The Weil Institute: Leveraging Precision Medicine to Change the Face of Healthcare
The following was submitted for publication by April Kriebel, a Ph.D. candidate in the Department of Computational Medicine & Bioinformatics. She works in the lab of Dr. Joshua Welch studying the integration of single cell molecular data with anatomical and physiological response data. April wrote this article in collaboration with Michigan Science Writers, an organization at the University of Michigan that trains graduate students and post-doctoral fellows in science writing and editing.
How is data used in healthcare?
With the increasing attention and excitement given to modern technology and medical advancements, it's somehow disconcerting to go in for a yearly physical and find that most areas of medicine appear to be untouched by the rapid progress we hear about on the news. Yet, unbeknownst to us at the patient level, the same data and information that doctors have always collected is now being used and understood in new and refreshing ways. The end goal is personalized medicine, or precision medicine, the idea that medicine and treatment plans are tailored to a specific individual.
The vision of precision medicine is to assess all relevant attributes of a patient while treating their disease. Ideally, genetic, biological, environmental, and lifestyle factors would all be considered when evaluating and caring for a patient. [1] Dr. Can Ince, Director of the Department of Translational Physiology at the University of Amsterdam, details this wave of data as an “overwhelming amount of patient data, including historical clinical information as well as continuous online data regarding the condition of the patient, which changes from moment to moment.” [2] The idea is to use all of those small bits of information about a patient to help doctors make more informed decisions about treatment plans and patient recommendations, allowing health practitioners to use algorithms to help understand each patient’s story, and discover novel insights into a patient’s health profile. To uncover key insights for each individual, researchers in precision medicine compile many patients’ data together, and look to classify patients into groups or uncover general trends within the data based on specific characteristics of a subset of the patient group. By looking at general trends across groups and subgroups, clinicians are able to offer more accurate diagnosis, treatment, & patient monitoring for patients within those groups. Precision health is a research priority because as the models improve and their usage expands, the more positive impact will be felt by patients. From a patient perspective, the changes in healthcare can seem slow, but, behind the scenes, there is a large investment of time and resources devoted to expanding the way precision health is used, and the ways it can benefit the patient population.
Sepsis is a key example of where big data has been used for more accurate diagnosis and treatment. Sepsis could be considered a complication of an infection, where widespread inflammation and immune response begin damaging organs and tissues. [3] While annually 1.7 million adults in America will develop sepsis, it has previously been treated as a single disease. [4] Research done at the University of Pittsburgh used the electronic health records of over 20,000 patients to identify four distinct sepsis types. [5] This stratification of the patient population is critical, because these different groups of sepsis patients responded differently to treatments. For example, a treatment previously established as ineffective against sepsis was shown to be beneficial for one subgroup of sepsis patients, and harmful to another, underscoring the importance of using big data to stratify patient groups [5].
Disease prognosis is another important area of research in precision medicine. Traumatic brain injury (TBI), a contributing factor to 30% of injury-related deaths in the United States, is an area of medicine where disease prognosis is difficult to estimate. [6] In fact, while most TBI patients take 2 weeks to awake from their comas, most TBI patients are withdrawn from life support within 72 hours of hospital admission. [7] To address the need for more accurate patient prognosis, researchers used data from a patient’s clinical and head CT data, to predict a patient’s outcome 6 months after sustaining a severe TBI more accurately than 3 neurosurgeons. [8]
How is the University of Michigan progressing precision medicine?
Researchers at the University of Michigan Max Harry Weil Institute for Critical Care Research & Innovation have explored the use of data in patient monitoring in the context of clinical care and have produced PICTURE (Predicting Intensive Care Transfers and other UnfoReseen Events). [9] PICTURE works behind the scenes, focusing on using already available data in a patient’s Electronic Health Record or data collected during clinical care to predict whether a patient is deteriorating. On average, PICTURE is able to detect a deterioration event, such as ICU transfer or death, 6 -12 hours prior to the event taking place. [10] It then notifies the clinical care team, allowing ample time for preventative adjustments to the standard of care. The goal of PICTURE is to aid clinicians in identifying patients who are not obviously trending toward a rapid decline, alerting physicians and caretakers that this patient is in need of more medical intervention than they may have suspected. The PICTURE algorithm has proven extensible, as it has successfully been used in multiple hospital systems with consistent results, and is being extended to include pediatric populations. [11]
What are the current challenges of data science?
Using data to make healthcare decisions is not a straightforward task, even when the data already exists as part of a patient’s record. There are two primary obstacles to creating effective models: useability and extensibility.
Although data on patients is collected fairly autonomously in most hospital systems, there are significant differences in how each hospital system might collect and store their data. Brandon Cummings, a data scientist at the Weil Institute, describes how even something as simple as blood pressure readings can vary based on whether they are taken from a monitor, a cuff, or an atrial line. Additionally, he explained how even the lack of certain data, such a lab test that wasn’t ordered, can have significance. “In medicine,” he explains, “missingness is defined by when the doctors order labs, or when the nurses take a vital sign…. if you go to a hospital with different care patterns, and the model relies on those patterns to make decisions….that changes your model.” Issues can also arise if the model is trained on patient populations that lack diversity. As a result, scientists and clinicians have to carefully validate their model, and confirm that their results are consistent and reproducible across hospital systems and demographic populations.
Useability is another key concern when creating algorithms for use in health care. Dr. Kevin Ward, Director of the Weil Institute, explains that, “the last thing healthcare workers want is more alarms, especially if those alarms may be wrong.” As a result, its key that solutions are built with clinicians in mind. To accomplish this, there is increased emphasis on involving clinicians in developing and implementing new algorithms and models. At the Weil Institute, clinicians are often those that first identify a problem, and then bring it to a team of data scientists to help brainstorm a solution. They are intimately involved with the development process, ensuring that the final product is something that clinicians themselves have identified as beneficial.
Who pays for data science in health care?
Data science has incredible potential to help support clinicians and improve patient care, but the process of developing, implementing, and curating such models and diagnostics amounts to substantial costs. While the traditional funding source for such projects has usually come from a federal source, the percentage of funding for research and development activities at universities is changing. In 2010, 61.1% of funding for research and development activities at the university level came from federal sources, and 5.2% came from industry sources. In 2020, only 53.5% came from federal sources, and 6.0% came from industry. In total, industry investment has jumped from 3.31 billion 2022 dollars in 2010 to 5.38 billion 2022 dollars in 2020 [12]. Private funding tends to generate more applied research [13], where “conversations often lead directly to the creation of a statement or work or project plan, including an associated budget, specifically developed to solve the problem at hand [14].” Gloria Waters, vice president and associate provost for research at Boston University describes industry funding as “focused on applied research that will result in the development of products with immediate commercial application [15].” Certainly, the goal of precision medicine is to apply tools and models that are developed within a healthcare setting, and this aligns well with the efficiency and product-driven mindset that often characterizes industry settings.
The massive amounts of data being generated and stored for each patient in the healthcare system is creating a wealth of information that can be leveraged to improve patient care, guide clinician decision making, and support researchers as they explore the physiology of disease.
“Healthcare in 20 years is going to look very different,” Cummings remarks, “we need to monitor our models and make sure that they continue to perform as healthcare changes.” With perseverance and strategy, we can continue to utilize data to our advantage, issuing in an era of medicine with better clinician support, more accurate diagnosis, and a refined approach to personalized medicine.
About the Weil Institute, formerly MCIRCC
The team at the Max Harry Weil Institute for Critical Care Research and Innovation (formerly the Michigan Center for Integrative Research in Critical Care) is dedicated to pushing the leading edge of research to develop new technologies and novel therapies for the most critically ill and injured patients. Through a unique formula of innovation, integration and entrepreneurship that was first imagined by Weil, their multi-disciplinary teams of health providers, basic scientists, engineers, data scientists, commercialization coaches, donors and industry partners are taking a boundless approach to re-imagining every aspect of critical care medicine. For more information, visit weilinstitute.med.umich.edu.
Sources
Precision medicine readiness principles resource guide: Care integration. World Economic Forum. (n.d.).
https://www.weforum.org/whitepapers/precision-medicine-readiness-principles-resource-guide-care-integrationInce, C. (2017). Personalized physiological medicine. Critical Care, 21(S3).
https://doi.org/10.1186/s13054-017-1907-7NHS Inform. (2019). Sepsis. Nhsinform.scot.
https://www.nhsinform.scot/illnesses-and-conditions/blood-and-lymph/sepsisCenters for Disease Control and Prevention. (2023, August 24). What Is Sepsis? Centers for Disease Control and Prevention; CDC.
https://www.cdc.gov/sepsis/what-is-sepsis.htmlBig Data Reveals Hidden Subtypes of Sepsis. (n.d.). UPMC | Life Changing Medicine.
https://www.upmc.com/media/news/051919-seymour-senecaPICTURE predictive analytic tool proves adaptable and reliable in hospital outside of U-M. (2023, March 24). The Max Harry Weil Institute for Critical Care Research and Innovation. https://weilinstitute.med.umich.edu/latest-news/picture-hurley
Predicting patient deterioration with powerful analytics in a hospital setting. (2022, September 1). The Max Harry Weil Institute for Critical Care Research and Innovation.
https://weilinstitute.med.umich.edu/product-portfolio/picture-suite-of-predictive-analyticsA machine learning analytic that predicts pediatric general floor deterioration. (2022, September 15). The Max Harry Weil Institute for Critical Care Research and Innovation.
https://weilinstitute.med.umich.edu/product-portfolio/picture-suite-of-predictive-analytics-nwkatBrainLine. (2018, August 9). Get the Stats on Traumatic Brain Injury in the United States | BrainLine. BrainLine.
https://www.brainline.org/article/get-stats-traumatic-brain-injury-united-statesMachine Learning Model Can Steer Traumatic Brain Injury Patients to Life-Saving Care. UPMC | Life Changing Medicine.
https://www.upmc.com/media/news/042522machine-learning-modelPease, M., Dooman Arefan, Barber, J., Yuh, E. L., Puccio, A. M., Hochberger, K., Nwachuku, E. L., Roy, S., Casillo, S. M., Temkin, N. R., Okonkwo, D. O., Wu, S., Neeraj Badjatia, Bodien, Y. G., Duhaime, A.-C., V. Ramana Feeser, Ferguson, A. R., Foreman, B., Gardner, R. C., & Gopinath, S. P. (2022). Outcome Prediction in Patients with Severe Traumatic Brain Injury Using Deep Learning from Head CT Scans. Radiology, 304(2), 385–394.
https://pubs.rsna.org/doi/10.1148/radiol.212181American Association for the Advancement of Science. (2018). Historical Trends in Federal R&D. American Association for the Advancement of Science.
https://www.aaas.org/programs/r-d-budget-and-policy/historical-trends-federal-rdBabina, T., He, A. X., Howell, S. T., Perlman, E. R., & Staudt, J. (2020, December 1). The Color of Money: Federal vs. Industry Funding of University Research. National Bureau of Economic Research.
https://doi.org/10.3386/w28160How Is Industry Sponsored Research Different from Government or Foundation Sponsored Research? | Industry Engagement.
https://www.bu.edu/industry/2021/08/31/what-makes-industry-sponsored-research-different-from-other-sponsored-researchThe History and Future of Funding for Scientific Research | The Brink. (2015, April 6). Boston University.
https://www.bu.edu/articles/2015/funding-for-scientific-research/#:~:text=Three%2DWay%20Split,labs%20operated%20by%20private%20contractors.