OHDSI Home | Forums | Wiki | Github

OHDSI Symposium 2023 - Phenotype Development and Evaluation Workgroup meeting 8am to 12pm

#1

Are you interested in leading an OHDSI characterization study?

The Phenotype Development and Evaluation Workgroup is pleased to announce an in-person meeting scheduled at the 2023 OHDSI Global Symposium on Saturday, October 21, 2023, from 9:00 am to 12:00 pm. The OHDSI Phenotype Development and Evaluation has the Objective Key Result (OKR) to “a) harden Phenotype development and evaluation, b) Improve collaboration by enabling community wide participation on Phenotype Development and Evaluation, c) Promote the usage of OHDSI Phenotype library.” One of the key results of this workgroup is to enable scientists to use the OHDSI Phenotype Library to perform collaborative scientific research for peer reviewed publications.

To further this key result, we aim to utilize the OHDSI Symposium 2023 as a collaboration opportunity to enable scientists to lead OHDSI studies that leverage the Cohort Definitions available in the OHDSI Phenotype library. The leads will be enabled to use the characterization evidence generated for the cohort definitions in OHDSI phenotype library for peer reviewed publications. See last paragraph below on the opportunity to lead a paper.

We will start by first picking a subset of cohort definitions in the OHDSI Phenotype library that were chosen to be part of the ‘How Often’ incidence rate analysis. Once we determine they have reasonable operating characteristics, we will supplement the incidence rate results of HowOften with population-level characterization data derived from the OHDSI software CohortDiagnostics (CD). The combination of the two may be used by study leads and their collaborators to write scientific research papers for peer reviewed publications.

The session aims to achieve the following outcomes:

  1. Complete assessment of potential measurement errors associated with the cohort definitions featured in the ‘How Often’ study across all data sources involved.
  2. Perform descriptive analytics of the target/outcome cohorts in the ‘How Often’ study.
  3. Write a manuscript on the insights gained and submit for peer reviewed publications.

We welcome participation from:

  1. OHDSI collaborators and data partners interested in leading (i.e. first author) or contributing (co-author) significantly to a research paper on a phenotype.
  2. OHDSI data owners who would like to contribute to the population level characterization component of the study by running CohortDiagnostics in addition to Incidence Rate analysis of HowOften.
  3. Individuals who want to gain technical experience in running CohortDiagnostics on your data site.
  4. Individuals who are interested in reviewing and interpreting the output from CohortDiagnostics or phevaluator and learn how to identify potential measurement errors.

We are looking for volunteer study leads. If you wish to lead a characterization study as the lead investigator, we encourage you to contact the workgroup leaders via email at rao@ohdsi.org . If selected, similar to the OHDSI Phenotype Phebruary 2023, the workgroup leaders @Azza_Shoaibi, @Gowtham_Rao will offer help to the volunteer leads throughout the project. We will enable you in assessing the operating characteristics of definitions, identifying potential error sources, interpreting the characterization results, and facilitating manuscript preparation for peer-reviewed publications. Collaborators with all levels of experience are welcome.

How to register for the meeting:

If you haven’t registered for the 2023 Global Symposium, then you can register for both the meeting and the Symposium here.

If you registered for the Symposium, but have not selected which meetings you will attend, then log back into your Eventbrite account by using the confirmation email sent to you after signing up for the Symposium or via their website Eventbrite - Log In and Sign In You MUST be signed up for the Global Symposium in order to register for any of the WG meetings.

  1. Simply find your confirmation email you received when you registered through Eventbrite.
  2. Click on the “Go to My Tickets” button or the “View and Manage” your online order link. If you never created an Eventbrite account, you should do so to manage any registration changes you may have in the future.
  3. Once you are in your Eventbrite account, and in the Symposium event, click Edit on the right side. Scroll down to the 2 questions at the bottom, select your answers, and then press SAVE. You may change your selections at anytime by logging back into your account.

If you have any questions about registering, please reach out to Elisse Katzman at symposium@ohdsi.org

#2

Dear attendees of the OHDSI Phenotype Development and evaluation workgroup face to face meeting:

We are looking forward to meet you in person on the second day of the OHDSI Symposium 2023 on Saturday 21st 2023 starting at 8:00 am - 12:00 pm. We have more than 40 workgroup members registered for our 4 hours together. We are hoping that all of you will use this opportunity to begin our collaboration and write scientific papers on phenotype algorithms and their application.

To prepare, we ask that you read the following papers. @Jill_Hardin and @Joel_Swerdel will start by talking about this example.
a) example 1: Systemic Lupus Erythematous Using a data-driven approach for the development and evaluation of phenotype algorithms for systemic lupus erythematosus - PubMed
b) example 2: Phenotype algorithm Hidradenitis Suppurativa JMIR Dermatology - Phenotype Algorithms to Identify Hidradenitis Suppurativa Using Real-World Data: Development and Validation Study
3) example 3: Clinical Application - treatment pathway hidradenitis suppurativa https://www.sciencedirect.com/science/article/pii/S2666328723000846

This will be followed by Azza and me, talking about how to use a template like approach for our scientific work. Then we are hoping to divide ourselves into groups of 4-5 and work together on different phenotypes based on the framework laid out. To help, Anna OStropolets and Evan Minty will lay out some clnical motivation to kick us off.

Please bring your laptop to the session. We expect this to be hands on.

Gowtham

#3

List of volunteer leads (to be updated):

@MaximMoinat Optical Pathway Glioma definition (cohorts 307/308)
@aostropolets Pneumonia
@Christopher_Mecoli SCAR, Antisynthetase syndrome
@razzaghih Sickle cell syndrome
@Evan_Minty Major Non Cardiac Syndrome
@JudyRac (tentative) DILI
@zhuk perioperative aspiration
@pavelgoriacko Hyperkalemia

#4

HowOften cohorts https://github.com/ohdsi-studies/HowOften/blob/master/analysis_specifications/analysis1.xlsx

CohortDiagnostics output https://data.ohdsi.org/PhenotypeLibrary

#5

Facilities:

  • We are expecting a room with round tables, power supply, and flip charts.

Agenda:

  • 8:00am to 8:15am - Introduction: Goal, Background, and Opportunity.
  • 8:15am to 8:45am - @Jill_Hardin and @Joel_Swerdel to walk through an example.
  • 8:45am to 9:00am - @Azza_Shoaibi and @Gowtham_Rao to build on the work of @jweave17 from the 2022 OHDSI symposium face-to-face meeting (slides here). They will present and formalize the process for phenotype evaluation and validation.
  • 9:00am to 9:15am - Leads discuss their clinical idea and the clinical application question.
  • 9:15am to 9:30am - Break.
  • 9:30am to 11:00am - Breakout sessions. Work in groups of 5. Output: Review cohort definition, target clinical description, literature review, perform evaluation using cohort diagnostic, and build a PheValuator xSpec cohort.
  • 11:00am to 11:45am - Each group presents their insights and proposes a publication plan.
  • 11:45am to 12:00pm - Next steps and adjournment.

Workgroup leads: @Azza_Shoaibi and @Gowtham_Rao.
Volunteers: @Joel_Swerdel @Jill_Hardin

t