Dear all,
I will talk about phenotyping on April 1st, 2022 at 10am US Central Time (11am US Eastern Time). It is a Zoom seminar in the PowerTalk series by the UAB Informatics Institute. The abstract is here at the end.
Link to the series:
https://www.uab.edu/medicine/informatics/education/biomedical-informatics-powertalk-series
Direct link to register (for free) to my talk:
The talk tries to give an intuitive review of rule-based phenotyping, its function within a study, and how the task can look like today. From there it touches more abstract topics:
- Using three-valued boolean algebra to calculate phenotypes, revise their logic, and anticipate the impact of data missingness
- The relationship between data representation and phenotyping
- Phenotyping clinical scores from sets of data elements
- Computational complexity matters in phenotyping
Here in the OHDSI community we stumble upon the world’s greatest “phenotypers” – who am I to speak about the topic. 
My goal is just to contribute a cent or two, distilled from hands-on experience, and promote conversation.
Kindest regards,
FabrĂcio
Abstract
Electronic phenotyping is the process of identifying patient subjects for study, as well as their covariates, and is integral to the majority of secondary use cases of electronic health data sets. Rule-based phenotyping, the most common kind, starts from a definition in medical language, and manually produces a computer algorithm to model that definition based on how patient data is expected to exist, if it does. This presentation leverages examples from current phenotyping practice to walk the audience through its work process, and illustrate hindrances found along the way. Discussed topics include practical matters, such as deciphering mismatches in the results between phenotypes, uncovering logical gaps in medical definitions, comparing phenotypes definitions themselves, computing clinical scores from groups of data elements, as well as higher-level matters such as current software tools, and computational complexity in phenotyping.