Mar 8, 2021
How Sweden uses data and AI to improve the health of citizens
Since the COVID-19 pandemic hit Sweden and elsewhere, health data has naturally been essential in helping public authorities and researchers find the best ways to respond and recover. The fact that public healthcare providers are able to collect, record, store, share, open up, and analyse necessary data with its current speed is a result of dedicated policies, technologies, processes, and people.
In Sweden today, we can see the effects of such policies — health-related statistics are available as open data and an ecosystem of government- and non-governmental actors are joining forces to build “information driven care” with data and AI. Below is a brief summary of some of these interesting, innovative initiatives.
Open data from Statistics Sweden
Open data is an important stage of the data value cycle. By ensuring that health-related information and data are available in high-quality, open formats, it allows public authorities and researchers, as well as citizens, to re-use them more easily. Statistics Sweden has made it its strategic objective to publish as much official statistics as possible as open data[1]. As part of this objective, official monthly statistics on reported deaths are published on the Statistics Sweden website and the Swedish national data portal dataportal.se. All open data from the agency can be accessed via Application Programming Interfaces (APIs), without the need to register.
It does not end here. The agency has created a live data dashboard that helps citizens, residents, researchers and organisations to more easily interpret the data. One of the indicators of the dashboard is the number of deaths per region and month in 2020 compared to the average in 2015–2019, which can help us better understand the impact of the COVID-19 pandemic on mortality.
A data warehouse and federated mortality prediction in Halland
Gathering meaningful insights from data to improve the planning and delivery of healthcare services requires disparate data sources and large databases to be integrated and combined (e.g. clinical and financial data).
In 2020, the Region of Halland, Sweden’s seventh largest region with around 300 000 inhabitants, constructed a comprehensive data warehouse[2] as part of a project called Leap for Life to collect live data from across the entire regional healthcare system — including hospitals, primary and ambulatory care, as well as data from private providers. These data include Electronic Health Records (EHR), lab and radiology data, financial data and HR data.
By integrating these large, disparate sources of data into one single repository, the data warehouse bridges data- and knowledge silos, and has increased the capacity for the regional healthcare provider to produce algorithms and monitor the healthcare system using standard Business Intelligence tools and AI. As a result, healthcare providers and researchers[3] can more easily analyse patient pathways, identify trends and predict healthcare capacity, such as Incentive Care Units (ICU) — which is crucial during a pandemic.
To further make better use of data, Region Halland is developing federated learning capabilities together with Halmstad University. As a testbed project under the umbrella Federated Learning[4] initiative co-ordinated by the Swedish National Center for applied Artificial Intelligence (AI Sweden), they have developed federated mortality prediction, devising an algorithm predicting emergency care patients’ survival rate up to 30 days after they have visited the emergency care unit. This allows them to better follow-up and prevent patients from unnecessary illness and cost savings for the hospital. The region has also begun applying natural language processing in its work to enhance patient safety.
Government support to AI- and information driven care
Support from the central government in combination with a well-developed and diverse ecosystem of actors is important for increasing the use of data and AI in the design and delivery of public care services.
As part of the national programme Vision-driven Healthcare run by Sweden’s government innovation agency Vinnova, the agency is funding a 5-year project to implement information driven care in Sweden using AI. Co-ordinated by AI Sweden together with the Swedish Association of Local Authorities and Regions, Region Halland, Halmstad University and Karolinska University Hospital (one of the largest University Hospitals in Europe), it aims to identify, stimulate, promote and scale up successful AI-applications within the Swedish public healthcare by establishing a dedicated innovation environment. It all takes place in close partnership among public, private and academic parties. The data warehouse in Halland is part of this project.
Vinnova is also funding other relevant projects. In 2020, the agency launched an open call for tender to build Vinter — a digital platform that supports secure sharing and use of sensitive data about diabetes. The aim is for the integration of these data to support innovation in generating services valuable for people with diabetes, their families and care providers. The first calls to innovate on the platform is set to begin in 2021.
Sweden’s path towards a data-driven public sector
Since 2019, when the OECD published its Digital Government Review of Sweden and the results from the latest OECD Digital Government Index and OURdata Index were launched, the Swedish government has done a lot to improve the application of data within the public sector. Policies for better governance of data across the entire Swedish public sector will be critical in supporting ongoing and future sector-specific data initiatives, including those related to health. Below is a summary of relevant policies and initiatives:
- Creation of the Swedish Agency for Digital Government — DIGG
- Launch of the new Swedish framework for digital cooperation
- Development of a government digital infrastructure for information exchange
- Development of a national framework for base data
- Launch of national principles for making information (data) available
- Launch of a recommendation on open licenses and intellectual property
- Launch of the new national data portal dataportal.se
- Draft national open data action plan
For more information about Sweden’s work on data-driven care and data-driven public sector, visit Vinnova, DIGG, Statistics Sweden and AI Sweden. For more of the OECD’s work on data-driven public sector and digital government, visit oecd.org/gov/digital-government.
[1]https://www.scb.se/contentassets/15740059c3c74fde9310ff95126a9f21/malbild-sveriges-officiella-statistik_en.pdf
[2] A data warehouse is a central repository of integrated data from various disparate sources
[3] After standard Institutional Review Board (IRB) processes
[4] Federated learning is a new way of machine learning with the potential to allow use of sensitive data while complying with legal requirements.
