Clara MoreauChief Data Officer · Essays & perspective
Chief Data Officer / Europe
Clara Moreau
Open toBoard appointments · Speaking invitations
Imaging AI5 min read
Patient records need a purpose for every use
A CE mark and a vendor's study show that an imaging algorithm can work somewhere. Only a hospital's own scanners, patients and habits show whether it works there, and the cost of finding out belongs in the purchase rather than in the data team's spare time.
Hospitals have long shared research data by sending pseudonymised extracts to their partners. I argue that the extract should become the exception: partners bring their analysis to a controlled environment, and only checked results leave it.
Sponsors increasingly ask hospitals to describe their data as research-grade. No such general label is honest. Fitness depends on the question, and a hospital should answer each request with a short account of what its records can and cannot show for that purpose.
A hospital record is written to treat one patient; before it answers anyone else's question, I have to show it is fit for that question and safe to use.
I am the chief data officer of Fondation Hospitalière Aurevalle, a fictional private non-profit hospital group based in Lille. We run six hospitals and four imaging centres across the Hauts-de-France region, with 2,150 beds and 11,400 staff, and in 2025 we recorded €1.24 billion of operating revenue, most of it paid by France's national health insurance. Because the group takes part in the public hospital service, our records describe a whole regional population rather than a chosen slice of it.
My job is to decide how those records are kept, who may use them beyond the care of the patient, and on what terms. I came to it from the other side of the table. I spent ten years at a contract research organisation in Paris, in trial statistics and data management, and four and a half years in Copenhagen leading real-world evidence for a research-based pharmaceutical company. I have asked hospitals for their data far more often than I have been asked for ours, and I remember which of those requests I should not have made.
Since I joined Aurevalle in March 2021 we have opened a clinical data warehouse covering 3.1 million patients, stopped sending row-level extracts to research partners and begun testing every imaging algorithm on our own films before a clinician sees its output. Each decision cost something: research income, a regional grant payment, the patience of colleagues who wanted results sooner. I would take all three again because a hospital's permission to reuse patient data rests on patients believing that we use it carefully.
Analysis brought to the data
Since April 2024 external researchers work inside our analysis environment, and only checked results leave it. External projects rose from 21 extracts in 2022 to 38 projects in 2025, and partnership income reached €2.6 million, after one sponsor walked away from a €1.4 million study.
An imaging algorithm tested before it was trusted
Four months of silent running on 26,400 chest films showed a CE-marked triage tool performing as claimed on upright adult films but missing almost a third of pneumothoraces on portable intensive-care films. We kept it out of intensive care.
From request to data in eleven weeks
The median wait for approved research access fell from seven and a half months in 2021 to eleven weeks in 2025, through one data-access committee, standard agreements and a written fitness note for every request.
Patients told, and their objections honoured
Every project using our records appears on a public register. 3,140 patients have objected to research reuse since 2023, and their records drop out of every query from that day, including studies already approved.
Current focus
I am preparing the group for Europe's new framework for secondary use of health data, which will route many research requests through national access bodies, and deciding which of our own procedures to keep. I am also writing the monitoring terms for our next two imaging tenders, so that a supplier's duty to show performance at each site sits in the contract rather than in my budget.
Conversations I welcome
I am not looking to leave Aurevalle. I would consider one non-executive seat at a European healthcare company that processes patient data at scale, such as a diagnostics network, a health mutual or a digital-health business, joining its audit and risk committee and, in time, chairing a technology and data committee. I can give about twenty-two days a year, with my board's consent and after a conflicts review, and I accept a few invitations to speak to hospital, research-ethics and board audiences.
Beyond the page
The record behind the perspective.
In the boardroom
Association BelvérineNon-executive director of a non-profit regional cancer registry · 2022–present
Formation
Engineering diploma in statistics and economicsInstitut Varennier de statistique appliquée · 1999
Master's degree in epidemiology and biostatisticsInstitut Marsaguet de santé publique · 2000
Clara brought us the decision to stop sending extracts with the lost income already in the paper, and the name of the sponsor most likely to leave. She did not ask us to admire the environment she wanted to build. She asked us to accept its cost, and she was right about both.
Étienne VarlaumontFictional directeur général, Fondation Hospitalière AurevalleThe full record ↗
Contact
I am not looking to leave Aurevalle. I would consider one non-executive seat at a European healthcare company that processes patient data at scale, such as a diagnostics network, a health mutual or a digital-health business, joining its audit and risk committee and, in time, chairing a technology and data committee. I can give about twenty-two days a year, with my board's consent and after a conflicts review, and I accept a few invitations to speak to hospital, research-ethics and board audiences.
