Image: Caley Dewhurst (ODI)

Case study: Wellcome data science challenge

Thu Jan 28, 2021
$download_content = get_field('download_content');

Wellcome collaborated with the ODI to encourage innovation, and support more equitable funding by researching and designing a series of health data challenge prizes

Wellcome collaborated with the ODI to encourage innovation, and support more equitable funding by researching and designing a series of health data challenge prizes

Summary of the project

Artificial intelligence and data technologies are capable of transforming research and healthcare. Despite their importance, little funding is available for the foundational data and software tools used in health data research. Without the right tools, work with health data is slow, and only well-resourced institutions are able to make substantial progress. 

The Data for Science and Health team at Wellcome collaborated with the Open Data Institute (ODI) to encourage innovation, and support more equitable funding by researching and designing a series of health data challenge prizes.

A challenge prize is a financial reward to support innovative solutions to an otherwise unfixed problem. By funding open source, foundational data and software tools for science and health, with a focus on usability, adoption and long-term sustainability, Wellcome hopes to build public and patient trust, equip and motivate data scientists and create open and trustworthy data ecosystems. Data innovation can open opportunities for more cost-effective interventions, as well as providing entirely new insights that may have been overlooked through traditional approaches.

Effective health data innovation requires a diverse set of people from multiple disciplines to communicate and work effectively together: data scientists and research software engineers, clinicians and healthcare providers, biomedical researchers, patients and the public, policy and decision-makers. These groups often don’t have a shared understanding of problems and there are few opportunities for them to collaborate effectively. The ODI gathered insights on what health problems that could benefit from the challenge prize approach, and provided recommendations on how to design the programme.

Key facts and figures 

A shortlist of four topics were selected and researched to develop into challenge prizes:

  • Addressing anxiety and depression in young people and adolescents in the UK
    • 75% of mental illnesses start before the age of 18.
    • 70% of children and adolescents who experience mental health problems do not receive appropriate support.
  • Reducing deaths from snakebite in India
    • Snake bites kill between 81,000 and 138,000 people every year.
    • The World Health Organization aims to cut deaths and disabilities from snake bites by 50% by 2030.
  • Minimising antimicrobial resistance in Uganda, Malawi and Kenya
    • Drug-resistant diseases could cause 10 million deaths each year by 2050 if unaddressed.
  • Addressing public health risks in rapidly growing urban environments
    • Over 55% of the world’s population live in urban areas, expected to increase to 68% by 2050.
    • By 2030, two billion of the global urban population will live in slums, mostly in Africa and Asia.

What was the ODI’s role and what was the impact?

The Wellcome Trust is a politically and financially independent global charitable foundation, which supports science to solve urgent health challenges. Their strategy includes grant funding, advocacy campaigns and partnerships, and developing programmes of work that draw on Wellcome’s expertise across science, innovation and society to deliver ambitious goals.

Funded by Wellcome’s Data for Science and Health priority area, the ODI carried out a scoping project across the UK, sub-Saharan Africa and India to design a series of data science challenge prizes in the field of health. 

Progress in the field is inconsistent and hindered by three systematic blockers to the development of ethical, equitable technologies for healthcare, namely: there are too few data scientists working with health data for public good, trust in the use of data for health innovation is fragile, and it is very hard to get access to and use health data in combination with other data sources. 

One way to address some of these barriers is by offering a financial reward to support innovative solutions to an otherwise unfixed problem in the form of a challenge prize, motivating a wide pool of innovators to work together on a specific problem and create new solutions. This not only drives forward technological capacity in that field, it builds networks between communities of experts. 

A challenge prize is an open prize in which a financial reward is awarded to individuals and organisations to respond to a specific social or technical ‘challenge’ or problem which isn’t being solved. Challenge prizes can identify and draw attention to pressing health issues, allowing them to be meaningfully addressed using existing data, often combining non-traditional with traditional sources and using novel analytical methods to reframe issues. In addition, by putting meaningful uses of data in ways that help people and communities at the heart of the challenge, they can help develop broader data infrastructure in purposeful ways.

As part of the first phase of the project, a ‘theory of change’ was developed for the prizes, defining exactly what was being aimed for and how to achieve it.  These challenge prizes were not intended to create new datasets but to ensure the participants of the challenge can focus on what they can do with the data, not how to collect it. There is a huge amount of health data collected every day around the world, but it is very hard to get access to it, and very hard to use it in combination with other data sources. 

It was necessary that any prize responded to a clear and compelling problem that had been fully researched, including the legal and regulatory context of the problem and any stakeholders that ought to be identified and engaged. The problem needed to be broad enough to have space for a variety of approaches and solutions, but specific enough that measurable change could be seen through the work. Through extensive engagement with around 30 clinicians, researchers, data experts, patient groups and data scientists, as well as health practitioners across the UK, sub-Saharan Africa and India, the ODI identified a longlist of challenges. Topics included on the longlist but not selected for further work were:

  • Predicting risk for multimorbidities
  • The impact of air pollution on children’s environmental health
  • Diagnosis and treatment of neurodegenerative non communicable diseases, such as Alzheimers or Parkinson’s
  • Access to services in rural areas in order to support better maternal early child health

Each item on the longlist was scored by the ODI and Wellcome against a series of criteria, including the relevance and appeal of the problem, the availability of data, the scale of potential impact, how measurable the impact might be and any legal or ethical barriers. Four key topics were identified for the team to continue researching, and they were:

  • Data science for more effective mental health treatment for young people
  • Data science for reducing snakebite mortality
  • Data science for improving antibiotic prescription and use
  • Data science for public health risks in urban environments

More information about each of those topics can be found here.

We undertook in depth research into each of these topics, engaging with over 60 people globally, and identified the best prize models to deliver impactful innovations. This included an evaluation of the data landscape, the current state of scientific understanding in the field, identifying potential participants and what would influence them to take part: a good health challenge prize problem has a wide societal relevance so that practitioners from different disciplines will understand the problem and be motivated to help solve it.

The problems needed measurable solutions that enabled time-bound targets to measure the progress being made. Clearly demonstrating the value of the project can inspire interest from other funders and partners to invest and galvanise further work. It is important that these prizes have a longer-term impact, so the process built in time for improving data quality and access to data, and for taking any solutions that arise from the challenges into operation.

It was helpful to articulate the challenge in a single statement that could then be reviewed regularly throughout the process. Broad enough to iterate throughout the research process, the statements described the problem, its scale and then what the challenge prize would do to solve it without predetermining what specific solutions might look like. 

As part of this project, a health data challenge prize playbook has been developed and published by the ODI. It is primarily aimed at funders and people working with funders, or other organisations who are looking to solve problem-specific societal challenges related to health by leveraging data to stimulate innovation.

The playbook provides guidance on how to identify a health problem and assess whether it fits within the model of a challenge prize. It also provides a step by step guide on how to scope out the problem in more detail and design the challenge prize around it.

Who else was involved?

To understand the types of opportunities for using data science in healthcare, the ODI looked to the work of Understanding Patient Data who provide a framework for thinking about the impacts of data in healthcare, supporting individual care, understanding of disease, policy evaluation, planning health services, diagnosis, treatment and prevention, and patient safety.

What was challenging?

Scoping a global health project within the context of Covid-19. The project was around halfway through when the UK, and much of the world, went into lockdown. While we still managed to engage with a large number of stakeholders, it did mean we had to tailor that engagement to be suited to a remote format. It also added quite a high level of uncertainty around the viability for the programme and recommendations being proposed.

Prioritising the longlist of topics to further research. The initial phase of the project for identifying a longlist health topics was quite short (3 months) so only allowed us time to undertake a fairly high level of research. This did make the selection of the final chosen topics quite difficult; we could have potentially narrowed that down earlier and built in different criteria to allow us to do that. 

Getting up to speed with new topics quickly. While the ODI has done work in the health sector before, some of the topics we were looking at were very new to the team and required a lot of in depth research within a relatively short space of time. As well as desk based research, speaking to a wider range of experts, both within Wellcome and externally, really helped us to get a good depth of understanding. 

What went well?

It gave us the opportunity to use, test and refine some of our existing processes and tools. Undertaking in depth research into the shortlisted challenge prize topics allowed us to reuse and develop some tools and methodologies around assessing the data landscape in specific areas, such as creating data ecosystem maps and data inventories. It also allowed us to consolidate that guidance into the Data Challenge Prizes for Health Playbook to help others looking to undertake similar research. We’ll also be looking to iterate on this playbook through other ODI projects. 

We have broadened our expertise and network within the health sector. Undertaking research into the long and shortlisted topics gave us a much greater understanding of the challenges around data in the health sector, and how they might be addressed, which is a key focus for the ODI. We engaged with many experts in the sector, which was extremely helpful for this project and future work.

What have we learned

Always be prepared to adapt plans and be flexible. The Covid-19 pandemic put a stop to our ability to speak to people through in person interviews. We had to quickly think about how we could effectively connect with a large number of people globally on the topics we were researching, whilst the priorities within the health sector were completely changing. We still engaged with many people 1:1 through interviews, but undertook less workshops than we would have liked to. 

Try to create stories that people will relate to. Towards the end of the project, we worked with our comms team to develop tangible stories around the topics to strengthen the rationale for undertaking a challenge and to encourage buy in from stakeholders. We found these to be really useful and think they’re an important element to build in as early as possible.

Timebox research where you can. The topics we were researching were all quite large and It was challenging at times to know whether we were achieving the right breadth and depth to inform the recommendations. It’s important to remember that there is always more you could learn, especially in early stage scoping work. Timeboxing allows you to set the parameters for the level of knowledge you need to get to and build in the time to consolidate your findings.