Bringing Clinical Data to GCP for Automated AI Discoveries
summary
Imagia Deep Radiomics uploads deidentified clinical and imaging cohorts into Google Cloud Storage, runs thousands of pre-trained feature extractors across each image to find statistically validated signals (not immediate classifiers), and scales training and inference on Kubernetes Engine and ML Engine with TPUs. Every processing step records entries in per-process GCS buckets with trace logs so patient consent can be audited, and an automatic revoker that deletes derived data and retrains models removes a patient’s data and updates affected models for revocability and regulatory compliance.
AI-written summary · about
chapters (5)
- 0:00Introduction and vision
- 13:43Clinical AI platform overview
- 20:12Technical biomarker discovery process
- 37:34Trustability and future work
- 40:45Summary and Q&A
related talks
32mMachine Learning with Google Cloud Platform
09mHow AI is changing the healthcare system
23mBuilding Scalable, Observable MLOps Systems on Google Cloud
15mGCP for Data Scientists
03mAI solutions for healthcare
27mKubernetes + AI Joining Forces in the Battle Against Cancer
50mInnovation and data analytics: Solving real-world challenges
26mFrom Pixel to Patient: Making Medical AI Work in the Real World