Ways to better insulin control through data gathering and analysis
Recently I participated together with one Diabetologist in curing patients with diabetes, in a panel organized by Abbott, on the topic of different tech advances, and possibly promises that AI can bring in different aspects of health in general and this chronic disease in particular. Since it was an interesting and fruitful conversation, I wanted to share some remarks and insights.
First, it was interesting to note how the definition of health also changes with time.
Since we are getting more entangled with different technological accessories, it is worth asking what the limits of our health are and when it is indistinguishable from our “equipment”?
Can we transcend our technological limits? For example, can a person say that he is not healthy if his pacemaker is dysfunctional?
In general, one of my previous research projects showed, that different tech and AI experts believe that Health is one of the most important areas where AI developers both will and should invest money/resources. We can only imagine what will be the differences between the countries applying advanced AI health-related products to their population, and those who fail to do it (for any number of reasons). There is room for cooperation between government institutions and the private sector, to make the most of the financial resources, human capital, but also data that is sometimes available only to state-owned institutions.
Some are claiming that exactly AI and its applications are bringing quantum change in the fight with diabetes. But why is it important, especially with this disease?
Numbers differ a bit, but there are estimations that around 1o to 12% of overall health costs (that are in the range of 760 billion per year, worldwide?) are attributed to diabetes patients.
There are around 422 million people with diagnosed diabetes worldwide, but some statistics are warning us that only 1 out of 2 persons with diabetes are diagnosed, and others are living in at least temporary ignorance.
Additionally, some papers (Dankwa-Mullan, Rivo, et al, 2019) that a staggering 31% of diagnosed patients, stop using proscribed medications after 3 months?!
Diabetes care is harder because of the permanent lack of real-time data that could be used. And there are at least 3 segments where more data, and on-time data could help our health systems:
1. Prevention of diabetes
2. Detection of diabetes
3. Management of diabetes
Projects that are under development will be implemented in the different parts of the ecosystem — on patients, with health systems as a structure, and with medical doctors and nurses that could all have benefited from Artificial Intelligence and Machine Learning.
The ecosystem could be defined with a few elements:
A. Wearable sensors and devices (a great example is Abbotts FreeStyle Libre)
B. Smartphones and apps using and processing data (from sensors)
C. E-Health records and data mining possibilities (if they are filled with data properly)
E. Online communities and social media (which are getting more active and vibrant ben before)
New possibilities are opening for better blood glucose control, fewer episodes of hypoglycemia, and less comorbidity and complication of diabetes patients.
When we gather enough and wright type of data, it will give researchers opportunities to observe and aim for a few objectives that were elusive until recently:
Spotting patterns of behavior (that can lead to disease development, or connected with good management, etc.; Early diagnosis, especially for those who would miss it; and finally famed personalized treatments and recommendations. Even in the cases when medical staff has enough resources and time to deal with every patient personally, new amounts of data will allow them to use not tens or hundreds of data points, but thousands or more, and in a structured and user-friendly way (hopefully). Then personalized approach will get to new levels of personalization and the overall picture will have a much better “resolution”.
Wearable devices together with already used tools could better connect carbohydrate intake and insulin dosing on an individual level because of the integration of more information sources such as blood parameters, dietary habits, and physical activity. It is reasonable to expect that it would lower the number of hypoglycemia episodes, especially those happening during the night.
Some projects are promising even to make some diagnoses in a noninvasive way, just using cameras and pictures that could detect for example foot problems and warn patients to react on time, thus saving a significant number of amputations that are happening because of diabetes-related problems.
In general, there is an important psychological aspect of those advances. Diabetes obviously is a complex disease, that is not easy to understand and monitor and have under control for a long period of time that it requires it. Consequences are not immediate but visible very often after years or decades, which is causing a higher level of neglect for some proportion of patients.
In this way, tools, and trends that I mentioned could give patients better control, include and engage them on a deeper level in the process, giving them better explanations and insights about the current and future state of their bodies and a chance for better understanding of the whole process.
As some patients commented to me: it could be really revolutionizing the care for diabetes and lives of the all people involved in it. We might say that one revolution already happened, and it will be even more impactful when the data and opportunities will be more adopted and used.