EMPTY

Huge volumes of information are produced by health care systems every day, creating a data goldmine that has yet to be exploited. With computer power and analytical skills advancing apace, it is only a matter of time before data starts to deliver.

Or so goes the theory. In practice, many of the large datasets generated by health systems are poorly coordinated, and plagued by duplication and inefficiency. Technical issues aside, part of the problem is a lack of agreement on standardised outcomes that matter to patients and difficulty pulling together data from disparate sources.

With so many private and public labs taking an interest in the field, a shared approach to collecting, analysing, reporting and interpreting data is essential.

Big-budget collaboration

The €3.3 billion Innovative Medicines Initiative 2 is tackling this in its Big Data for Better Outcomes (BD4BO), programme, an umbrella under which several projects will look to improve how data is used to address Alzheimer’s disease, leukaemia, multiple sclerosis and cardiovascular disease, and to develop a European distributed data network.

It is a mammoth task, according to Magda Chlebus, Director of Science Policy at the European Federation of Pharmaceutical Industries and Associations (EFPIA), which coordinates the programme on behalf of its pharmaceutical company members. “We need smarter ways to manage healthcare budgets,” she told Healthy Measures. “One way to do this is to look at the outcomes that health interventions result in.”

While BD4BO has started with Alzheimer’s disease and has several more therapeutic areas in the pipeline, its real value will be in convening a diverse constellation of players under one roof. “If you want to embark on such an immense project, you need to be able to test the model and bring everybody into the tent,” Chlebus said. “It is important to have a safe harbour – a place where we can check the outcomes approach in real-life.”

Defining outcomes

The programme will address some of the biggest gaps in applying data analytics to health outcomes. For starters, it seeks to define ‘outcomes’. “The current definition of an outcome may differ depending on who you ask,” according to Chlebus. “Health authorities, doctors and patients may not have the same priorities. Similarly, regulators and reimbursement officials do not always see things in the same way.”

Finding a methodology to balance these sometimes-conflicting perspectives and settling on a universal definition is an essential but challenging task. Once the outcome is defined, researchers will need to determine what kind of evidence should be collected and analysed. “How will we measure the outcome and where is the data we need to evaluate?” Chlebus said.

Then the companies and academics that run the projects will explore how to access and analyse the data they need. To this end, other IMI projects are coming up with useful tools and methodologies for using real-world data that could complement BD4BO.

For example, RADAR CNS, which is using wearable and mobile technologies to monitor patients with depression and epilepsy, is also generating a wealth of data and fine-tuning analytical methods. Another, WEB RADR mines social media channels for adverse events data.

“The BD4BO programme is not just the five projects we have already launched,” said Chlebus. “It builds on other existing projects and tools. The beauty of it is that everything sits in one place and we can exploit work being done by other projects – that’s what makes it so exciting.”