Big Data Becomes Real Data in Precision Health

Last week I had the privilege of speaking at Stanford Medicine’s 7th Annual Big Data in Precision Health conference. Over the years, the event has represented an inspiring forum where leaders from industry, academia, and government can share ideas around the opportunity for data to impact human health. What made this year’s event special was that each speaker presented results of their work, making it undeniably clear we have now moved beyond big ideas and are seeing the real-world impact of data in transforming health.

Jeff Dean of Google kicked off the conference with a tour de force presentation of work from the brilliant team at Google demonstrating the impact of machine learning and data on everything from diagnosis to clinical care. One major theme of Google’s work was the power of images when combined with deep learning architectures. Jeff showed impressive results using retinal images, computed tomography (CT), and pathology images combined with deep learning to enhance physicians’ ability to diagnose complex diseases. In fact, image-based methods are now breaking new, unexpected ground. For example, detecting aspects of human biology in retinal images beyond what physicians can do, and even improving variant calling in genomics data!

A big theme across the conference was the use of technology and machine learning (ML) to gather meaningful health data from people. Stanford HUMANWIDE seeks to collect large, multi-layered datasets spanning molecular genomics to phenotypic and behavioral data from study participants. I was especially impressed by the work of Dina Katabi of MIT, who talked about repurposing Wi-Fi signals to monitor patients in their homes. Her team is able to use “Invisibles” as she calls the approach to monitor for falls and sleep patterns, even detect heart rate and respiratory rate! Katabi followed Zhenan Bao of Stanford who envisioned a future where technology becomes fully integral to the human form, from devices to implants. Seems like Implantables and Invisibles will battle it out as we find more creative ways of interacting with technology remotely.

As part of the incredible team at Recursion, my focus is on drug discovery and I was thrilled to share our work alongside companies in the space such as J&J, Novartis and Regeneron. We at Recursion have been particularly interested in the work coming out of Janssen, with leaders like Hugo Ceulemans breaking ground on using multi-task learning architectures to predict bioassay activity from historical data. Beyond discovery, Emma Huang gave a great talk about how J&J Innovation is thinking about evaluating ML-based opportunities across the entire R&D pipeline (with an awesome landscape slide that I’d like to get ahold of!). Representing Novartis, Mimi Huizinga presented some big ideas on data science efforts spanning clinical development and commercialization. Also impressive was the scale of Will Salerno of Regeneron’s approach to leveraging massive, human genomics datasets to understand human disease causation like never before. 

I’m missing a lot of great speakers in this post and can’t quite do justice to them all, but overall Euan Ashley and Janet Kalesnikoff organized a superb event. The message was clear that we’ve moved far beyond ideas, and we’re breaking ground on tech producing real results in healthcare. Over the next few years, I think some of the big questions will not be what we tackle, but how we begin to deploy solutions to rethink our workflows in healthcare and drug discovery. We’ll need to establish processes for radiologists, pathologists, and other clinicians to directly interface with ML-powered tools in their day-to-day. On the drug discovery side, we’ll need to implement predictive tools and ML-powered endpoints into the workflows of drug discovery R&D. Importantly, it’s not about replacing steps or practitioners, it’s about reimagining how practitioners engage with data in their domains to make better decisions more efficiently.

If you attended Stanford’s Big Data in Precision Health event or followed the live stream online, what stood out to you? It would be great to hear from others on their takeaways from the event.