Beyond Bilbao: Patients Must Help Shape Health AI

Pedro Carrascal on patient-generated data, meaningful participation and why trust begins with shared governance.

What should health artificial intelligence help us improve? Clinical outcomes matter, but so do the everyday realities of living with illness: maintaining independence, managing fatigue, sustaining relationships and finding a treatment that fits a person's life. Understanding those priorities requires patients to help define the questions from the beginning.

In our recent Health Data Forum conversation, Pedro Carrascal of Spain's Plataforma de Organizaciones de Pacientes (POP) made a clear case for patients to become active partners in health data ecosystems. Their contribution must extend across the collection of data, the governance of its use and the evaluation of the solutions it enables.

As we continue the dialogue beyond Bilbao, his message offers a practical direction: build participation into the way health data and AI are developed.

Measuring what matters to patients

Carrascal described POP's contribution of a first data catalogue to Spain's national health data space. The initial contribution is limited in scale, he explained, but its strategic significance is substantial: it gives practical expression to the role of patient organisations as data providers.

The catalogue draws on work around chronic illness and wellbeing. Behind it lies a broader effort to define unmet needs through the experience of patients and their associations.

This matters because the information available to a health system influences the questions it can answer. When important aspects of life with illness remain unmeasured, they can also remain absent from research priorities and innovation decisions.

During the conversation, Carrascal distinguished between patient-reported outcome measures, or PROMs, and patient-reported experience measures, or PREMs. PROMs capture outcomes as reported by patients, including aspects of health and quality of life. PREMs focus on their experience of care and services. Both contribute valuable knowledge, and each answers a different set of questions.

For Carrascal, patients should also help shape those questions and the measures used to explore them. Their knowledge can reveal needs that existing datasets were never designed to capture.

Participation needs a permanent structure

Meaningful participation requires a place in the decisions that shape a data ecosystem.

Who determines which data should be collected? Who identifies the needs that deserve attention? Who sets priorities for investment? And who assesses whether the resulting work has delivered value?

Carrascal argued for patient involvement throughout these processes, with complementary roles for individuals, disease-specific associations and umbrella organisations such as POP. Each brings a different perspective and capacity to contribute.

He described POP's participation model as bringing together four dimensions: mechanisms for participation, objectives, metrics and an organisational structure. Consultation alone cannot sustain that work. Participation needs clear responsibilities and arrangements that endure.

For organisations developing health data initiatives, this is a useful challenge. Inviting patients into a discussion is a starting point. Giving them a defined role in shaping decisions makes that involvement consequential.

Trust creates the conditions for better data

Asked about safeguards around the secondary use of health data, Carrascal emphasised transparency, oversight and people's ability to understand how their data is being used.

He presented these safeguards as conditions that allow health data ecosystems to grow. People need confidence in the purposes, protections and governance behind requests to contribute their information.

That becomes especially relevant as health data extends further into everyday life. Clinical records provide essential knowledge, while patients can contribute additional understanding of how illness and treatment affect them outside a healthcare setting.

Carrascal sees considerable potential in bringing these perspectives together. Realising it depends on communication and governance that patients can trust, alongside a clear account of the benefits the work seeks to deliver.

Patients must help evaluate AI

The same principle applies to artificial intelligence.

In Carrascal's account, patients have a role in supplying relevant data, helping define what an AI solution should address and evaluating whether its results meet their needs. Professionals remain essential to that evaluation, and patients contribute knowledge about the impact on their lives.

This gives health AI a concrete test of purpose. A technically impressive result still needs to demonstrate useful change: better decisions, more appropriate care or improvements that people can recognise in their daily lives.

The conversation also touched on personalised medicine, genomics and biomarkers. Carrascal saw an opportunity to connect those advances with patient-reported needs and real-world experience, while recognising the challenges of financing, access and equity.

Personalisation becomes more meaningful when the person's priorities help guide the use of increasingly detailed biological information.

Taking the dialogue beyond Bilbao

Carrascal's contribution gives the next stage of the health data conversation a clear focus. Patient participation can improve the relevance of the data we collect, the legitimacy of the decisions we make and the usefulness of the technologies we develop.

For the Health Data Forum community, the practical question is how to make that participation part of everyday work. Research programmes, data initiatives and AI projects can begin by defining where patients influence priorities, how their contributions shape decisions and how they take part in evaluating results.

Bilbao brought people together around the future of health data. Beyond Bilbao, the work continues through those choices.

Patients must have the opportunity to help shape that future—and to judge whether it is improving their lives.