PreciDIAB

Pr. Jean-Baptiste Beuscart

University professor – Hospital practitioner in the acute geriatric medicine department, specialist in the course of care, drug interactions and artificial intelligence.

For this new issue, Prof. Jean-Baptise Beuscart presents his work in the field of artificial intelligence and e-health in order to better understand the objectives attached to these technological applications as well as the proposal of the National Center PreciDIAB to answer to these new challenges.

“Can you tell us about your professional background and what made you jointly interested in drug interactions and artificial intelligence? ”

I am a geriatrician with nephrological training. The older people I care for often have drug-related side effects, and kidney function is central. In addition, I did a thesis in statistics applied to multistate models, dedicated to situations where a patient presents several possible states over time.

These models make it possible to describe and analyze care paths, a theme that is also central in geriatric settings. Finally, I was lucky enough to be able to participate in ambitious large-scale projects on (1) the reuse of data and algorithms applied to drug risk (European PSIP project) and (2) the care pathway for subjects elderly (national project PAERPA). These trainings and experiences have convinced me that we are reaching a point where the quantity of available data, the quality of the mathematical tools, and the computing power of the machines offer extraordinary possibilities in terms of research and structural aid for the benefit of patients and caregivers.

“What is e-health and how can this area change the care and care of people? ”

“E-health” includes all the applications of information and telecommunications technologies in the service of health. It is therefore a very broad field, which includes in particular the use of digitized health data. As such, it is necessary to differentiate the reuse of health data (data reuse), which corresponds to analyzing data by changing the purpose of this data (for example, using billing data to describe care paths), from artificial intelligence, which is one of the many methods of analyzing this data.

“E-health” is therefore already present on a daily basis in hospitals and outpatient clinics, if only through computerized prescriptions and software specific to our professions, but the reuse of digital data in real time is just around the corner. his beginnings. The potentials are enormous: improvement of prediction scores, detection of risky situations in real time, optimization of care pathways, security, information and communication, prescription assistance, and many other applications!

However, the challenges are just as important because it is necessary to ensure a data processing respectful of the General Data Protection Regulation (GDPR), to control the quality of this data, to ensure their interoperability, etc. Above all, special attention must be paid to the processing of the information produced: it must be integrated into the existing information circuit and be addressed to all the interlocutors concerned, from a multi-professional perspective and involvement of the patient himself.

“What is the proposal of the National Center PreciDIAB to respond to these new challenges? ”

Part of the PreciDIAB program aims to optimize anti-diabetic treatments during hospitalization. With my team and other academic and industrial collaborators, we will develop strategies for the prevention of iatrogenism (ie adverse effects caused by taking one or more drugs) using computerized decision support systems in respecting two key stages: (1) design of dynamic, reliable and specific rules; (2) a thoughtful and concerted integration of these alerts into the multi-professional care process, adapted to the care context.

We will also develop pre-screening tools that can take advantage of the rich but heterogeneous data in the Lille University Hospital data warehouse. These tools will be based on first level artificial intelligence (terminological alignment, automated annotation, etc.) and second level (supervised and unsupervised classification). Ultimately, the development of this project will make it possible to better detect and correct situations at iatrogenic risk and / or requiring therapeutic optimization in hospitalized diabetic patients, in conjunction with the outpatient sector and the patient.

In addition, a detailed analysis of the data concerning diabetic patients hospitalized at the Lille University Hospital will make it possible to determine the recruitment capacities of diabetic patients at the Lille University Hospital, to carry out pre-screening for inclusion in clinical trials, and to clarify the scope of some clinical questions.

These essential functions will be able to meet the expectations of institutional and industrial research.