Big Data Takes the Stage at Stanford

Theral Timpson

We're currently developing a series on big data here at Mendelspod.  So we jumped at the chance to attend the first 'Big Data in BioMedicine Conference' put on at Stanford in conjunction with the University of Oxford.  The conference gave a great overview of the topic, reaching not only into all that omics data, but health IT and public health as well.  

Stanford is a fitting place for the topic. It's the workplace for what I call the Big Three: Atul Butte, Mike Snyder, and Russ Altman  (all of whom have joined us at Mendelspod.)  Luminaries in the field from UCSC and Berkeley joined as well, including David Haussler of Human Genome Project fame, and Steve Brenner.

Watching some of the conference on livestream (video will be posted soon at, I made sure to make it in person to the panel 'Public Health and Big Data Policies'.  I wanted to hear a talk by Forbes columnist and GNS Healthcare CEO, Colin Hill.

Colin has been on my radar as several guests at Mendelspod have suggested him for an interview.   I have seen his face in dozens of emails, as he co-chairs O’ Reilly Media’s Strata Rx Healthcare Big Data Conferences.  He began his presentation with a quote by John Wanamaker, thought by some to be the father of modern marketing (not a character on Mad Men):

“Half the money I spend on advertising is wasted.  But I don’t know which half.”

Based in Kendall Square in Cambridge, MA, (another hub for big data and bio innovation), GNS Healthcare is using big data analytics to help in everything from drug development to patient diagnosis.

“Just as Pay Per Click solved Wanamaker’s dilemma, we are showing what is working for whom in heatlhcare.  We have the data,” Colin said in a steady, confident presentation.  

The holy grail for GNS and others in the field is to update standard of care with individualized treatment algorithms.  For example, GNS uses three dimensional modeling to analyze patient characteristics and predict drug effectiveness. 

Another example is their predictive test for “metabolic syndrome”.  This is a condition based on a combination of a patient’s baseline health data that, when occuring together, increases the risk of developing cardiovascular disease and diabetes.    If you are high in three of five basic areas: blood sugar, blood pressure, body mass index, cholesterol, and triglcerides, you are considered to have metabolic syndrome.  Colin pointed out that 4% of Americans are diagnosed, but that it’s estimated that 25% of us have it.  According to Hill, GNS’ test has a predictive power for the syndrome of 88%.  

As Colin spoke I began to get the big data big picture and consider all the possible algorithms and tests that are possible to run at companies like GNS.  Ways of improving drug development, clinical trials, omic profiling such as that done by Mike Snyder and 23andMe, drug effectiveness and toxicity--you name it.  And how are doctors going to keep up with all this information?  It’s impossible.  And how much of this will be regulated by the FDA?  

It’s obvious that private commercial enterprises such as GNS will be playing a major role in the future of healthcare.  I’m beginning to understand what Eric Topol is talking about by “homo digitus” and the patient as consumer.  Hill says that in today’s world of so much data, standard of care has become outdated.  

“Let’s not waste the data,” he said.  “Big data can solve knowledge blind spots and create as much impact as the ACA (Obamacare).”

I was eager to hear from Yael Garten, a data scientist at LinkedIn.  Is this social media giant up to something cool for healthcare?   Yael trained as a bioinformatician at Stanford, a protege of Russ Altman before joining LinkedIn.    It turns out Yael didn’t really have anything practical to offer yet.  Just fancy graphics of big omics data side by side with some LinkedIn Data, and yes, they looked similar.  But so what?  You can see a connector map like that on the back of your Delta Airlines napkin.  

She continued with some statistics and trends.  There’s a big rise in postings for data scientists at LinkedIn.  Fifty percent of email is now attended to on mobile devices.  Usage of desktops graphs differently through the day than that of tablets.  Data is being democratized.  So how is LinkedIn using this data to improve health outcomes?    The connection wasn’t made.

The final two presentations on the panel were by Dennis Wall from Harvard and from two scientists at the NIH.

Wall has been working on new diagnostic tools for autism, which he said has become an epidemic.  I had no idea that the numbers of autistic persons in the US was increasing at such an alarming rate.  It’s gone from one in 110 people to one in fifty!  Wall showed (graphically of course) that many affected by autism are in rural areas and do not have access to qualified caretakers.  What Wall and his team have done is to create a program which can look at amateur video footage of persons and diagnose with a high degree of accuracy whether those individuals are autistic.  

Peter Lyster is the program director for NIH’s BICB, or biomedical informatics and computational biology program.  He had the important task, along with a colleague whose name I didn’t catch, of communicating the strong commitment the NIH is making to informatics.  About $1 billion of the NIH’s $30 billion budget goes to the sector, and Peter announced a new large initiative of $100 million in grants.  

The presentations were followed by a lively Q & A with the four speakers.  My favorite question came from John Hornberger, an economist and member of an ASCO ethics panel.  He wanted to know of Colin what considerations GNS has made for the ethics of big data.  How are they monitoring the quality of their tests?  Do they look to the FDA for oversight.  The question was aimed at Colin, but applied to all the panelists and all of us at the conference.   Is it right for LinkedIn to use data gathered from all our accounts and postings and be sold to healthcare companies?  What are the dangers of diagnosing autism not in person by a physcian, but by a computer using a two minute video clip?    Is there enough oversight of the natural market forces which use our data in new ways?  

Colin’s response was somewhat cavalier as would be expected from an entrepreneur.  He pointed out that GNS is currently looking at 100 million lives through various studies.  That it’s happening now.  That the standard of care is outdated because it is not data driven and people need help.

At the break I met Hornberger to secure an interview for our upcoming series to discuss ethics and big data.

“This is a new industry, a new conference,” he acknowledged.  “We don’t want to throw out the baby by regulating too early.  But there is a discussion to be had.”  

In addition to Hornberger and Hill,  Atul Butte, the conference ring master and face of all things bioinformatics agreed to come on again as part of our own big data series.  And David Haussler from UCSC will be joining us for a discussion about what will be the winning utility platform for genomics data.  

For being the first stab, the conference was superbly organized with a glamorous stage setting at the Li Ka Shing building at Stanford.  (The Li Ka Shing Foundation was the major underwriter for the conference.)  Not surprisingly big data folks are big tweeters.  You can read much more about the event at #bigdatamed on Twitter.

New to Mendelspod?

We advance life science research, connecting people and ideas.
Register here to receive our newsletter.

or skip signup