The Data Exchange Podcast: Omer Dror on healthcare applications of data exchanges, privacy-preserving tools and methods, language models, and more.
Subscribe: Apple • Android • Spotify • Stitcher • Google • RSS.
In this episode of the Data Exchange I speak with Omer Dror, CEO and co-founder of Lynx.md, a startup that enables data exchanges and markets in the health and life sciences. Data exchanges match data providers and suppliers, with data buyers and consumers. A couple of years ago I devoted a keynote talk on data exchanges at a moment when landmark privacy regulations like GDPR were coming online. At the time there were early efforts to build data exchanges that focused on privacy, while also providing incentives for potential data suppliers to participate.
Fast forward two years later and the topic of data exchanges – both the technology and use cases – has started popping up across different industries and sectors, including healthcare and financial services. In this episode, Omer and I focused mainly on healthcare and life sciences and we covered many topics including:
- Healthcare data, and various privacy-preserving tools and techniques for working with sensitive medical data.
- Natural Language models and tools for healthcare.
- How the pandemic impacts how organizations handle and use healthcare data.
- Public data exchanges (examples: AWS Data Exchange, Azure Data Share, Snowflake Marketplace)
- Data exchanges in healthcare and the life sciences.
- Trends in AI and Data in Healthcare.
Omer is a proponent of differential privacy, particularly as an alternative to standard de-identification tools use in healthcare:
- ❝ I think health care is one of the biggest drivers for differential privacy. So differential privacy is still at very early stages, NIST has just come out with a challenge centered on differential privacy. And there are a lot of people working on this, including us. One of the advantages of differential privacy is that when you get access to data, you still can’t use it to re-identify individual patient records. And I think one of the most risky things about old style de-identification is now when there’s so much data out there, data volumes are so great, and it’s so easy to cross reference that data with social media and other sources. You can actually re-identify patients, even if they went through HIPAA Safe Harbor de-identification or any other type of de-identification.
Subscribe to our Newsletter:
We also publish a popular newsletter where we share highlights from recent episodes, trends in AI / machine learning / data, and a collection of recommendations.
Related content and resources:
- A video version of this conversation is available on our YouTube channel.
- Navigate the road to Responsible AI
- Download the 2020 NLP Survey Report and learn how companies are using and implementing natural language technologies.
- Yishay Carmiel: “End-to-end deep learning models for speech applications”
- Dan Geer and Andrew Burt: “Security and privacy for the disoriented”
- Rumman Chowdury: “The State of Responsible AI”
- Krishna Gade: “What businesses need to know about model explainability”
- Alan Nichol: “Best practices for building conversational AI applications”
[Image: Medical Care – from Wikimedia.]