Guy Fagherazzi, a researcher in diabetes epidemiology at the Centre for the Epidemiology of Mental and Physical Health (CESP – Inserm/Université Paris Descartes/Université Paris-Sud/Université de Versailles Saint-Quentin-en-Yvelines) discusses new monitoring devices for people with diabetes.
To monitor their diabetes, be it type 1 or type 2, 422 million people affected by the illness worldwide must follow a daily routine of regularly checking their blood glucose, taking their medication and adapting their diet and lifestyle.
The advent of digital technology now offers patients the opportunity to introduce connected solutions to that routine for a more effective monitoring of various key parameters. Already, people living with diabetes are much more “connected” than patients suffering from other chronic illnesses. However, paradoxically, that data is still underused in terms of improving medical monitoring and furthering research.
After the genome and the microbiome, the time of the “digitosome”
The study of the genome (the genetic material of an organism) and, more recently, that of the epigenome (all the markers, or tags, attached to the genome that regulate gene expression) and the microbiome (all the genomes of the bacteria inhabiting an organism) has produced, in recent decades, a vast amount of data. Similarly, the online data generated by an individual throughout their lifetime is a wealth of information for research.
That is why our research team is particularly committed to developing the concept of the “digitosome” for the purposes of medical research. The term refers to all the data produced on an individual and originating from social media, smartphones, connected items, connected medical devices, etc.
Connected objects and fitness trackers improve behaviour
Connected objects such as fitness trackers and connected watches, as well as smartphone applications, have proven useful in the primary prevention of type 2 diabetes. They encourage wearers to adopt healthier behaviours, preventing the onset of the illness. Connected objects incite high‑risk individuals to step up their physical activity, adopt healthier diets and improve the quality of their sleep, among other benefits.
However, it is important to bear in mind that these tools are merely facilitators, not drivers, for adopting healthier behaviours. To preserve the benefits they provide, other forces for change must exist and take over in the long run. Improvements could include enhancing the fun aspects (gamification) of connected objects, improving ergonomics or user experience, strengthening the feeling of belonging to a community, etc.
People who already have type 2 diabetes also benefit from using connected tools or mobile applications, as they make daily monitoring easier. The patients who use them see their glycaemic control improve drastically compared to those who do not.
An exploitable source of data
The quantity of data generated on the lifestyle of people with diabetes is a gold mine for researchers. By analysing it, they can better understand the influence of modifiable risk factors linked to lifestyle (diet, excess weight, etc.) in the onset of type 2 diabetes, and its complications (cardiovascular illnesses or retinopathy, retinal damage that can lead to blindness). Artificial intelligence (AI) algorithms already use such data to predict individuals’ glycaemic responses following a meal. The era of personalised nutrition is dawning…
As well as the connected objects available to the general public, there are many connected medical devices, specific to type 1 or type 2 diabetes, which have radically changed how to monitor the condition. Recently, the most striking example is that of the Freestyle Libre, a flash glucose monitoring system that simply scans the patient to generate their glucose level, trend (rising, falling) and history over the last 14 days. Soon, the device will operate via Bluetooth. We now know that the users of such devices are less at risk for developing a cardiovascular disease than those who monitor their blood glucose in the traditional way. That is mainly because using the devices reduces glycaemic variability and the frequency of hypoglycaemia. The time spent at target glucose levels, i.e., with optimal glucose levels, set with diabetologists, increases.
Once again, the quantity of data generated by these devices opens many avenues for research. It has already been possible to identify new metrics for patients and health professionals to follow, as well as the simple blood glucose and glycated haemoglobin (HbA1C) measurements, which are only too imperfect in their reflection of the reality of the illness.
Artificial intelligence methods and diabetology
If artificial intelligence were a car, then data would be its fuel. AI methods have already entered the field of medical research, including diabetology. ‘Deep learning’ methods are particularly suited to imaging data, and algorithms have already made it possible to diagnose and rate diabetic retinopathy with a reliability exceeding that of a consensus of experts using images of the retinal fundus.
Ongoing developments also allow for the possibility, in a few years’ time, of a working artificial pancreas that could precisely predict the virtually continuous evolution of blood glucose and thus instantly trigger the necessary insulin response to ease the daily lives of people undergoing insulin therapy.
Finally, algorithms fuelled by vast databases of computerised medical files make it possible to predict optimal courses of treatment for new patients in given clinical and symptomatic states. Although these various breakthroughs are only in their infancy, they allow a glimpse of how medical practice will change, with the addition of automated assistants capable of accomplishing many tasks, without, however, taking over completely.
Today, the main challenge is to develop algorithms using various interoperable data sources (clinical, genetic, connected devices, smartphones, social media, medical files, etc.) simultaneously.
Factoring in patients’ emotions through digital technology
In conjunction with the practice of randomised controlled clinical trials, the advent of “real life” data and digital technology has also brought about a new form of epidemiology, venturing into the fields of medical informatics and data science.
Major cohort studies such as the E4N study have already opted for multi-source data collection (online questionnaires, texts, connected objects, bodily samples). Soon, in France, the Health Data Hub will make it possible to fuse various data sources on a national scale to optimise the use of French medical data. However, a key challenge is currently to include the feelings of participants and patients in such studies.
The online data generated by patients with diabetes (i.e.,”digitosome”) has an impact on every area of diabetes, from research and prevention to monitoring. According to “Digital diabetes : Perspectives for diabetes prevention, management and research”, Guy Fagherazzi, Philippe Ravaud, Diabetes and Metabolism, 2018.
Current studies are more suited to assessing the lifestyle, socio‑economic status, treatment and state of health of participants. They are not so effective when it comes to considering emotions and psychological factors, yet it is a fact that deeply distressed patients who cannot come to terms with their diabetes run greater risks of having complications and a poor quality of life.
That is the reason for hybrid projects such as the World Diabetes Distress Study. In this study, the online activity of patients with diabetes is monitored, and the content shared by worldwide communities of patients on social media such as Twitter is studied through an automated analysis. This approach correlates patients’ emotions with more objective biological and clinical criteria. In France, patients with diabetes can actively contribute to research by participating in new initiatives such as the ComPaRe or Diabète Lab (Diabetes Lab) studies.
What about in the future?
We are entering an era in which patients with diabetes are not simply characterised by one or two recent blood glucose or glycated haemoglobin measurements, but, potentially, thousands of values for key parameters, whether clinical, biological, genetic, emotional or environmental. Generated by digital technology, this data and the AI methods that make it possible to analyse it will call into question the consideration of the different types of diabetes and their management.
People with diabetes are the most connected patients suffering from a chronic illness in the world. According to “Digital diabetes : Perspectives for diabetes prevention, management and research”, Guy Fagherazzi, Philippe Ravaud, Diabetes and Metabolism, 2018.
These new opportunities are promising, but also imply their own challenges, which should not be neglected. Beyond the technological challenges, this evolution must systematically benefit patients, who, thanks to digital tools, should have more influence than ever in the management of their own condition and all related decisions.
Limiting social inequalities linked to digital technology, ensuring an ethical approach to the analysis of these vast quantities of data, improving the training of health professionals and researchers and helping patients to embrace these developments are some of the major challenges facing us in the present and in the future.
Finally, the end of the silo mentality in the field of research, the funding of large-scale projects and the development of the Open Data/Open Source format in medical research should help to speed up the scientific discoveries that will have a true impact on the lives of those with diabetes.
This article is adapted from the literature review "Digital diabetes: perspectives for diabetes prevention, management and research" published by Guy Fagherazzi and Philippe Ravaud in "Diabetes & Metabolism".