What it takes to process healthcare data?

on YouTube ↗47m

" summary

IQVIA processes 600 million de-identified patient records from roughly 1 million data feeds and 140,000 suppliers, so engineering focuses on scale, sliding cost, and constant ingestion rather than batch rebuilds. Systems implement privacy impact assessments to measure re-identification risk, support re-bulking and opt-out removal, map many local drug/diagnosis dictionaries into a global ontology, and choose stacks like Cloudera/Databricks for ingestion and Snowflake for BI. Operational choice favors format-specific plugins over forcing a single input standard.

AI-written summary · about

" chapters (12)

1. 0:00Processing healthcare data intro
2. 2:29IQVIA healthcare data scope
3. 5:01Big Data in healthcare
4. 6:07Scaling healthcare data challenges
5. 10:25Complexity of medical data
6. 12:57Regulations and data privacy
7. 16:24Privacy impact assessment
8. 20:19Data supplier agreements
9. 28:51Standardizing data formats
10. 31:58Data dictionaries interoperability
11. 36:33Healthcare data challenges summary
12. 38:20Q&A

" related talks

• 31mAI in healthcare: Data driven transformationShahidul Mannan · Global Artificial Intelligence Conference 2020
• 41mAccelerating privacy-enhancing data processingFlorian Stefan · PyCon Italia 2025
• 33mHealthcare "big" Data from Data Management to Web AppsAndrew Nguyen · Clojure/West 2014
• 07mData Collection and ReorganizationAndy Arm, logan wilding · Strata Rx Conference 2013
• 54mUntangling Healthcare with Spark and DataflowRyan Brush · Philly ETE 2016
• 10mHealth Data Anonymization and PrivacyKhaled El Emam · Strata Rx Conference 2013
• 17mHealth Care Analytics: The Key Is IntegrationTom Davenport · Strata Rx Conference 2013
• 28mHerding Data to the CloudPatty Bragger-Wilkinson · Indy Cloud Conf 2020