Different from most of the people, patients with diabetes carry the responsibility of controlling their disease at all times and wherever they go. Without the proper tools and knowledge, this can go downhill when interfering with their social life, physical activity, convenience, and overall quality of life.
Today, apps and devices integrate machine learning algorithms to make the life of those with diabetes a little bit easier. In this post, we will review three impactful examples of AI systems that are improving the quality of life in diabetes.
The artificial pancreas
From its rudimentary start as a “do-it-yourself” device that combined a continuous glucose monitor (CGM) and an insulin pump, the artificial pancreas is now more reliant on artificial intelligence. The system is composed of a closed loop that simulates the normal physiology of the pancreas, controlled by an algorithm with the primary function of sensing glucose levels changes and infusing insulin at the right time.
Several algorithms were tested to work with the artificial pancreas by using virtual patients, a computer simulation of patients with diabetes. In 2010, the MD-Logic artificial pancreas integrated a “fuzzy logic theory” algorithm to imitate the behavior of a caregiver in the control of diabetes. This system showed to get better control of blood glucose levels while at the same time decreasing the risk of hypoglycemia.
Image from FDA.gov
Later, Medtronic adapted the MD-Logic technology and created the first hybrid artificial pancreas where the patient needs to infuse insulin before every meal, but the initial levels of insulin are adjusted automatically. The FDA approved it as the MiniMed 670G in 2016. According to a user, there’s still a lot to desire from the currently available model; however, he experienced a more stable pattern in his glucose levels.
With the MiniMed, patients still have to calibrate the CGM periodically with a capillary blood glucose measurement. New CGM systems and insulin pumps have been released since then, with aesthetically pleasant designs and smartphone integration, as well as other closed-loop systems that compete with the Medtronic device.
How can artificial pancreas improve the life of a person with diabetes?
Patients with insulin-dependent diabetes need to be vigilant of their blood glucose control to avoid ketoacidosis, hypoglycemia, and other serious complications. For this, they need to calculate their insulin dose, measure their blood glucose regularly, and monitor their carbohydrate intake. This strict routine can become time-consuming, discouraging, and lead to a lower quality of life.
The end goal is an artificial pancreas that collects data from the CGM sensor without the need for input from the user. This would save time in calibration and calculation of doses and carbohydrate loads; it can also improve activities such as exercise, prevent poor treatment, and predict adverse events.
Apps for the daily life of patients with diabetes
Described as the “first mobile prescription therapy”, BlueStar is an app focused on education for patients with diabetes that collects data from the user’s behavior otherwise never available to healthcare providers.
In an interactive way, it shares personalized and manageable pieces of information from evidence-based medicine. This is especially useful for those who have just been diagnosed with the disease but, as the system learns from the patient, it will adapt to the level of knowledge and provide information that will be individualized and relevant to the user.
Moreover, the app makes a summary of the information, and the patient can share it with his/her care team, making the communication easier during a consult and facilitating the decision-making in the treatment plan.
In a randomized clinical trial, patients using BlueStar decreased their A1c by 1.9% compared to a 0.7% decrease without the app.
Registered as a medical device, the mySugr app is heavily focused on data collection. It can be integrated with other apps and devices, Apple Health, and Google Fit, and it gamifies therapy goals motivation by introducing challenges led by a monster character. It also takes into account problems that are very specific to patients with diabetes, such as forgetting a dose of insulin, experiencing issues with the insulin pump, or the stress of eating out.
One of its most distinctive features is the estimation of A1c without the need of a blood test, which can give the patient peace of mind before getting the results from the doctor’s office. According to a randomized retrospective study, the app improved the quality of blood glucose control in patients by reducing an average of 16mg/dl in blood glucose, and 0.43% in eA1c, especially after week 8 when the app starts a coaching dialogue. They also concluded that the act of registering their blood glucose is more beneficial in those patients that are “less well-controlled.”
Glooko is an app more specialized in collecting data from CGM devices. The application simplifies this information and adapts it to real-world data for an accurate estimation of glucose levels.
Its algorithm was validated in conjunction with the interpretation of 57 clinical specialists, and it can also pull data from blood glucose meters, insulin pumps, and exercise trackers.
Smart Thermometers For Insulin
Ensuring that the insulin is always at a constant temperature is critical for the proper effect of the medication; however, the need to keep these conditions at all times can be limiting for patients that want or need to travel regularly, or that don’t have a reliable source of electric energy in their locality. Also, domestic refrigerators can fail and vary their temperature, and other situations could limit access to them.
As insulin is so crucial for the daily life of a person with diabetes, keeping the right temperature can be a cause of stress and inconvenience in many situations, interfering with a good quality of life. Moreover, the damage that warmth can cause to the medication will lead to inefficacy, inappropriate treatment, and higher therapy costs.
Thermostats are a form of artificial intelligence that is always taken for granted. Usually, patients keep their insulin in their fridge and rely upon the appliance’s thermostat to keep the medication at a consistent temperature.
The digital health company MedAngel developed a portable smart thermometer that sends data to a smartphone app alerting the patient of temperature changes and keeping a registry of these throughout time. In one of the reviews from the device’s users, the patient was surprised to find that the location where he was keeping his insulin in the refrigerator had a less than optimal temperature, and the smart thermometer helped him find the ideal spot for his insulin.
Working on the next life-changing app for diabetes? Get access to clean data for the development of novel apps and other softwares at the John Snow Labs Data Market and explore our diabetes curated datasets.
FDA- What is the pancreas? What is an artificial pancreas device system?
ADA- Atlas et al. MD-Logic Artificial Pancreas System A pilot study in adults with type 1 diabetes
Contreras et al. Artificial Intelligence for Diabetes Management and Decision Support: Literature Review
FDA- The 670G System – P160017
Welldoc- Shomali et al. A Data Science Framework for Mobile Health–Engagement and Outcomes
Hompesch et al. Clinically-Relevant Improvement in Quality of Blood Glucose Control in Well-Controlled Users of mySugr’s Mobile Diabetes Management Tool