Interview: Gary Shorter, Director of Data Science, Quintiles on Big Data for Healthcare

Gary Shorter has 20 years experience in the Pharmaceutical and Healthcare industries and he is Global Statistician for four FDA drugs through approval. He has spent the past 5 years consulting on Advanced Analytics across drug development and healthcare platforms.

Gary recently delivered a keynote speech at Big Data Innovation Summit 2014 held in Santa Clara on “Data Science Now, Next & Beyond”. He shared his views on data architecture and science status today, what's happening now and then more on what the future is likely to focus in with respect to more complexity in data science techniques to delve deeper into insight.

Here is my interview with him:

Anmol Rajpurohit Q1. Can Big Data help the US Healthcare system fix one of its biggest challenges - rising medical costs? What viable strategy would you recommend for that goal?

Gary Shorter: Larry Page, Google CEO, believes that its feasible to save thousands of lives through the use of Big Data, and certainly I would agree that with a much more integrated set of healthcare information and a focus on personalized healthcare it is possible to provide insight that will support better patient health, which is what Quintiles is all about.

As to rising medical costs, that’s a tough nut to crack for sure in the US. Unfortunately there is no one answer and there are many very famous economists, policy makers and colleagues at Quintiles working on this that are much more qualified than myself. I am going to stick with the provision of using big data to provide insight for better patient health. It’s likely to take the rest of my life to succeed but I prefer my chances over fixing the medical cost issue!

AR: Q2. Can you highlight some of the unique aspects of working with Healthcare data? What kind of data problems do you look into?

GS: Healthcare data is the biggest challenge of them all. You can tackle claims data and identify certain activities across an event that occurs differently depending upon region and type of event. You can tackle physicians EMR (Electronic Medical Record) data and surface insights over time concerning specific patient populations and determine more patient impacts to events that occur. You can connect directly with patients and determine personalized impacts and their views on what works and doesn’t work for healthcare.

You can find sub-populations that have specific meta-omic characteristics that could provide insight in personalization of patient outcomes. Within themselves these each provide challenges on the scale and complexity of the data and on the desire to find insight and the resulting validity of the insight found.

Running a regression model and returning a high correlation does not mean I have found causation. If I run thousands of models I will find something. If I run thousands of models and receive thousands of p-values, there will be significance in a portion of those.

The trick is in taking the step back and looking across the healthcare universe, collaborating with colleagues in the Patient, Payer, Provider and Pharma spaces that bring these different views to the table and using data across those disparate sources to both provide new insight but also confirmation of results across each source to justify and validate the results.

AR: Q3. What are the key merits of using Quintiles Infosario Platform? How does it enable data-driven decision making in clinical development?

GS: Quintiles Infosario® is a comprehensive, end-to-end clinical development suite of services that seamlessly integrates data, systems, processes and Quintiles’ knowledge. It enables a new way of developing and commercializing drugs and devices. It enables data transparency, continuous access to information, informed decision-making, and data- driven processes – all underpinned by Quintiles’ Technology. New services are being developed that utilize additional real-time data mixed with new data sources, providing a strong basis of ever expanding capabilities to both the Pharmaceutical and Healthcare industries.

AR: Q4. Can you explain the term "Topological Data Analysis"? How does this benefit Healthcare solutions?

GS: Topological Data Analysis looks at the shape of a cluster of data. The intent is to discover through cluster analysis a network of connected patient events but to then take that a step further in recognizing the shapes of those clusters and how over time differences occur in those shapes as a patient progresses through their events. For me this is the next stage in the development of how we view disease progression, the changes in patients health over time and hence how to monitor patient similarities to identify a more personalized health path based upon the collection of real world evidence.

The problem with healthcare data is that we need more of it, and as more of it comes to us we eventually could become drowned in its sheer volume.

Using the Topological approach we identify the relevant shapes by specific nodules, compressing the data into more manageable amounts. Then as new patients come in with progressively worsening health events they can be monitored in real-time due to the compressed data, looking at how they match their results to those of the Topological analysis and allowing the Doctor to see the expected long-term event progression and how other patients responded to differing treatment options.

Here is: Second and final part of the interview

Relevant posts:

Healthcare Analytics: Identifying Leaders and Key Trends

Interview: Xinghua Lou (Microsoft) on Mining Clinical Notes and Big Data in Healthcare

Data Mining Medicare Data – What Can We Find?