Database Enabled Insulin Pump (DEAP) David Bowen,
[email protected], 2009 When I think of the challenges of diabetes control, I think of two things: carbohydrate counting and insulin absorption. Many people live like I do, with rotating schedules where no two days in their lives are the same. Food comes from many different sources, and has greatly varying ingredients. Up until now, the best a diabetic could do is to only look at carbohydrates in food and give insulin accordingly. However, all diabetics are familiar with the pizza paradox, where the combination of carbohydrates and fat cause a bifurcated glucose response. This blood sugar roller coaster leaves diabetics feeling far worse than the rest of the world’s non-diabetic population after eating cheap pizza. Insulin pump manufacturers are aware of this problem, and have added a two peak bolus feature to insulin pumps, giving part of a dose now, and giving part of the dose later to handle both the carbohydrates and the fat. This is fantastic; pizza can now be handled. Unfortunately, we live in the real world where all food isn’t as simple as pizza, and no one should be eating pizza for every meal. Why can’t all foods have customized glucose responses, like the bifurcated response of the two peak (pizza) bolus?
Figure 1. Generic glucose response curve.
My proposal for an improved diabetes control device is the following. Ipods and Iphones and all manners of PDA can store gigabytes of data. Why can’t an insulin pump remember food glucose responses and allow them to be used in a user friendly way? Consider an insulin pump that looks more like a Blackberry®. Within the memory is a personal database of food. These foods include commercially available restaurant food as well as common generic foods. This database is also not static. New foods can be added
by downloading upgrades from the manufacturer’s website, or customized by the user. Stored along with each different food in the data base will be the data for the glucose response (See figure 1). At the touch of a menu, the particular food pops up displaying all nutritional information including the glucose response curve. Well, simply having the glucose response curve may be informative for a diabetic, but not particularly useful. To make the database at all useful, the glucose response curve can be administered by the pump. The pump will not act only to bolus for a few minutes, but will tailor the dose over many hours to achieve the most stable glycemic response as indicated by the particular foods glucose response curve stored in the database. Of course, rarely do people eat just one particular thing, so selecting many foods from the database will effectively be administered simultaneously. The database enabled insulin pump (DEAP) will sum the glycemic responses from a bouquet of foods in a single meal and form the appropriate and complex insulin response.
Figure 2. Example of a database page containing the food info for a Subway tuna sub.
As stated earlier, the glucose response curves would be available for download from the pump manufacturer. These glucose response curves can be determined empirically through non-diabetic trials, or calculated based on the carb-protein-fat signature of a food. The curve is scaled based on quantity consumed and the diabetic’s particular carbohydrate ratio and initial blood glucose. In the mid 1980’s, much research was conducted studying the glycemic response of people after eating “mixed meals” (meals with identifiable carbohydrate, protein, and fat components). The correlation was made to a single foods glycemic index, which is defined as the area under the glucose response curve of the food divided by the area under the glucose response curve of white bread. The three papers reviewed [1]-[3] indicated a constellation of factors that effect the real glucose response of a person. This fact makes the approach of bolusing for carbs and basal for everything else method of diabetic control seem outdated. It was nice to have instantaneous control after the introduction of lispro to the market, but it has been over ten years and more sophisticated techniques are needed. Storing the glucose response of thousands upon thousands of foods into a database enabled insulin pump (DEAP) that will intelligently combine the glucose responses of each food in a meal is the next best thing to an operating pancreas.
Figure 3. Glucose response data for mixed meals [3].
The benefits of the DEAP are critical on one external factor: insulin absorption. Insulin pump infusion sets should be changed every three days in order to maintain a high rate of insulin absorption. However, three days is an average time, and particular irritation of the site or the insertion of the cannula into scar tissue could result in poor absorption long before three days is up, making boluses ineffective and blood sugars erratic. To further aid diabetics in their control and improve the operation of the DEAP (or any insulin pump), I propose the following. A micro electro-mechanical (MEM) salt sensor be placed at the tip of the cannula to monitor the rate of insulin absorption. The subcutaneous site around the tip of the cannula will have a predictable change in salinate level from the injection of insulin. This level will dissipate back to an equilibrium salinate level as the insulin is absorbed. If the insulin is being absorbed below some threshold rate, the salt level being monitored by the MEM sensor will not return to
equilibrium quickly enough, and the user should be alerted to change infusion sites as soon as possible. Thus, high insulin absorption will be monitored and maintained. The concept of the DEAP replaces the function of the pancreas as an insulin controlling organ. The improved infusion site monitoring will account for the external nature of the DEAP and the problem of delivery. With the DEAP and near perfect insulin absorption, unexplained blood sugar spikes will be a thing of the past. References: [1] Wolever, Thomas. Jenkins, David. “The use of glycemic index in predicting the blood glucose response to mixed meals”. American Journal for Clinical Nutrition, 43, 167-172, 1986 [2] Chew, Irene. Brand, Janette. Thorburn, Anne. Truswell, A. “Applcation of glycemic index to mixed meals”. American Journal for Clinical Nutrition, 47, 53-56, 1988 [3] Gulliford, Martin. Bicknell, E John. Scarpello, John. “Differential effect of protein and fat ingestion on blood glucose responses to high- and low-glycemic index carbohydrates in noninsulin-dependent diabetic subjects”. American Journal for Clinical Nutrition, 50, 773-777, 1989