Skip to main content
Sign In
 

Seminar Series



Revisiting Propensity Score Modelling Assumptions
Debashis Ghosh, Professor and Chair
Department of Biostatistics and Informatics

Tuesday, May 21 | 12:00 - 1:00 PM
Education 2 North, Room 1201


Debashis Ghosh is currently Professor and Chair of the Department of Biostatistics and Informatics at the Colorado School of Public Health. He previously worked at the Department of Biostatistics at the University of Michigan and in the Department of Statistics at Penn State University. Dr. Ghosh has received a variety of honors and recognition throughout his career, including the 2013 Mortimer Spiegelman Award for outstanding early career statistical research in public health and the 2015 Myrto Lefkopolou Distinguished Lecturer from the Harvard School of Public Health. He currently co-directs the Data Science to Patient Value (D2V) Analytics Core with Dr. Tell Bennett.


Lundberg headshot.jpgExplainable AI for Science and Medicine
Scott Lundberg, PhD Candidate and NSF Fellow
University of Washington

Thursday, April 25 | 11:00 AM - 12:00 PM
Education 2 South, Room 2201


Scott Lundberg is a PhD candidate in the Paul G. Allen School of Computer Science & Engineering at the University of Washington working with Su-In Lee. His research focuses on explainable artificial intelligence and its application to problems in medicine and healthcare. This has led to the development of broadly applicable methods and tools for interpreting complex machine learning models.


Caverly_0.jpgLightening the Load: Personalizing Substantive, Everyday Decisions in Primary Care
Tanner Caverly, Assistant Professor of Learning Health Sciences
University of Michigan

Tuesday, March 19 | 12:00 - 1:00 PM
Education 2 North, Room 1202


Presented in conjunction with the ACCORDS Patient-Centered Decisions Seminar Series

Attendees of this talk will be able to:

1. Give examples of substantive, high-volume decisions in primary care.
2. Explain when a medical service might be considered more vs. less preference-sensitive.
3. Describe the elements of a targeted, brief shared decision making approach.


Why D&I?: Using Dissemination and Implementation Methodologies to Advance the Field of Data Science and Health Care Delivery Research
Bethany Kwan, Assistant Professor, Family Medicine
Jenna Reno, Dissemination Research Instructor, Family Medicine

Tuesday, February 19 | 12:00 - 1:00 PM
Education 2 North, Room 2301

Despite the demonstrable benefits of many new healthcare and data science discoveries, the field as a whole has been slow to put research findings into practice. The challenge of moving health research innovations from discovery to practice is complex and multifaceted. Dissemination and implementation (D&I) methodologies provide a promising solution for addressing this translation challenge. In this presentation, Drs. Bethany Kwan and Jenna Reno will share how D2V is using D&I principles to advance the process of transforming data science to patient value.  ​


Identifying, Assessing, and Balancing Competing Risks of Multiple Hospital-Acquired Conditions (HACs)
Mark Gritz, Associate Professor
Division of Health Care Policy & Research

Tuesday, January 15 | 12:00 - 1:00 PM
Education 2 North, Room 1304

Hospital-acquired conditions (HACs) are common, costly, and a national patient safety priority. In 2010, there were over 5.9 million HACs with an estimate of over $62.8 billion in incremental hospital costs. Preliminary data suggests there may be a divergence in progress on CAUTIs and HAPUs compared to falls. It is possible that hospitals are focused on interventions that reduce a single HAC are not taking into account unintended consequences and trade-offs that may increase the risk of other HACs. To date, there have been no quantitative assessments of multiple HAC competing risks and trade-offs. Our objectives are to develop an empirical framework to identify, assess and predict competing risks of multiple HACs, assemble a multisite database to implement this empirical framework, develop a Multiple HAC Competing Risk Dashboard to inform clinicians, administrators and patients decisions related to indwelling urinary catheter (IUC) use, and pilot test this dashboard.

hoch headshot.jpgBig Data for Childhood Respiratory Disease: Pitfalls and Promise
Heather Hoch, Assistant Professor, Pediatric Pulmonary Medicine
Children's Hospital Colorado

Tuesday, November 20 | 12:00 - 1:00 PM
Education 2 North, Room 1304​


Heather Hoch is an Assistant Professor in Pediatric Pulmonary Medicine at Children’s Hospital Colorado.  She graduated from Washington University in St. Louis with an undergraduate degree in Biology and Anthropology (with a focus on Medical Anthropology) and completed her medical training, masters in clinical science, residency and fellowship at the University of Colorado, Anschutz Medical Campus.  Dr. Hoch is also trained in health services research, with a focus on adherence to asthma medications, and she has a special interest in using big data to evaluate outcomes for children with asthma.  She has completed big data evaluations to evaluate outcomes in children with cystic fibrosis across the US, and for children with asthma in Colorado.


Thinking Outside the Check-Box: Culturally Effective Research, A Framework to Achieve Better Papers and Obtain Bigger Grants
Paritosh Kaul, Professor
Department of Pediatrics

Tuesday, October 16 | 12:00 - 1:00 PM
Education 2 North, Room 1304

By 2035, minorities are expected to make up over 40% of the US population. Cultural competency is important now more than ever, and is an essential component of providing effective healthcare2. While many educational programs for health professionals now include a cultural competency component in their curriculum, the strategies for translating cultural competency techniques into practice lack evidence. Cultural competency is necessary to addressing the underrepresentation of minorities as participants in research in addition to ensuring that such groups are beneficiaries of the results. To reduce health disparities for the growing population of minority groups, research focused on developing effective and evidence-based strategies for teaching cultural competency to health professionals and implementing cultural competency techniques in healthcare settings is essential. This seminar will provide an introduction to culturally effective medicine, its practices, and its incorporation into medical research.


High- and Low-Value Cardiovascular Care: An Analysis of Colorado All-Payer Administrative Claims
Vinay Kini, Assistant Professor
Division of Cardiology

Tuesday, September 18 | 12:00 - 1:00 PM
Education 2 North, Room 1304​

Improving the value of cardiovascular care has proved challenging. Cardiovascular diagnostic tests such as cardiac stress tests and echocardiography have received particular scrutiny owing to their widespread use, concerns about overuse, and their expense (these tests account for over 40% of Medicare Part B spending on medical imaging). Professional societies have developed guidelines such as Choosing Wisely that explicitly define clinical situations when testing is considered high-value (improves patient outcomes) and low-value (provides no benefit to patients), but little is known about whether these efforts have been successful.

In this project, we use a large all-payer administrative claims database from the state of Colorado to examine hospital-level variation and correlation in rates of use of two high-value and two low-value cardiovascular tests. We gain insights into the degree of variation in high- and low-value testing, factors that influence healthcare value, and begin to understand whether efforts to improve value in diagnostic testing have been successful.


A Policy Analysis of Colorado Medicaid Primary Care Payment Bump
Martha Meyer, Postdoctoral Fellow
Data Science to Patient Value

Tuesday, August 21 | 12:00 - 1:00 PM
Education 2 North, Room 1304


The Accountable Care Act expanded Medicaid eligibility to most low income adults and an additional 7 million were anticipated to be eligible.  Access to primary care has long been considered a challenge in Medicaid and low provider payment rates are often given as a reason. Section 1202 of the Health Care and Education Reconciliation Act of 2010 required all States to increase Medicaid payments for eligible primary care visits to at least the same level as Medicare for two years (2013-2014) in an effort to increase provider participation in Medicaid.
  
Colorado was one of 16 states plus Washington DC that authorized the continuation of the primary care payment bump for Medicaid.  The first continuation was authorized from January 2015-June 2016 at the Section 1202 rate and removed some implementation barriers.  The second rate continuation was authorized for July 2016-June 2017 with a reduced payment amount from 100% to 85% of Medicare rates and limited the number of primary care services with the higher payment.  Prior evidence of the impact of the Section 1202 bump is mixed; however, some believe the impact was mitigated by the temporary nature of the initial payment bump. Therefore, Colorado’s payment continuation offers us the opportunity to examine if ongoing changes in primary care payment rates is an effective policy tool to influence provider behavior.  


Wasfy_Jason.jpegAdvancing Value in Cardiology Care
Jason Wasfy, Medical Director for Population Health Management
Massachussetts General Hospital

Tuesday, July 17 | 12:00 - 1:00 PM
Education 2 North, Room 1304
Video
The growth of alternative payment models have changed incentives for health care providers.  Given that cardiology care nationally is both very common and expensive, cardiologists have had to generate substantial new knowledge about how to deliver higher value care.  Here we review how both programs and policies have demonstrated changes in both quality and cost for patients with heart disease and outline the challenges as payment models continue to evolve.​


Project Management: It's a Thing
Kelley Burns, Program Manager
Data Science to Patient Value

​Wednesday, June 20 | 12:00 - 1:00 PM
Education 2 North, Room 2302
Video
Kelley Burns, a certified Project Management Professional (PMP), will introduce a brief history of project management, current industry standards, and its application to academic projects.


graciela.jpgDigging Deeper into the EHR: The Case for NLP
Graciela Gonzalez-Hernandez, Associate Professor of Informatics
University of Pennsylvania

Tuesday, May 15 | 12:00 - 1:00 PM
Education 2 North, Room 1304​
Video
Dr. Gonzalez Hernandez is a recognized expert and leader in natural language processing (NLP) applied to bioinformatics, medical/clinical informatics, and public-health informatics. After 11 years at the Department of Biomedical Informatics at Arizona State University, she joined the University of Pennsylvania and established the Health Language Processing Lab within the Institute of Biomedical Informatics. Her recent work focuses on NLP applications for public-health monitoring and surveillance and is funded by R01 grants from the National Library of Medicine and the National Institute of Allergy and Infectious Diseases. 


Orr Headshot 2017.jpgAdventures in Clinical Data
Jeremy Orr, MD, MPH
PhD candidate, Department of Epidemiology

Tuesday, April 17, 2018 | 12:00 - 1:00 PM
Education 2 North, Room 1304

EMR data is now ubiquitous, but a variety of factors make it difficult to leverage for care improvement.  The Boston based analytics firm Humedica (now Optum Analytics) pioneered a method to clean and map EMR data on a massive scale.  This allowed them to provide health systems with trustworthy clinical intelligence and predictive models.   But even after having excellent analytics, many health systems failed to operationalize around it, and were unable to influence clinician behavior to improve quality and lower costs.  Stanson Health, spun out of Cedars-Sinai Medical Center in Los Angeles, developed a different approach to improving quality and cutting costs.   Stanson took evidence based clinical guidelines such as Choosing Wisely, put them directly into EMR workflow, and then tested and iterated on this process to maximize effectiveness and minimize provider fatigue.  

Both approaches teach us multiple lessons about how to utilize the EMR and EMR data for research, improving outcomes, refining care processes, and increasing the benefit to annoyance ratio for providers interacting with such systems.


Kittelson_Andy_308[1].jpgUsing Health Systems Data to Inform Clinical Monitoring: Case Example in Total Knee Replacement Rehabilitation
Andrew Kittelson, PT, DPT, PhD
RESTORE Group, Physical Therapy Program

Tuesday, March 20, 2018 | 12:00 - 1:00 PM
Education 2 North, Room 1304

The hope is that health systems data can be harnessed for a variety of useful applications (e.g., quality improvement projects, characterizing practice patterns, informing real-time healthcare decisions).  However, to be most valuable, these applications must be well-defined and informed by the right data. In this seminar, we present our work in Total Knee Replacement (TKR) rehabilitation.  Our goal is to use routinely collected outcomes data to improve clinical monitoring via novel informatics. We will highlight the data collection procedures and specific outcomes required, and speak to the partnerships needed to make such efforts successful.​


YangFan13_0782web.jpg
Using Post-Quality of Life Measurement Information in Censoring By Death Problems
Fan Yang, Assistant Professor
Department of Biostatistics and Informatics

Tuesday, February 20, 2018 | 12:00 - 1:00 PM
Education 2 North, Room 1304

Many clinical studies on non-mortality outcomes such as quality of life suffer from the problem that the non-mortality outcome can be censored by death, i.e. the non-mortality outcome cannot be measured if the subject dies before the time of measurement. To address the problem that this censoring by death is informative, it is of interest to consider the average effect of the treatment on the non-mortality outcome among subjects whose measurement would not be censored under either treatment or control, called the survivor average causal effect (SACE). The SACE is not point identified under usual assumptions but bounds can be constructed. The previous literature on bounding the SACE uses only the survival information before the measurement of the non-mortality outcome. However, survival information after the measurement of the non-mortality outcome could also be informative. For randomized trials, we propose a set of ranked average score assumptions that make use of survival information before and after the measurement of the non-mortality outcome which are plausibly satisfied in many studies and develop a two-step linear programming approach to obtain the closed form for bounds on the SACE under our assumptions. We also extend our method to randomized trials with noncompliance or observational studies with a valid IV to obtain bounds on the complier survivor average causal effect. We apply our method to a randomized trial of the effect of mechanical ventilation with lower tidal volume vs. traditional tidal volume for acute lung injury patients. Our bounds on the SACE are much shorter than the bounds obtained using only the survival information before the measurement of the non-mortality outcome.


Fatima-Paruk.jpgBreakthroughs with Neural Networks and AI: Predictive Modeling Comes Alive
Fatima Paruk, Chief Medical Officer
AllScripts Analytics

Tuesday, January 16, 2018 | 12:00 - 1:00 PM
Education 2 North, Room 1304

This session provides an overview of the Allscripts artificial intelligence (AI) initiative to leverage Microsoft Azure solutions for predictive models for patient cost and care. Analytics play a key role in the need for accurate risk stratification, predicting patients with rising costs, and enabling optimized care pathways for improved outcomes. Incorporating insights into actionable information is key to bringing value to existing data assets. We will highlight success stories and lessons learned in the race to deriving value from big data, and demonstrate how utilizing predictive models can drive business decisions and promote adherence to current, evidence-based guidelines in chronic disease management.  


Can Faculty Development Be Done Through Intelligent Tutors?
Janet Corral, Assistant Professor of Educational Informatics

Tuesday, November 21, 2017 | 12:00 - 1:00 PM
Education 2 North, Room 2302


Time-pressed faculty, fellows and residents are required to teach, though often do not have time to invest in multi-hour faculty development events integral to honing their skills. Innovative methods for providing evidence-based coaching to instructors are needed. Multiple interventions have been trialed over the last 5 years, the most recent of which is a just-in-time intelligent tutor system that delivers content in the time of need. This complex project funded by the Department of Medicine PACE grant weaves together multiple disparate data sources: evaluation ratings of faculty; faculty self-assessment; evidence-based tips for teaching; student feedback; and, faculty and learner clinical schedules. This D2V session will present a works-in-progress update by sharing the details of the needs assessment as well as the intelligent tutor + machine learning design, while asking audience members at three junctures to engage in Boyer’s (1990) scholarship of application: to debate and discuss from your own research perspective what you consider to be appropriate next steps. Bring your lunch, and your willingness to dialogue and debate!​​


Campbell.jpgMcQueen.jpgValue of Information: Active Learning Through User-Friendly Computations and Application Discussions
Jonathan Campbell. Associate Professor
R. Brett McQueen
, Assistant Professor
Center for Pharmaceutical Outcomes Research

Tuesday, October 17, 2017 | 12:00 - 1:00 PM
Education 2 North, Room 2302

Have you ever wondered about whether you’re making the right decision? It turns out there is a science for that called decision sciences. Within decision sciences, there is an approach called decision analytic modeling that gathers up the best available evidence, relates and forecasts this evidence under certain circumstances and alternatives, and spits out quantitative metrics that are important for making a decision.    
During the first part of this seminar, we will give an overview of modeling methods and motivate the topic of value of information. Value of information can be generally considered as the expected cost of uncertainty in the decision (i.e. the probability and consequences of making the wrong decision).  During the second half-hour, through active learning, we will engage the audience by providing a one-page simplified exercise of calculating one measure of the value of information. We will walk the participants through the one-page exercise using visual aids.  

The objective of this seminar is to offer participants a practical primer on understanding the calculation, presentation, and policy implications of value of information. The seminar will conclude with value of information applications and our proposed D2V pilot project.  


Larry Hunter.jpg
Artificial Intelligence and Medicine
Larry Hunter, Professor of Pharmacology

Tuesday, September 19, 2017 | 12:00 - 1:00 PM
Education 2 North, Room 2302


Contemporary artificial intelligence methods are increasingly penetrating clinical practice.  This talk introduces the subject, discusses their strengths and limitations, and describes research directions in overcoming some of the most critical limitations. It is a relatively brief talk, intended to stimulate discussion (which there will be plenty of time for). ​


dan matlock.bmp
​Patient Decision Making: T1 - T4
​Dan Matlock, Associate Professor, Division of Geriatrics

Tuesday, August 15, 2017 | 12:00 - 1:00 PM
Education 2 North, Room 2301
Slides


This presentation will use work in patient decision making and cardiovascular diseases to highlight concepts and scientific questions spanning the T1-T4 spectrum.  T1 will discuss some of the research on the basic psychology of decision making. T2 will discuss some of the strategies that are being studied to improve patient decision making clinically. T3 will discuss the implementation strategies to actually get these strategies implemented in the real world. T4 will discuss some of the policy and cultural advances in improving patient decision making.


labeikovsky.jpg
Version Control Systems as Catalysts for Collaboration and Reproducible Research
Wladimir Labeikovsky, Health Sciences Library Bioinformationist

Tuesday, July 18, 2017 | 12:00 - 1:00 PM
Education 2 North, Room 2301



Scientific research is inherently a collaborative enterprise. Technical advances and pervasive connectivity allow scientists to share data and analyses on a large scale. With these powers, however, comes organizational debt that most researchers still approach in a rudimentary fashion. In this talk, we will discuss the basis and advantages of software engineering systems for version control of code and their application in different aspects of biomedical research. We will then describe the landscape of specific tools currently available for this purpose.​


​Expanding the Reach of Palliative Care: Adapting a Psychosocial Intervention for Latino Persons with Advanced Cancer
David Bekelman, Associate Professor of Internal Medicine
Stacy Fischer, Associate Professor of Internal Medicine

Tuesday, June 27, 2017 | 12:00 - 1:00 PM
​Education 2 North, Room 2301

There are major opportunities to improve the value of health care for people with serious illness. For many with serious illness who are near end of life, the default is to provide intensive treatments (i.e., intensive care unit, cardiopulmonary resuscitation, mechanical ventilation) regardless of patient preference. Furthermore, people with serious illnesses have poor quality of life and are often depressed. These problems are even more significant in Latino persons with serious illness. Palliative care improves the value of health care by addressing these problems. In this project, we take an initial step to extend the reach of palliative care to Latino persons in the community by adapting a palliative care psychosocial intervention for Latinos with advanced cancer.  We will describe the process of culturally tailoring this intervention by engaging Latino persons and other experts. The psychosocial intervention is part of a more comprehensive community-based early palliative care intervention for Latinos that we will test further in subsequent research.