Data is critical for Artificial Intelligence (AI) and Machine Learning (ML) applications in healthcare. Their quantity and quality directly impact the accuracy and inclusiveness of AI-based solutions, determining if predictions can be generalized to different populations, especially for underrepresented subjects. Thus, developers need to be aware of their limitations and find ways to overcome the hurdles they generate.
Trust and trustworthiness are the foundational aspects concerning the sociotechnical factors to working with health data at scale. Trust comprises a multifactorial myriad of elements, including purpose, governance, and ethical and legal considerations. Many recent policy documents, such as the AI Act or the Data Act, have approached it.
Anonymized and aggregated data
However, many stakeholders, such as social care organizations that contribute to health provision and integrated care services across Europe, are still mostly unaware of how the European Health Data Space can work and how their contribution can be mutually beneficial. And how can they trust something they don’t know? Most data-sharing activities, foreseen to maximize the use of health data for innovation, do not require access to individual data – at least not at the patient level. Instead, anonymized and aggregated data are needed. Therefore, privacy and security challenges may not be as difficult to overcome as they seem at first sight. It means that more clarifications are required to define the requirements for different specific scenarios, as well as the professional shared responsibilities on anonymizing data at the organisations’ level to ensure all will work well. Also, within this same scope, common interoperability standards play an essential role in addressing some challenges around data quality, trustworthiness, and the potential for generalization.Not using data can harm patientsIt is still acknowledged that most citizens need further information regarding data storage, access and sharing, as well as better digital skills. This requires professionals and policymakers in this area to simplify language and preferably use materials with visual information that makes it more accessible to people in different contexts, literacy, and education levels. During the COVID-19 pandemic, the use of some digital solutions, such as the COVID certificate or tracking tools, may have opened the way for people to better understand the potential benefits of such tools and thus be an enabler for the future. Also, many risks, for example, cybersecurity, were better acknowledged as extremely important to address.