Open data in health and education as high-value data?

When we think of open data our first intuition is usually directed towards data generated by public sector bodies in the exercise of their functions and made available for reuse by citizens and businesses, i.e. public sector open data or open public data. This is natural, because public sector information represents an extraordinary source of data and the intelligent use of this data, including its processing through artificial intelligence applications, has great transformative potential in all sectors of the economy, as recognised by the European directive on open data and re-use of public sector information.

One of the most interesting novelties introduced by the directive was the initial but expandable definition of 6 thematic categories of high-value datasets, whose re-use is associated with considerable benefits for society, the environment and the economy. These six areas - Geospatial, Earth Observation and Environment, Meteorology, Statistics, Societies and Corporate Ownership and Mobility - are the ones that in 2019 were considered to have the greatest potential for the creation of value-added services and applications based on such datasets. However, looking ahead to 2021, which is almost a year into the global health crisis, it seems clear that this list misses two key areas with a high potential impact on society, namely health and education.

Indeed, we find that on the one hand, educational institutions are explicitly exempted from some obligations in the directive, and on the other hand, health sector data are hardly mentioned at all. The directive, therefore, does not provide for a development of these two areas that the circumstances of the covid-19 pandemic have brought to the forefront of society's priorities.

The availability of health and education data

Although health systems, both public and private, generate and store an enormous amount of valuable data in people's medical records, the availability of these data is very limited due to the very high complexity of processing them in a secure way. Health-related datasets are usually only available to the entity that generates them, despite the great value that their release could have for the advancement of scientific research.

The same could be said for data generated by student interaction with educational platforms, which is also generally not available as open data. As in the health sector, these datasets are usually only available to their owners, for whom they are a valuable asset for the improvement of the platforms, which is only a small part of their potential value to society.

The directive states that high-value data should be published in open formats that can be freely used, re-used and shared by anyone for any purpose. Furthermore, in order to ensure maximum impact and facilitate re-use, high-value datasets should be made available for re-use with very few legal restrictions and at no cost.

Health data are highly sensitive to the privacy of individuals, so the delicate trade-off between respect for privacy and the need to support the advancement of scientific research must always be kept in mind. The consideration of health and education data as high-value open data should probably maintain some particular restrictions due to the nature and sensitivity of these data and promote figures such as the donation of data for research purposes by patients or the exchange for the same purpose between researchers. In this sense, the 2018 regulation on data protection introduced the possibility of reusing data for research purposes, provided that the appropriate pseudonymisation measures and the rest of the legally stipulated guarantees are adopted.

The importance of public-private partnerships

Education and health are two areas where the private sector or public-private partnerships are making exciting strides in converting some of the potential of open data into benefits for society. Open data publishing is not the exclusive preserve of the public sector and there is a long tradition of private-public collaboration, largely channelled through universities. Let us look at some examples:

• There are a number of initiatives such as the pioneering The UCI Machine Learning Repository founded in 1987 as a repository of datasets used by the artificial intelligence community for empirical analysis of machine learning algorithms. This repository has been cited more than 1000 times, the highest number of citations obtained in the computer science domain. In this and other repositories also managed by universities or foundations with donations from private companies, we can also find open datasets released by companies or in which they have actively collaborated in their creation or development.
• Also large technology companies, no doubt inspired by these initiatives, maintain open data search engines or repositories such as Google's dataset search engine, AWS's open data registry, or Microsoft Azure's datasets, where datasets related to health or education are increasingly common.
• In terms of data that can contribute to improving education, for example, The Open University publishes OULAD (OpenUniversity Learning Analytics Dataset), an open learning analytics dataset containing data on courses, students and their interactions with the virtual learning environment for seven courses. However, there are very few comparable datasets whose joint use in projects would undoubtedly allow further progress to be made in areas such as detecting the risk of students dropping out.
• As far as the health sector is concerned, it is worth highlighting the case of the Spanish platform HealthData 29, developed by Fundación 29, which aims to create the necessary infrastructure to make it possible to securely publish open health datasets so that they are available to the community for research purposes. As part of this infrastructure, Foundation 29 has published the Health Data Playbook, which is a guide for the creation, within the current technical and legal framework, of a public repository of data from health systems, so that they can be used in medical research. Microsoft has collaborated in the preparation of this guide as a technological partner and Garrigues as a legal partner, and it is aimed at organisations that carry out health research.

At the moment the platform only has available the Covid Data Save Lives (COVIDDSL) dataset published by the HM Hospitales University Hospital Group, composed of clinical data on interactions recorded in the covid-19 treatment process. However, it is an excellent example of the potential that we may be missing out on globally by not collecting and publishing more and better data on patients diagnosed with covid-19 in a systematised way and on a global scale. The creation of predictive models of disease progression in patients, the development of epidemiological models on the spread of the virus, or the extraction of knowledge on the behaviour of the virus for vaccine development are just some of the use cases that would benefit from greater availability of this data.

Education and health are two of the great concerns of all developed societies in the world because they are closely related to the well-being of their citizens. But perhaps we have never been more aware of this than in the last year and this represents an extraordinary opportunity to drive initiatives that contribute to unlocking more open health and education data. Whether as high-value data or in any other form, these datasets are key to enabling us to better react to future health crisis situations but also to help us overcome the aftermath of the current one.

Content prepared by Jose Luis Marín, Senior Consultant in Data, Strategy, Innovation & Digitalization.

The contents and points of view reflected in this publication are the sole responsibility of its author.