Reusing healthcare data to develop value-added tools for radiology
Stéphane Roche, Miguel Guevara, Ludovic de Rochefort , Alexandre Vignaud
Journées Francophones de Radiologie, 4-7 octobre, Paris, France
Abstract :
Objectif
Health data, particularly imaging data, is at the heart of the healthcare system and of cohort research projects. Their secondary re-use, although a source of innovation, is subject to specific regulations. With the development of artificial intelligence tools and specific areas for access to this sensitive data within health data warehouses, secondary use is particularly conducive to speeding up the marking of digital medical devices, particularly imaging devices. During this talk, we will contextualise the example of quantitative magnetic susceptibility (QSM) imaging.
Patients and Methods
Open data from the National Health Data System (SNDS) and the Centre for Epidemiology of Medical Causes of Death (CépiDc) were used for the epidemiological, co-morbidity and socio-economic study. Imaging data for the study of normal ageing came from the SENIOR cohort.
Results
The number of Alzheimer’s-type dementia patients will remain stable at around 713,050 in 2021 (43.4% of all dementias), with a total prevalence of 1.04% in 2021. Alzheimer’s disease accounts for almost 18.2% of deaths in France each year. Associated with this disease, these patients have also been diagnosed with another dementia, generally Parkinson’s disease (13.01%), the after-effects of a stroke (respectively 12.21%), cardiovascular disorders such as rhythm disorders (21.17%), chronic coronary artery disease (12.33%) and chronic heart failure (9.61%). These patients are also being monitored for mood disorders (12.5%) and are receiving antidepressants (24.57%) and anxiolytics (21.57%).
Using imaging and health data from the SENIOR cohort studying normal ageing shows that certain pathophysiological risks are similar, particularly cardiovascular risks (24.05%). Depression was also present in 6.3% of subjects, and particularly in women (80%). We controlled for the absence of anaemia in the elderly in order to exclude bias in intracerebral iron measurements. We demonstrated that cardiovascular risk had an influence on the measurement of intracerebral iron in the globus pallidus and the putamen using the magnetic susceptibility quantification (MSQ) technique, even though the subjects had no symptoms of neurodegenerative disease.
Conclusion
The combination of the reuse of population health data and the prospective constitution of a cohort comprising the health data of study subjects and relevant imaging data, combined with robust, uniform and centralised QSM post-processing, has the potential to accelerate the development of an innovative medical device making the most of cohort data including imaging.