Research Fellow, Mass General Brigham
1 active project
Temporal Representation for Improving Accuracy
Scientific Questions Being Studied
EHR diagnosis codes use for billing are known to be rife with inaccuracies. For example, a patient may be evaluated for diabetes mellitus but then have a diabetes mellitus code assigned to their chart. As a result, using the structured diagnosis codes in an analysis may give inaccurate conclusions. For large genetic studies, one way to improve the code accuracy is to only consider patients as having the disease as those with at least two diagnosis codes on different days. There are other simple heuristics that could be applied to improve accuracy across various disease by incorporating temporal patterns of various features. This project aims to study these temporal patterns to see if there is a rule that can be used to improve the accuracy of identifying patients with specific diseases.
- Methods Development
My plan is to use the survey data and EHR data to create more refined definitions for phecodes. Phecodes are a group of labels used to define a specific disease. I will augment these phecodes with temporal patterns to improve their accuracy. We will benchmark their performance against the survey data.
I will potentially learn how certain definitions might work better for certain populations.
Demographic Categories of Interest
- Race / Ethnicity
Data Set UsedRegistered Tier
- Zachary Strasser - Research Fellow, Mass General Brigham
You can request that the All of Us Resource Access Board (RAB) review a research purpose description if you have concerns that this research project may stigmatize All of Us participants or violate the Data User Code of Conduct in some other way. To request a review, you must fill in a form, which you can access by selecting ‘request a review’ below.