ADLM Meeting of the Minds

Lab directors (and/or assistant directors); Lab managers (supervisory and/or non-supervisory); Physicians; Pathologists

Intermediate

A laptop or personal device for notetaking and audience participation. Please note that charging stations are not available at audience tables, bring a fully charged device for this activity.

• Presentation slides
  
• Sample code
  
• Synthetic datasets
  
• Further reading/resources

After participating in this course, participants will be able to: 

• Articulate the concepts of diagnostic utility and value in clinical laboratory medicine.
  
• Assess and select real-world data sources for comparative diagnostic studies.
  
• Apply foundational and advanced epidemiological causal inference methods to diagnostic evaluation.

• Moderator: Lee Schroeder, MD, PhD, University of Michigan
• Andrew Admon, MD, MPH, University of Michigan
• Ryan O'Connell, MD, MBA, University of California, Irvine

(10 minutes) Welcome & introduction (Lee Schroeder)

Dr. Schroeder will open the course by outlining the agenda and reviewing logistical details. This introductory segment will frame the importance of focusing on diagnostic value and set expectations for the interactive and hands-on nature of the course.

• Why focus on diagnostic value?
• Agenda and logistics

(20 minutes) Framing value in diagnostics (Lee Schroeder)

Dr. Schroeder will introduce key concepts related to diagnostic value and utility, including how diagnostic testing contributes to patient outcomes and health system performance. The session will address strategies for selecting appropriate outcomes to target when assessing diagnostic value. Participants will then engage in a hands-on group exercise to evaluate a set of value-based diagnostic targets and discuss their relevance and feasibility.

• Defining diagnostic value and utility
• Selecting appropriate outcomes to target
• Hands-on exercise: Groups evaluate a list of value-based diagnostic targets

(40 minutes) Causal inference toolset for diagnostic utility evaluation (Andrew Admon)

Dr. Admon will introduce core causal inference methods for evaluating diagnostic utility, comparing randomized trials with real-world evidence approaches. Participants will explore tools like Directed acyclic graphs (DAGs), target trial emulation, and propensity scores, discuss study design strategies, and work collaboratively to build and interpret a diagnostic strategy DAG for a selected project.

• RCTs vs. real-world evidence
• Core methods: DAGs, covariate selection, target trial emulation, propensity scores
• Study design strategies for diagnostic comparative effectiveness
• Hands-on group exercise: build and interpret a diagnostic strategy DAG using prior project identified

(20 minutes) Data sources for comparative diagnostic studies (Ryan O’Connell)

Dr. O’Connell will review key data sources for comparative diagnostic studies, including EHR platforms, institutional warehouses, standardized data models, and centralized databases. Participants will examine strengths, limitations, and practical considerations, and engage in discussion on selecting the right data source for specific diagnostic questions.

• Overview: EHR platforms (Epic, Cerner) and institutional data warehouses
• Data models: OMOP for standardization and multi-site studies
• Centralized databases: Epic Cosmos, CMS VRDC, OptumLabs
• Assessment: Strengths, limitations, and practical considerations• Discussion: Matching diagnostic questions to appropriate data sources

(15 minutes) Break 

(25 minutes) Extracting and preparing real-world data (Ryan O’Connell/Lee Schroeder)

Drs. O’Connell and Schroeder will guide participants through identifying cohorts and outcomes in EHR systems, including Epic Clarity. The session will address common data challenges and feature a live demonstration on preparing datasets for diagnostic strategy comparisons.

• Identifying cohorts and outcomes in EHR (Epic Clarity)
• Managing data challenges: mapping, missing data, and loss to follow-up
• Live demo: preparing a dataset for diagnostic strategy comparison

(30 minutes) Hands-on analysis: Assessing diagnostic utility (All) 

Faculty will lead a hands-on session on assessing diagnostic utility, covering propensity scores, analytic comparisons in R, and calculation of diagnostic performance measures. Participants will complete a guided analysis using sample datasets and code.

• Propensity score generation, weighting, and comparisons in R
• Calculating and interpreting diagnostic utility measures
• Guided hands-on analysis with sample code and data

(15 minutes) Interpreting and Applying Results (All) 

This section will focus on interpreting analytic results and translating findings into practice. Faculty will discuss strategies for communicating results to both clinical and administrative audiences and for moving from analysis to meaningful practice change.

• Communicating findings: clinical and administrative perspectives
• Moving from analysis to practice change

(10 minutes) Closing and Q&A (All)

The course will conclude with a summary of key points and takeaways, a review of resources for continued learning, and an open discussion to address remaining questions and identify next steps for participants.