Tuesday, June 21, 2022: Restructuring Health Systems for Learning: Building Equity into the Learning Health System
Tuesday, June 14, 2022: LHS Collaboratory Workshop: Operationalizing Learning Communities
View all workshop presentation slides
This virtual workshop will review the concepts behind Learning Communities, which are foundational to Learning Health Systems. We will explore the phases of all learning communities, regardless of size or problem of interest: planning, initiating, implementing, and sustaining. Participants will engage in a collaborative activity to develop a learning community. Please join us either individually or as part of a team.
Tuesday, May 14, 2022: LHS Collaboratory Workshop: Learning Health Systems 101
This virtual workshop will review the basic concepts behind Learning Health Systems including the learning cycle, infrastructure, and learning communities. Participants will engage in a collaborative activity to design a learning cycle.
Dr. Friedman's presentation slides
Tuesday, April 19, 2022: Medical AI and Learning Health Systems
This recording contains Dr. Xiao Liu’s talk as well as the moderated discussion with Drs. Liu and Karthikesalingam.
Medical AI - Three Common Myths on the Path from Code to Clinic
In this talk, Alan Karthikesalingam will discuss lessons learned in Google's experiences of taking medical AI systems from early research to clinical implementation.
Medical AI - Raising the Bar on Evidence Standards
In this talk, Xiao Liu will discuss existing and new clinical evidence standards as applied to medical AI systems. Her talk will focus on recently published standards to ensure transparency and reproducibility of clinical evidence underpinning medical AI systems, including reporting guidelines such as SPIRIT-AI and CONSORT-AI.
Dr. Xiao Liu's presentation slides
Dr. Karthikesalingam's presentation slides are not public due to proprietary considerations.
Tuesday, March 22, 2022: Network-Based Research
PCORNet and the PaTH Subnetwork
In this talk, Kathleen McTigue describes the vision of PCORNet, its organization, and its value to the field of clinical research. PCORNet is divided into regional subnetworks one of which is PaTH. The organization of PaTH along with its priories will be discussed.
UM’s Site within PCORNet/PaTH
David Williams, PhD
The University of Michigan is an institutional member of PaTH/PCORNet.
In this talk, David Williams describes the organization and processes of the UM site within PCORNet/PaTH, studies in which UM participates, and resources for UM investigators interested in participating in PCORNet studies.
Dr. Kathleen McTigue's presentation slides
Dr. David Williams' presentation slides
Thursday, February 24, 2022: The Invisible Infrastructures and Infrastructuring Enabling Federated Data and Research Networks
The session will describe the landscape history, current status, and future of federated health data networks that are used to support a Learning Health System. Dr. Brown will describe the creation, infrastructure, operation, and uses of several networks from the perspective of a network coordinating center. Dr. Harris will describe insights from participating in multiple networks as a network partner, including infrastructure, governance, and operational lessons learned.
Dr. Jeffrey Brown's presentation slides
Dr. Marcelline Harris' presentation slides
Tuesday, January 18, 2022: Single and Multi-Institutional Experience in Radiation Oncology Constructing a Learning Health System for Machine Learning from Real World Data
This presentation will explore how Big Data Science and Informatics research can overcome deficiencies within the electronic health record and optimize real world data collection. We will discuss examples of how standardized nomenclature integrated into clinical workflow can enable statistical AI methods to advance clinical decision support and improve outcome models. Our successes in radiation oncology come from single multi-institutional, multi-national and multi-professional society collaboration.
Tuesday, December 7, 2021: The Current State of US Interoperability and Implications for Learning Health Systems
Interoperability is considered a key capability of a high-performing healthcare system and has been a top policy priority for more than a decade. Implementing interoperability is, however, a complex undertaking – requiring stakeholder coordination that tackles incentives, governance, technology, standards, and more. In this talk, Dr. Adler-Milstein will describe current approaches to interoperability and where we stand with respect to current levels of national adoption. She will then discuss the implications for Learning Health System efforts at different levels of scale.
Thursday, November 11, 2021: Special Fall Workshop: Operationalizing Learning Communities
This virtual workshop will review the concepts behind Learning Communities which are foundational to Learning Health Systems. We will explore the phases of all learning communities, regardless of size or problem of interest: planning, initiating, implementing, and sustaining. Participants will engage in a collaborative activity to develop a learning community. Please join us either individually or as part of a team.
Charles P. Friedman, Department Chair of Learning Health Sciences, Josiah Macy Jr. Professor of Medical Education, Professor of Information, Professor of Public Health
Building Learning Communities: Thinking about Culture as a Form of Infrastructure
Alexandra Vinson, PhD , Assistant Professor of Learning Health Sciences, Director of the Medical Education Scholars Program
Lisa Ferguson, MSI, Program Manager, Department of Learning Health Sciences, University of Michigan
Michelle Williams, MHI, Project Manager, Department of Learning Health Sciences, University of Michigan
Tuesday, October 19, 2021: Making Sense of Unstructured Data
Unlike structured data, unstructured data are often buried within free text clinical narratives that are difficult to analyze and interpret to derive useful insights. Free text cannot be easily categorized in the same way that a structured, numerical data point can, and unstructured data often have nuances that are not easily captured or represented in structured data.
This session will cover methods and techniques for interpreting and converting unstructured text into useful research data using two related, but distinct approaches: (1) Natural Language Processing (NLP), a specialized branch of AI focused on the interpretation and manipulation of human-generated spoken or written data; and (2) information retrieval, which often underlies many search engine technologies. This session will also highlight EMERSE, an open-source information retrieval tool that has been designed to help everyday users work with the free text documents (i.e., clinical notes) in medical records that is now being adopted by other academic medical centers.
Finally, attendees will hear directly from researchers about how they have used these methods and tools to enhance their research by accessing and harnessing the power of unstructured data.
Bringing data to the people: How a secure, self-service, free text search tool can empower clinical research teams and improve productivity
David A. Hanauer, MD, MS, FACMI, FAMIA, Director of MICHR Informatics Program, Associate Professor of Learning Health Sciences, Associate Professor of Informatics U-M School of Information, Clinical Associate Professor of Pediatrics
Promise of Unstructured Data
VG Vinod Vydiswaran, PhD, Associate Professor of Learning Health Sciences, Associate Professor School of Information
Using EMERSE to Improve Research Involving Rare Cancers
Christina Angeles, MD, Surgical Oncology, Director, Sarcoma Research Working Group
Making Sense of Unstructured Data
Xu Shi, PhD, CCMB Affiliate Faculty, Department of Biostatistics, University of Michigan
EMERSE at UCSF Unlocking our De-Identified Unstructured Data!
Leslie Yuan, MPH, Chief Information Officer, Clinical and Translational Science Institute (CTSI), University of California San Francisco
Thursday, September 30, 2021: LHS Collaboratory Kick-0ff
Janet and Bernard Becker Professor and founding Director of the Institute for Informatics (I2), Associate Dean for Health Information and Data Science, Chief Data Scientist, Washington University in St. Louis
The Learning Health (Record) System
Much has been written about the challenges associated with the use of current EHRs, however the promise of these technology platforms remans vast and mostly under-realized. This presentation will explore the ways in which Biomedical Data Science and Informatics research are helping to realize the potential of EHR technologies in the context of creating an LHS, from the optimization of workflow and human factors, to the generation of reproducible and systemic clinical phenotypes, to the delivery of emergent knowledge to both providers and patients via advanced clinical support systems.
Research Assistant Professor, Department of Biomedical Informatics, Scientific Director for PheWAS Core, Vanderbilt University
Techniques and Challenges for EHR Phenotyping
Electronic health records (EHR) contain a wealth of real world data that can be used for research purposes. However, extracting phenotype information from EHRs can be challenging. EHR phenotyping can be divided into two types: (1) Fast phenotyping which seeks to capture a broad swath of the medical phenome, and is often accomplished using coded EHR data (e.g. billing codes) and (2) slow phenotyping that seeks to achieve high precision and recall for a single phenotype, and often uses multiple EHR data types (e.g. medications, text, lab results). This talk will describe specific use-cases for both fast and slow phenotyping, and review challenges that are commonly encountered in creating research-grade EHR phenotypes.