Global Health Monitoring

Global COVID-19 Impact Monitoring

SurveyDebiasBenchmarkRankPredictDistribute

Led at Boston Children's Hospital. Led analytics and machine learning for the Meta/UMD COVID-19 Trends & Impact Survey, benchmarking self-reported signals against official case data worldwide.

114+ countries30M+ survey responses

How it works

Survey Collection

Billions of daily impressions across dozens of languages reached respondents in 114 countries.

Data Ingestion

Automated ingestion, weighting, and anomaly detection using Python and R.

Bias Correction

Bayesian approaches adjust for sampling bias and improve representativeness.

Quality Assurance

Anomaly detection and QA protocols identify data issues before analysis.

Benchmarking

Survey signals are compared against established epidemiological metrics and vaccination data.

Modeling

Machine learning models predict COVID-19 positive test cases using engineered features.

Privacy Safeguarding

Secure aggregation, suppression thresholds, and QA reviews run prior to any data sharing.

Distribution

Delivery via API feeds and dashboards to WHO, CDC, and national health agencies for situational awareness.

Outcome

Co-first author; dashboards adopted by WHO, CDC, and national agencies for real-time pandemic situational awareness.

Citations

PNAS paper