12/ As an ID fellow doing EHR research, this represents my path toward independence - using informatics to understand host-pathogen interactions.
Huge thanks to Josh Denny lab at NHGRI, #AllofUs Research Program, and www.niaid.nih.gov/about/infect... for making this possible! 🙏
#MedSky #IDSky

11/ For the #IDSky #EpiSky #AcademicSky community:
What other episodic infectious diseases need computable phenotypes? UTIs? Pneumonia subtypes? C. diff?

I'm building expertise in ID EHR methods - reach out for collaborations! Always happy to help with phenotyping approaches in and out of ID.

10/ What's next from our NHGRI lab:
🧬 GWAS to identify genetic risk factors for severe respiratory infections
📈 Clinical determinants of hospitalization across viruses
💊 Diagnostic stewardship for multiplex respiratory panels

Phenotypes are just the beginning! #IDSky

9/ Important caveats (transparency matters):

Low sensitivity = can't estimate true prevalence
#AllofUs demographics ≠ general US population
Oversamples severe disease, misses mild infections

Remember: Always validate your phenotypes

But, methods likely applicable to any US EHR dataset

8/ Want to use these phenotypes or create your own EHR methods? Let's collaborate! 🤝
✨ Code available in #AllofUs Community Workspace
🧬 Perfect for GWAS, health disparities, clinical outcomes research
📊 Framework adaptable to other episodic infectious diseases

DM me - my passion is EHR phenotyping

7/ Unexpected finding: Our phenotype oversamples severe disease 🏥
Non-flu/non-COVID viruses showed high hospitalization rates vs. known epidemiology. Why?

Limited outpatient testing
No specific treatments = less testing motivation
Diagnostic value mainly in hospital settings

6/ Our phenotypes matched CDC surveillance beautifully 🎯
✅ Expected seasonal patterns
✅ Geographic distribution aligned with national data
✅ Virus type proportions matched NREVSS/GISRS
Mean absolute error: <6% for all viruses vs CDC positivity rates

5/ Plot twist: COVID disrupted everything 😷
Common coronavirus ICD codes spiked during the pandemic (clinicians coding concern for COVID) despite no positive tests. Excluding hCoV codes after Feb 2020 corrected counts and PPV.

Lesson: Always validate phenotypes across time periods!

4/ Performance varied dramatically by virus:
High ICD sensitivity: Influenza (67%), RSV (55%)
Moderate: COVID-19, adenovirus, hMPV (33-45%)
Low: Rhinovirus (9%), parainfluenza (8%)
But, PPV was consistently high (90-97%) for non-flu/non-COVID viruses 📈

3/ Our approach combines 3 data types within 90-day episodes:
🔹 Virus-specific ICD codes (e.g., J12.1 for RSV pneumonia)
🔹 Laboratory results (PCR, antigen, culture)
🔹 Antiviral prescriptions (flu/COVID only)
This integration > any single component alone, with varying breakdowns by virus.

2/ Why does this matter? 🤔
Respiratory infections are the most common human diseases, but research typically focuses on influenza, RSV, and COVID-19. Meanwhile, rhinovirus, hMPV, PIV, and common coronaviruses cause comparable illness but remain understudied.
#EHRphenotyping could unlock this. #IDSky

What an important conversation. Thanks for putting it out there!

DC! So lovely seeing all of the activity across the world for #standupforscience2025.