Expert opinion Julien Rutard, Vice President, Healthcare Sector (Capgemini Invent) - Data in the hospital environment

In January, Future4care launched the January4data campaign to highlight the essential role played by healthcare data in the value chain. To this end, we solicited the expertise of members of our ecosystem, which has given rise to these notes. The genesis of healthcare data, technological developments and uses, ethics and applications in the hospital sector are just some of the topics covered in this series of articles.  

I'm convinced that hospitals in France play a crucial role, built around three fundamental pillars: care, teaching and research. It's a complex mission, and the rapid evolution of technology, particularly in the field of artificial intelligence (AI), is profoundly transforming these activities. Today, we are witnessing a veritable explosion of data and technological capabilities, creating immense possibilities but also raising questions of interoperability, quality and security.

The integration of data into hospital structures is nothing new. For over fifteen years, biobanks and associated databases have played a major role in medical research. However, we now understand that simply accumulating biological data, without enriching it with contextual information (imaging, therapeutic strategies, pathological trends), is insufficient. The richness of data lies in its quality, granularity and interconnection. Clearly, without precise structuring and a coordinated effort to improve their quality, these data cannot fully meet the needs of modern healthcare.

Unprecedented technological evolution

Advances in computing and storage capacities, notably with the emergence of quantum computing, are revolutionizing the way we approach data processing. This technological evolution makes it possible to process massive volumes of information with greater precision and speed. For example, modern AI models can analyze data in seconds that would take months to process manually. But what does computing power matter if the data processed is incomplete or of poor quality? This question illustrates one of the main challenges we face: data interoperability and relevance.

Take medical imaging, for example. The widespread use of MRIs and scanners, combined with ever more powerful capture tools, has democratized access to massive volumes of data. However, without a rigorous labeling process, an MRI image can only be partially exploited. It is the human expert who must, for example, identify and delimit a tumor to enable the AI to train accurately. This crucial step, known as labeling, is often underestimated, but is the foundation of any reliable analysis.

Increased storage capabilities also enable a complete patient history to be maintained, including genomic data, medical history and interactions with different treatments. This paves the way for unprecedented personalization of care, but only if this data can be exploited and interconnected between different healthcare systems.

The challenges of data quality and security

Beyond simple acquisition, there are challenges linked to data management and processing. Biases in acquisition mechanisms, artifacts and extrapolation problems can all affect data quality. Take a concrete example: data from the same patient may vary according to the machines used or the protocols applied, introducing differences likely to bias analysis results.

What's more, interoperability between different hospital systems remains a major challenge. For example, the same patient being treated in several different facilities can generate data that is scattered and difficult to correlate. These shortcomings hamper a global and coherent vision of the care provided.

Another major challenge is data security. In a society oscillating between trust and mistrust, it is vital to strike a balance. Too many restrictions hinder access to data, even for noble objectives such as research or public policy. I've been confronted with this complexity myself: the delays and constraints imposed by systems such as CASD (Centre d'accès sécurisé aux données = Secured data access center) are often a deterrent. Yet these restrictions do not always prevent data leaks, often caused by trivial human error.

Furthermore, examples of medical data leaks remind us of the importance of re-evaluating our systems. Some recent leaks have highlighted flaws in internal access management, underlining the need for stronger auditing and accountability protocols.

The role of ethics in healthcare data management

Ethics play a central role in our approach to health data. It guides us when the law is silent. But too much control can have the opposite effect, fueling conspiracy and holding back projects. We need to ask ourselves: is it ethical not to fully exploit available technologies to improve public health?

The ethical framework is also crucial for regulating the use of AI in analysis processes. Take the case of machine learning algorithms: how can we ensure that they do not introduce systemic biases? Ethics must also address the dilemmas linked to the balance between innovation and personal data protection.

Conclusion

Data is at the heart of the evolution of our healthcare system. If we are to harness its full potential, we need to rebalance our approaches between openness and protection, while ensuring that strong ethics are maintained. It's time to meet the organizational, technological and ethical challenges to build a future where healthcare is accessible, innovative and secure.