DS0804 - Systèmes informatiques et numériques, phénotypage, organismes et pathologies virtuelles, Recherche méthodologique, informatique et statistique pour répondre aux défis conceptuels et technologiques du développement de la recherche en santé

Modeling of insulin and blood glucose as accelerometry data and heart rate during stress and physical activity in type 1 diabetic relationships to improve the performance of Artificial Pancreas – DIABELOOP_AP

DIABELOOP_AP

Modeling of insulin and blood glucose as accelerometry data and heart rate during stress and physical activity in type 1 diabetic relationships to improve the performance of Artificial Pancreas

Initial objectives of the project

The scientific aim of the Diabeloop_AP project is to develop a model of type 1 diabetes that takes into account physical activity and emotional stress in addition to insulin infusion and interstitial glucose levels in order to optimize therapy and bioregulation of subjects’ blood glucose levels. This project is a methodological research programme made possible by the acquisition of suitable data within the context of the Diabeloop project (artificial pancreas) and the Diabrasport project (measurement of qualified physical activity to improve treatment, steered by CERITD). Application of this model will fit naturally into future versions of the artificial pancreas and will coincide with the availability of miniaturized sensors worn by subjects, integrated into the medical device currently being developed. The accompanying technological aim is to develop a DT1 patient simulator to provide a bank of virtual subjects enabling benchmarking of the bioregulation algorithms currently being developed. It should be noted that the 2 currently available DT1 patient simulators (University of Virginia, University of Cambridge) do not allow physical activity to be taken into account, thus considerably hindering the development of an artificial pancreas system. Our simulator can subsequently be made available to teams working on bioregulation algorithms in order to test their performance.

WP1: Collection of missing data: To construct this new virtual patient model, it is necessary to collect data to match signals measured by the accelerometer and heart-rate sensors in response to physical activity or stress with variations in blood glucose levels and, consequently, insulin requirements. Various databases already existing and provided by the project partners allow this question to be approached about physical activity.
WP2: Link between accelerometry/heart rate signals and activity and stress levels:
The objective of this task is to develop two indicators of physical activity level and stress level. Both indicators will be used to fine-tune Hovorka’s model parameters.
To be successful, this task requires close cooperation between CEA-LETI and CRNH-RA to “make the link” between raw sensor data and medical knowledge.
WP3 – Impact of physical activity and emotional stress on the Hovorka parameters: The objective of this project is to model and develop in available databases the relationship between the set of Hovorka parameters and indicators developed in WP2. This will be carried out using the Diabeloop and Diabrasport databases and the studies developed in SP1. The results will be incorporated in software (virtual patient) within this project.
WP4: Validation of the virtual patient: The aim of this subproject is to validate the virtual patient in silico and Clinical validation of the HRM-augmented and accelerometer-augmented algorithms.

The outstanding results are:
• Consolidation of the energy expenditure estimation (EE) from actimetry (3A) and / or heart rate data. 2 papers have been submitted with these results (one accepted and one under revision) in international scientific journals.
• The application of these methods / models to data from Type 1 diabetics patients
• The development of a predictive model of the glycemic drop as a function of the estimated EE

The results of the Diabeloop_AP ANR project would be integrated in the second version of the Diabeloop Artificial Pancreas system to further improve it. It can also be included in other Artificial Pancreas systems and be used to improve the guidance and training of diabetes patients, whether they use an Artificial Pancreas or not. As such this research project will have a strong medical benefit by reducing hypoglycaemic events, severe complications and improve everyday quality of life for a large number of type 1 diabetes patients.

Publication :
1. H. M. Romero Ugalde, M. Garnotel, M. Doron, P. Jallon, G. Charpentier, S. Franc, E. Huneker, C. Simon, S. Bonnet. An original piecewise model for computing energy expenditure from accelerometer and heart rate signals

. M. Garnotel, T. Bastian, H. M. Romero-Ugalde, A. Maire, J. Dugas, A. Zahariev, M. Doron, P. Jallon, G. Charpentier, S. Franc, S. Blanc, S. Bonnet, and C. Simon. Prior automatic posture and activity identification improves physical activity energy expenditure prediction from hip-worn triaxial accelerometry (Submission 16/06/2017; reviewing 17/07/2017)

Type 1 diabetes affects 180,000 people in France, increasing 3-4% per year and with a significant number of children and teenagers. Patients must perform self-administration of insulin several times daily at doses based on their carbohydrate intake and blood glucose reading, with the level of physical activity (PA), emotional and psychosocial stress also being factors. These last factors are the most difficult to manage, even for patients completely familiar with this aspect of their treatment. PA is the cause of most hypoglycaemic accidents and is a contributing factor in blood sugar imbalance and its attendant ills. Recurrence of hypoglycaemic episodes can significantly impair quality of life. Increased stress levels will have an opposing effect, reducing the effectiveness of insulin and leading to higher blood glucose levels. Maintaining blood glucose levels in the correct range with all these factors is thus very challenging for the patients. However failure to maintain the range will lead, on top of hypoglycaemic events, to severe complications (retinopathy, amputations, renal and cardiac failures,…)
There are several avenues being explored for curing type 1 diabetes (stem cells, islet grafts,…) however they are either unpractical for a whole population, prohibitively expensive or will not be available before many years. As such, the best approach is to improve the current treatment. One possible solution is an artificial pancreas, a system that mimicks the insulin delivery of a non-diabetes person. The Diabeloop consortium is currently developing such a system, which incorporates continuous glucose measurement, a patient interface (to input meal, PA and other patient data), a calculator to determine insulin pump settings, and telemonitoring capability. This project started in 2011, and in 2014 2 clinical studies demonstrated the effectiveness of a prototype on 35 patients. The consortium is targeting 2018 for availability to the patients of the system.
However during the development of this Artificial Pancreas system the PA and Emotional Stress have been problematic. We know they are key factors in the everyday life of diabetic patients, but their effects are under-represented and not very well studied in the clinical studies and literature. In particular the physiological models of insulin and glucose supporting the algorithms do not account for these 2 factors. The Diabeloop_AP ANR project proposes to better understand the impact of PA and emotional stress and to develop an innovative model enabling their integration using continuous measurement of heart rate and accelerometry. To do this we will rely on a number of existing databases we have access to (including from a previous ANR project, “SVELTE”) as well as collect specific data initially with patients. We will then build upon existing, well described physiological models to improve them by including these 2 factors. Finally we will validate this new model during a clinical study.
To achieve this result we assembled a strong team, with partners bringing complementary expertise to the project: the CEA for the algorithm and modelling, the CRHN-RA with their long experience in Physical Activity with regards to Diabetes, CERITD with their Artificial Pancreas and long history of type 1 research, and CHSF and CHU Grenoble bringing their clinical and diabetes expertise.
The results of the Diabeloop_AP ANR project would be integrated in the second version of the Diabeloop Artificial Pancreas system to further improve it. It can also be included in other Artificial Pancreas systems and be used to improve the guidance and training of diabetes patients, whether they use an Artificial Pancreas or not. As such this research project will have a strong medical benefit by reducing hypoglycaemic events, severe complications and improve everyday quality of life for a large number of type 1 diabetes patients.

Project coordination

Guillaume CHARPENTIER (CERITD)

The author of this summary is the project coordinator, who is responsible for the content of this summary. The ANR declines any responsibility as for its contents.

Partner

CERITD
CEA - LETI Commissariat à l’énergie atomique et aux énergies alternatives - CEA
CRNH Rhône-Alpes Centre de Recherche en Nutrition Humaine Rhône-Alpes

Help of the ANR 388,603 euros
Beginning and duration of the scientific project: September 2015 - 36 Months

Useful links

Explorez notre base de projets financés

 

 

ANR makes available its datasets on funded projects, click here to find more.

Sign up for the latest news:
Subscribe to our newsletter