Big Data - Revealing the unseen

Big Data is one of the major trends in the IT world and Roche is investigating ways to learn from the data.

“At Roche, we use ‘big data’ as an umbrella term for methods and technologies that enable us to process vast amounts of data,” says Simon Ulrich, Head of Business Intelligence & Master Data Management in Pharma Informatics.

Three key aspects and, at the same time, key challenges associated with big data are the sheer volume of data, the diversity of data formats, and the speed with which it needs to be generated, processed and used. Until 2003, for example, around five billion gigabytes of data were generated by humankind on the internet. In 2011, the same volume of data was generated within 48 hours, and today it only takes a few minutes. What does that mean for Roche?

Matthias Essenpreis, Chief Technology Officer for Diagnostics: “Big data is an opportunity for us and the whole industry. The main challenge is not so much the technical hurdles, but obtaining the right data from multiple sources and understanding how that data can be interpreted and used so that we can generate valuable outputs.” Roche has identified four areas to focus interest on.

Zeroing-in on the Roche approach

The first is ‘social analytics’ and involves working with online communities which allow patients to share their stories and experiences. “Roche wants to gain access to this data to use it for research purposes. Besides issues such as the language of the patients’ comments, considerations on legal and regulatory requirements need to be taken into account as well as data privacy protection.

Another field is “data exploration,” which uses statistical methods and “data mining” techniques like the analysis and detection of meaningful structures within unstructured text. One goal is to analyze that data in order to generate predictions for the future instead of only looking into the past. The third area is “data warehouse augmentation.” These applications are designed to improve understanding of large amounts of unstructured raw data, for example by combining and comparing it with structured data sets. Lastly, “operational analysis” deals with data created with sensors from production machines and others, using the resulting data streams for projections and optimizations.

To fail quickly and cheaply

At present, nine big data projects are under way at Roche and more pilot requests are under review. “We need to be aware that big data technology only starts to emerge. Looking at all sectors as a whole, 85 percent of all projects and pilots do not deliver any insight that can be converted into a direct business benefit,” says Simon Ulrich.

“Pilot projects in pRED are trying to find out, how typical questions in research can be answered by searching through huge text data bases. For example, a platform has been created to make medical abstracts and other academic publications more accessible to researchers”, says Bryn Roberts, Head of pRED Operations.

Global Technical Operations is currently investigating big data technology to optimize biotech production, for example by looking at the incoming data streams. The long-term goal is to develop a system that can help to predict the success or failure of production steps.

A joint Global Product Strategy, Medical Affairs and Product Development project is aiming to use data which has been generated outside of controlled clinical studies. These records could still be relevant to demonstrating the effectiveness of medicines.

“Even the external IT market cannot offer us all solutions we are looking for,” says Simon. “However, with our platform, we can respond flexibly to new developments. So when it comes to big data, Roche is well prepared.”

A piece of a puzzle

Change is occurring across the global healthcare system today.  Some of this change is driven by cost and regulatory pressure to adopt more efficient, technology-driven solutions. At the same time, patients are increasingly empowered with technology to measure and manage their own health and to increasingly connect with one another.  The emerging innovations – together referred to as “Healthcare Technology” – often rely on or produce vast quantities of data.  Big Data is a piece of a larger revolution in Healthcare Technology in which we must work with new technology tools for our industry and develop innovative partnerships across healthcare as a whole.

Tags: Science