EIT Wild Card

BLOG: Making European healthcare smarter

The IDC has estimated that by 2025, the amount of data generated by humans could reach 163 zettabytes, or 163 trillion gigabytes. The amount of available data is growing exponentially, and by 2025, it is estimated that 90% of this data will have been generated in the preceding two years alone.

In addition to computers and smartphones, this data is increasingly being generated by wearables and other Internet of Things devices, providing a wide variety of data sources to draw upon. Networked medical devices are playing an increasingly important role in allowing patients and their healthcare providers to monitor, send and store biometric data in real time, 24 hours a day, 365 days a year.

This automated round-the-clock insight into patient wellbeing opens the door to a new realm of personalised medicine. On the level of individual patients, data from these devices may allow healthcare providers to tailor interventions to specific patients, or improve compliance with existing treatments. The vast datasets generated by these devices and other big data sets also hold the promise of enabling improved models for the early detection of conditions which traditional automated screening tests have struggled to characterize.
On a broader scale, big data gives health services the ability to develop accurate predictive models, allowing them to assess trends in populations, identify at-risk demographics, and even when and where to invest in new technology, staff and infrastructure. This has the potential to not only improve patient outcomes, but also reduce healthcare costs by developing more effective care processes.

The problem

Big data holds a mass of valuable information for healthcare services, hospitals and primary care providers; but storing, exchanging, and making sense of it all presents major challenges. The variety of actors in the healthcare data space has led to a “data silo” effect, limiting systems interoperability. The lack of widely adopted healthcare data standards makes the international collection of large, structured datasets a major challenge.

The scale of these data poses a challenge as well. While traditional structured data sources such as electronic medical records are modest in size, valuable unstructured data sources including images, audio files and video pose a greater technical challenge – a single 3D CT scan can be as large as 1 gigabyte! Processing these large files en masse requires the application of sophisticated big data architectures and analysis methods.

These unstructured data also require the development of machine learning and other artificial intelligence techniques to be effectively analysed. While robust machine learning methods have been developed and applied to a number of fields, healthcare has been slow to adapt them for use in clinical settings.

Most hospitals and health services are burdened with day-to-day operational challenges which leave little time for implementing ambitious technological projects. These tend to be low priorities as they are perceived as complicated and time-consuming with many complications relating to costs and patient privacy. As such, solutions need to be practical, goal-oriented and relatively straightforward.

Wild Card: The search for smarter healthcare

The EIT Health Wild Card programme is about finding the best and brightest minds to tackle some of Europe’s most pressing public health challenges.

For the SMART HEALTH challenge, we’re seeking entrepreneurs to develop novel methods to obtain, interpret and apply big data and AI techniques to move from limited analysis to build and structure bigger datasets to facilitate its use in diagnostics. The proposals should ideally be within existing regulatory frameworks, cross border, at scale and fast paced. Interested applicants should consider these questions before submitting their entry:

  • How can we use AI in a practical way? Hospitals and health services must deliver optimal service with highly scrutinised budgets and in tightly regulated environments. How can EU health services roll out an AI solution that will demonstrably reduce costs and improve patient outcomes?
  • Can the solution be implemented on a pan-EU scale? GP clinics, hospitals and health services operate differently from country to country. An optimal AI/big data solution must be simple and robust enough to implement across European member states within existing regulatory frameworks.
  • How can we make sure patient data is secure? Personal health data is extremely sensitive, and the General Data Protection Regulations set to go into effect in the EU in 2018 stress the need to handle these data in a responsible manner. How can patients and other stakeholders be assured that their data is being kept safe and only being accessed appropriately?

Interested in taking part? Learn more about the SMART HEALTH challenge.

Douglas Spangler
Uppsala University, Sweden

Douglas Spangler is a health services researcher at the Department of Public Health at Uppsala University and a project manager at the Uppsala University Hospital. Currently, Douglas is leading an effort to develop and implement machine learning models to predict outcomes for 911 callers, with the goal of safely enabling a diversified response to low-priority calls.

Uppsala University is a core Partner of EIT Health.