BIG data: How pharma can use it to its advantage

MDCPartners founder and CEO, Mireille De Cré, explains how pharma R&D can make the most of the big data revolution.

The big data hype train has been gaining momentum for a number of years and its rise has been well documented. However, pharma companies are still perplexed by the scope of what big data can do, and why, how and where they should use big data tools to their advantage.

With focus moving away from sheer volume of data and inherent variability and complexity of large unstructured datasets towards user-friendly analytic tools that can help businesses plan for the future, now is the time for big, small and medium size pharma to seize the ‘big data’ opportunity.

Trailing behind

Despite being a sector priding itself on innovation, pharma has trailed behind when it comes to adopting new technology. For example, the timid approach that pharma has taken with social and new media (due in part to regulations and the industry’s code of conduct) and in the selection of clinical research partners.

Programme teams are still relying heavily on their existing networks or CRO partnerships, rather than tapping into the vast amount of data out there on scientific expertise and patient populations. Nevertheless, data analytics can open a whole new world of possibilities in the research of new compounds with the power to radically improve multiple stages of the drug discovery process.

In clinical development, phase II–IV, new technologies have emerged ‘en masse’ since the early nineties, but only a handful have attempted to go beyond the value proposition of merely smart reporting of in-clinic generated data within controlled and randomized clinical trials.

This is not surprising given that the data entering the pharmaceutical data hub is unstructured and messy, including many duplicates or errors. The formidable challenge here is to clean and link the data, building a consistent and reliable foundation upon which any model of value-adding analytics can be developed. Once that hurdle is taken, data can become a fertile ground for deriving intelligence. This is the fundamental proposition that big data offers; to speed up process and improve decision making.

What is meant by ‘analytics’?

Algorithms and statistics can start making ‘models’ of real world activity, which can be tested according to real world measurements. Machine learning takes this activity one step further — rather than handcrafting a relatively simple model, a model with up to hundreds of parameters is ‘trained’ using large amounts of input data. Everyday examples of machine learning, intrinsically linked to computational power, are face or fingerprint recognition, tumour cell classification and analysis of microarrays.

The concepts of analytics have also been extensively applied in the linking of information related to clinical trial metrics, investigational site locations, patient recruitment metrics, investigator and medical expert identification.

Unifying data on multiple parameters, when combined and semantically matched, helps to connect the dots for efficient clinical trial planning and patient recruitment. Whereas teams used to perform manual searches on publicly available sources (such as publications, trial registries and conferences) automated analysis can now provide companies with an improved picture on drug development programmes to assist business decision making.

External vendor support

Despite considerable investments in data science analytics teams, more pharma companies are using external vendors to support the implementation of a big data strategy. One such tool, ta-Scan, founded 10 years ago by MDCPartners has been employed by global pharma companies to change the way companies use data. One of the elements that has been most sought after has been the ability to consolidate public and internal unstructured data from the large data silos that still reside in pharma companies.

The quality of data output from tools such as ta-Scan that analyse vast public data repositories has earned the trust from big pharma and has piqued their curiosity to expand the analysis to their internal silos and real world data. It has been purported that real world data can cost in excess of $20 million with no clear measurable or KPI to evaluate the impact of interventions on patient cohorts.

Indeed, opening the lid on this internal data and providing it to experienced data companies may reveal valuable analyses that can translate data to information, guiding business intelligence and initiating alternative strategic objectives.

What is clear is that big data is here to stay and will continue to have a growing infl uence in terms of its impact to business decision making across a number of sectors. Some of those in pharma are firmly grasping the opportunity that analytics-based decision making can bring… And those that aren’t will need to catch up before they lose out.