Pivot

on YouTube ↗05m

" summary

Takes ~1M adverse-event reports, expands them into ~360M patient–drug–drug buckets, and analyzes them with gamma Poisson shrinkage estimator to score interaction confidence. Data ingestion runs on Hadoop/MapReduce, using 20 map-reduce jobs and then explores results in R and renders graphs with Gephi. Visualization surfaces expected clusters (HIV, breast/prostate cancer) and surprising cross-cluster links — e.g., a blood thinner connecting via loxanine to takapron — suggesting metabolic pathways worth biological follow-up.

AI-written summary · about

" chapters (5)

1. 0:00Understanding adverse events
2. 1:04Challenges in data analysis
3. 2:00Big data methodology
4. 3:08Visualizing drug interactions
5. 4:40Implications and conclusion

" related talks

• 31mAggregating combinatorial biomedical graph ranking results for drug repurposi...Daniel Korn · DeclMed 2023
• 10mA Story about Big Data in HealthcareHenry Wei · Strata Rx Conference 2013
• 18mTranslating a Trillion Points of Data into New Insights into DiseaseAtul Butte · Strata Rx Conference 2013
• 32mData Mining in the EMR Gold RushDavid Uhlman · OSCON 2010
• 07mData Collection and ReorganizationAndy Arm, logan wilding · Strata Rx Conference 2013
• 44mTransforming FHIR Data for horizontally scalable analyticsBashir Sadjad · DevDays 2023
• 19mHitting the Target(s) of Data OrchestrationAlexandros Kouretsis · posit::conf(2023)
• 14mDigging a Pit of Success for Your OrganizationAaron Chafetz · rstudio::conf(2022)