Saturday, May 31, 2014

How can data serve the diabetes community?

We at Sensotrend tend to promote cooperation between companies quite a lot. As a part of this activity, we're quite active in providing feedback to other startups developing their services. Below you can find an example of this.

The process was triggered by the tweet:
The text below is based on the email discussion that followed, formatted into a blog post.

Hi Alex,

I fully share your vision of combining data from medical devices and wellness trackers, and using that data to improve health and healthcare.

Regarding your questions, here goes:

  1. Tell us a little bit about yourself :).

    Male, age 38, software professional (user experience design, programming, project and product management, sales, consulting) for 16 years, type 1 diabetic for 15 years. A startup entrepreneur for less than a year now.

    Regarding my diabetes, I’m currently using an insulin pump. My HbA1c levels (the ‘official metric’, a rough indicator of my blood glucose levels lately) has been between 6.9 % and 7.3 % (using the old NGSP scale still widely in use, 52 - 56 mmol/mol on the new IFCC scale), so fairly OK.

    My HbA1c used to be worse. Around 4 years ago, I had a kind of a diabetes burnout. I did inject insulin and measured my blood glucose, but I could never really estimate my blood glucose without measuring, nor explain the result after a measurement. The measurements just were anywhere in 3.5 - 22.5 range (again, different scales exist, I’m using mmol/l here, I know mg/dl is used in the U.S.), and there was no visible trend, they were just scattered all over the place, a shotgun pattern. My HbA1c was 9.4 %. It was really frustrating, and overwhelming.

    What helped me out of that burnout was data from a continuous glucose monitor (CGM). It helped me see patterns and trends. Even when it did not always provide clear answers to why something was happening, it gave me better tools to approach those questions myself.
  2. Please describe how an average day in your life looks like? Specifically, please let us know where the medical system plays a role...

    I prick my finger and measure my blood glucose several times per day. Usually when I wake up, immediately before and sometime after each meal, before, during and after any exercise, and before bed time.

    Whenever I eat or drink something containing carbohydrates, I need to think of the effect it has on my blood glucose, and dose insulin accordingly. The same with all physical exercise, and also lack of any exercise. Stress and illnesses also affect glucose levels and insulin consumption.

    Currently, my blood glucose meter is my command center and dashboard for daily operation. It has some intelligence built in, known as a bolus calculator. I have configured it with some parameters: Carb Ratio tells how much each gram of carbohydrates elevates my blood glucose, Insulin Sensitivity tells how much each unit of insulin brings it back down. There are settings for Stress, Illness, two Types of Exercise, and also for Menstrual Cycle, which I’m not using. With all these parameters, the bolus calculator helps me calculate the correct dose of insulin for each situation.

    When I woke up today at 6:49, my glucose was 12.9, a bit high. At 8:17, before breakfast, it was 18.2, so quite high already and rising fast. I need to think why that’s happening. My assumption is that my glucose has been low during the morning hours, causing the liver to release extra glucose to my blood stream, but since I don’t have any measurement data from the night time, I would need a continuous glucose monitor (CGM) to verify that. There may be other factors as well. Sometimes the infusion site (where the pump injects the insulin to my body through a cannula) gets corrupted somehow. Sometimes just pure stress causes my glucose levels to rise. It might also be that the meter reading was off for some reason, but I tend to rely on it being within the 15 % threshold. It’s been really quite usual for my second measurement being higher than the first one, but not by this much. I think about the basal insulin (continuously dosed background insulin) rate during the night hours. If my glucose levels have in fact been too low, it’s an indication that there has been too much basal insulin. However, I did not do any particular exercise the previous day (after sports I need to reduce the basal rate), and as mentioned, usually the trend has been upwards in the morning, rather an indication of too little basal insulin. Then again, it might be that the rising trend only starts when I wake up, as a result of stress, and when asleep the trend is in fact downward. And yet again, we’re just past the 4 day weekend due to Easter holidays. This may have affected the levels of my cortisol hormone. I might have less stress, and therefore the glucose levels have dropped last night, whereas usually the stronger cortisol keeps them up. Who’s to tell? There are several hypotheses, but not enough data.

    I then looked at my meal. Breakfast is usually the easiest of the day, as it tends to be pretty much the same each day. I entered my guesstimate of 50 grams of carbohydrates in my blood glucose meter. It told me to dose 6.3 units of insulin for the carbs, and another 6.5 to bring the high glucose back to the target range. I accepted and pushed OK. The meter communicates with the pump over Bluetooth, and the pump starts to dose the insulin. This is called bolus insulin, one that you dose per event, in addition to the background basal insulin.

    After breakfast, I worked at home for some time, then walked to the office. The walk takes less than 20 minutes and I don’t really count that as exercise, but a pedometer application on my phone keeps track of my steps. When I first started to use the pedometer it took me by surprise that I can walk less than 500 steps on some days when just programming at home, and nearly 20 000 steps on others, when there are several meetings in different places, without really paying attention to it. But it surely has an effect on my glucose levels and insulin consumption, so nowadays I do check the amount of steps some time during the evening.

    At work, I noticed the first symptoms of hypoglycemia, blood glucose dropping too low. I measured for confirmation and got the confirmation from my meter, 3.9 mmol/l. I always carry glucose with me for these situations, and took 6 tablets, enough to restore my glucose. The extra insulin to fix the high glucose at breakfast was probably too aggressive. I’ve created a mental note for myself never to correct a high with more than 5 units with a meal and not more than 2 units when not eating, but somehow managed to ignore that at breakfast. I could adjust the insulin sensitivity parameter in my bolus calculator, but it seems to be working fine with smaller corrections. So I’ll leave it as it is, for now, and try to remember my own rules for bigger corrections. An hour later, before lunch, my meter shows 5.1, a good number.

    My day is filled with wondering and pondering like that. With exercise, it becomes even more complicated. Exercise has been the biggest challenge for me regarding my diabetes treatment. If I don’t do any, my blood glucose values tend to go up, no matter how I try to compensate with additional insulin. However, when I do exercise, I find it hard to adjust my insulin levels accordingly.

    When eating, I need to consider whether I’m going to be active within a few hours after the meal, and in that case lower the amount of bolus insulin. I also should lower the basal insulin up to two hours before beginning to exercise. When exercising, I need to think how long it’s been since I last ate, does the level of exercise match the level I estimated when setting the basal rate. I also need to pay specific attention to my mood and stress. When I go jogging, my glucose levels go down, but when I play soccer, they tend to go up, due to the competitive nature of sports and the related adrenalin. And after the exercise, I need to think about how I should adjust my basal rate for the following hours, the following night, and the next day.

    Lately stress has become a major factor as well. When pitching my startup for a room of investors, I can feel my pulse increasing and a certain nervousness. Nothing that would prevent me from performing, but something clearly affecting my glucose levels (from 5.9 to 18.6 without any extra insulin to compensate the effects of adrenalin).

    You’d think one would eventually get a hang of the required insulin doses for each situation. I’ve had diabetes for almost 15 years now, and I’m still learning. Our soccer season is just starting. The first games are a shock to my body, and it takes time until it adapts, so each game is different. And each new season of soccer has a different impact. My body is not the same it was last year, the winter was poor in Finland and I didn’t ski at all, but I played futsal. All these things play their role in the equation, and it’s not a simple one to solve.

  3. On a regular basis, what data (e.g. blood pressure, activity, etc.) are you most interested in referencing? What do you use it for?

    Most important for me are the blood glucose levels. Are they on or off target, do I need to correct? Also, I need to see the trends, and try to learn from them.

    I monitor my physical activity and exercise quite extensively, to see how it affects my blood glucose and to learn to adjust the insulin doses accordingly.

    When my blood glucose levels are high and won’t come down for some reason, I check my blood for ketones, an indication of possible diabetic ketoasidosis.

    I’ve tracked my blood pressure occasionally. However, the levels seem to be quite static and in a safe range, so I only measure occasionally nowadays, around once every two weeks.

    Regarding urine tests you mentioned, I get checked for microalbuminuria once a year, with a bunch of other lab tests. It is a bit awkward to transfer the sample container to and from home for night time urine collection, but really not a problem I would personally pay any real attention to.

  4. With #2-3 in mind, where are some areas where you feel you could use some help or that someone could improve your daily life?

    I’d appreciate having the possibility to wear a CGM for longer periods of time. Nowadays I can get one from my diabetes clinic for one or two weeks, but I feel that’s hardly enough time to learn anything. There are so many factors involved.

    I’d like to be able to log all nutrition in a more automated way. I’ve seen some attempts in this direction, devices analyzing the blood stream non-invasively and software recognizing the macronutrients from a picture of a meal. I think the need for this kind of measurements is there, but personally don’t trust any of the available solutions enough.

    I’d also like to track my stress levels. I know there are things you can measure, but I don’t currently have the tools. I also don’t know which would be more suitable; heart rate variability, galvanic skin response, or some other metric.

    When analyzing the glucose measurements, whether from a glucometer or a CGM, I’d highly appreciate the possibility to have the results annotated in detail: what kind of exercise was involved (aerobic or non-aerobic, competitive or not), what nutrition, etc., and also the possibility to search and sort the results (show me all the game days from previous soccer season, and also their next days).

    I know the bionic pancreas is being developed for people with type 1 diabetes, but think its commercial deployment is years and years ahead. Still, even now some kind of alerts and advice throughout the day would be welcome.

  5. Given the above-mentioned context, what are the top 3 things you would like to see someone do for you?

    First thing for me would be the diabetes diary with automated data collection from multiple wellness trackers. That’s something I know I can do myself, so that’s what our startup Sensotrend is concentrated on. What would help us are API’s for different services and devices. All of the devices should really transfer their data to cloud services automagically. I know some people think it’s empowering for the patients to enter the data manually to logbooks. No, it isn’t. It’s simply too much of an effort. Too prone for errors, omissions, too. It’s much better to save the efforts for analyzing the data.

    As mentioned, I’d like to be able to track my nutrition easily. I currently have a great incentive to enter my carb estimates to the bolus calculator. But I don’t really track the glycemic index or other macronutrients of my meals, although I would like to have that information present when analyzing my glucose measurement data. Just a great meal logging app with superb assistance would be cool, but all fully automated solutions are very welcome as well.

    Another thing I can think of is the measurement of stress. I’d like to be able to track that without any effort, and be able to concentrate on matters that cause the stress. I’d also like to quantify it better than just based on how I feel.

    Finally, data alone does not give you answers. For me, seeing trends from CGM allowed me to see some patterns, as opposed to the shotgun chart of fingertip measurements. But I needed to find the ways of stabilizing that rollercoaster. So it is super important to get high quality instructions and assistance.

    The current healthcare system gets me from bad condition to OK, but how about when I want to go from OK to great? I understand the resources of the skilled doctors are best spent in bringing as many people as possible to the OK level. On the other hand, when I’m OK, and still have the drive to learn more and keep improving, I don’t think I’d be requiring so much handholding. Just high quality resources I could study on my own. Resources I can trust. I find it rather hard to find those.

    Even better than reading material would be services that analyze the data I’ve tracked, find patterns there, and provide explanations and suggestions. That’s where I hope to lead Sensotrend someday, and really hope we won’t be there alone. We’re a long way from there, though. Plenty of research is required before that. And plenty of data. We’re hoping to advance step by step, first just visualizing the data from different sources in a single user interface optimized for diabetes treatment, then perhaps sharing that information with peers and healthcare professionals, and gradually introducing algorithms making data analysis easier for people, before ultimately letting algorithms propose changes to treatment parameters.

  6. Anything else related to this area that you think someone should be solving or that you would like to see?

    I’d like to see healthcare services increasingly adopting the data from wellness trackers and other devices the patients use themselves. I’m already seeing a lot of that and think that’s a clear trend. It could still be accelerated, though.

    I’d like to see the big healthcare IT giants open up their interfaces for third parties to introduce services on their platforms, but kind of understand why that’s taking some time to happen.

    I’ve heard some doctors have mixed feelings regarding information provided by the patients. All of the healthcare professionals that have participated in treating my diabetes appreciate the data that reflects the real life of the person with diabetes. They do prefer the glucometer log over a pen and paper logbook, though. So I’ve never really seen that trust issue, but have heard it exists in some other fields of medicine.

I’m happy to elaborate on any of the answers. I’d also be happy if I would have been able to write them in a shorter format, but this is the shortest I could do this time. So apologies for the length, I hope you find the information valuable, though. :)

No comments:

Post a Comment