance semi-mechanistic modeling techniques for complex dynamic systems, especially for surrogate modeling and model calibration. The main clinical area is colorectal cancer and infectious diseases. With these research endeavors, we have developed computational models, solution algorithms, and analysis techniques to address critical but unmet needs in clinical practice, healthcare delivery, and public health, as well as biomedical instrumentation.

Nan Kong, PhD

Associate Professor, Weldon School of Biomedical Engineering

College of Engineering, Purdue University

Related Links

Professor Nan Kong, teamed up with Indiana University School of Medicine researchers and Eskenazi Health administrators, to forecast critical care use during COVID-19 pandemic

Professor Kong on optimizing the regional trauma network via bi-level integer programming, currently funded by National Science Foundation (NSF)

Professor Kong receives NSF funding for research to optimize the nursing facility staff schedule via Bayesian stochastic programming

A team of Purdue BME faculty, including Professor Kong, plans to engineer critical wearable technology and data analytics solutions to combat the opioid epidemic

Professor Kong’s work on inpatient discharge strategy assessment

Professor Kong’s two current PhD students became two of the five John W. Anderson Foundation Endowment winners in 2020, given by Purdue’s Regenstrief Center for Healthcare Engineering

Professor Kong’s former student Dr. Yan Li has exemplified the power of applying computational modeling to health policy research