Integrated Diabetes Management

B E A R S

B R E A K I N G

B O U N D A R I E S

I n f o r m a t i o n Te c h n o l ogy for Society • White Paper Competition

INTEGRATED DIABETES MANAGEMENT

!"#$ %&'()$$

TEAM MEMBERS:

CHRISTOPHER HANNEMANN graduate student, Department of Mechanical Engineering, UC Berkeley SARAH BETH EISINGER graduate, Department of Electrical Engineering and Computer Science, UC Berkeley

1 9 3 4 S t u a r t S t r e e t / B e r k e l e y, C A 9 4 7 0 3 • t e l e p h o n e : 7 0 3 . 4 0 2 . 7 0 6 4 • e - m a i l : c h r i s h a n n e m a n n @ b e r k e l e y. e d u

INTEGRATED DIABETES MANAGEMENT

Abstract

Diabetes is one of the most prevalent chronic illnesses, afflicting over 20 million people in the United States and over 180 million worldwide. Tight control of blood glucose levels is crucial in preventing complications related to diabetes, which include blindness, kidney failure, severe nervous system damage, stroke, and heart disease. This control requires the patient to carefully monitor blood glucose levels, diet, exercise, and insulin or other medication intake, as well as understand how each affects his or her glucose control. Without comprehensive and accurate record-keeping, it is difficult for both doctors and patients to effectively manage the disease. Current diabetes data management techniques fall short in terms of convenience, accessibility, and useful analysis features, resulting in inadequate data management for most patients. This paper proposes the use of information technology to improve and increase both the acquisition and presentation of data for diabetic patients. A system integrating glucometers and insulin pumps with wireless data transmitters, wireless mobile devices (such as smartphones), personal computers, and a web-based records database is presented, and technical and market feasibility is assessed. Such a system can leverage the power of information technology to provide significantly better analysis, feedback, and self-management tools for diabetics, potentially improving the glucose control and, subsequently, the life of millions of diabetics worldwide.

I. Introduction I.A. Statement of Need

Diabetes is one of the most common chronic illnesses and biggest public health problems in the world [1]. The World Health Organization (WHO) estimates that, as of 2006, more than 180 million people worldwide have diabetes, with that number expected to double by 2030 [2]. 2.9 million deaths per year can be attributed to diabetes and its direct complications. In the United States alone, there are 20.8 million people––7% of the population––afflicted with diabetes [3]. While it ranks at the 6th leading cause of death in the US as listed on death certificates, this is a glaring underrepresentation; in reality, only 35% – 40% of deaths related to diabetes have it listed, and only 10% – 15% have it listed as the underlying cause. The overall age-adjusted risk of death is twice as great for people with diabetes. In addition to the direct effects on the endocrine system that are characteristic of the disease (the inability of the body to produce and/or effectively use insulin, and in turn to properly metabolize sugars), there are numerous complications associated with diabetes [3], [4]. Table 1 lists several of the most common complications experienced by diabetics as reported by the Center for Disease Control and Prevention (CDC) in 2005.

Many of these complications arise due to the excess flow of free glucose in the blood stream, which interacts

adversely with organs and generally causes a decrease in blood circulation. Other problems related to diabetes include acute life-threatening events, such as ketoacidosis and extreme hypoglycemia, and a greater susceptibility to other illnesses. In general, diabetics are typically more likely to exhibit worse effects from common illnesses and have higher death rates associated with infections such as influenza and pneumonia. The economic burden associated with diabetes is immense. In 2002, a total of $132 billion was spent in the US on diabetes [3]. This amounted to an average of $13,243 per capita for diabetics versus $2,560 for other individuals, an increase of 2.4 times after age-adjustment factors.

H a n n e m a n n / E i s i n g e r

1

Integrated Diabetes Management

Table 1. Common complications associated with diabetes in the United States [3]. Leading cause of new adult blindness Leading cause of of kidney failure 60% - 70% of diabetic adults have mild to severe nervous system damage 73% of diabetic adults have high blood pressure or take prescription medication for hypertension > 60% of non-traumatic lower limb amputations occur in diabetics Nearly one-third of diabetics have severe periodontal disease Risk of heart disease is 2 to 4 times greater Risk of stroke is 2 to 4 times greater

Constant control of blood glucose levels can greatly reduce the risk of microvascular complications (eye, kidney, and nerve diseases) as well as decrease the frequency of severe hypo- and hyperglycemic events [5], [6]. The CDC reports that such risk is reduced by 40% for each percentage point drop in HbA1c levels 1 [3]. Adequate glycemic control requires patients to carefully balance their diet, exercise, and medication, and to understand the effect that each has on their blood glucose levels. For some, maintaining a healthy diet and regular exercise regimen is enough to keep glucose levels controlled. For others, including all Type 1 diabetics (who produce no insulin and absolutely require daily injections or other subcutaneous insulin delivery to live), a significantly more rigorous plan is required. Measuring blood glucose levels two or more times each day using a glucometer, counting carbohydrate intake, understanding the complex effects of exercise, and regularly adjusting insulin dosages is a minimum for acceptable glucose level maintenance. The importance of each of these factors can be better understood through explaining the use of basal and bolus insulin dosages.

Insulin dosages are typically administered using two overlapping time-scales: long-acting and short-acting. Long-acting insulin is used to cover the normal metabolization of sugars in the body and serves as the daily foundation. If delivered via injection, this insulin is either a 12-hour shot given twice each day (such as NPH) or a 24hour shot given once daily (such as insulin glargine, or Lantus). This long-acting portion of the insulin therapy is called the basal dosage. Insulin pumps use low-dose, frequent pumps of short-acting insulin instead of single doses of long-acting insulin for the basal. Short-acting insulin is used whenever a patient eats or needs to correct an abnormally high glucose level. This type of insulin (such as insulin lispro, or Humalog) may be given several times each day, and is called a bolus. The bolus varies based on current blood glucose level, activity, and food intake. It is typically dosed using Equation (1),

Bolus [ units] =

1

€

Carbohydrates [ g]

 g  Carbohydrate Sensitivity Factor unit 

+

 mg   Current Glucose Level    dL  

 mg    dL 

- Target Glucose Level 

 1 Insulin Sensitivity Factor units ⋅

mg

(1)

dL  

HbA1c––or glycosylated hemoglobin––is used to measure the average plasma glucose concentration over an extended period of

time. It is commonly used by doctors to get a measure of “average” blood glucose levels for a patient, thereby assessing their glucose control and risk of complications.


2


The need for complete and accurate data becomes apparent when considering this formula. Not only are situation-specific values (current glucose level and amount of carbohydrates) required for the patient to deliver an appropriate bolus, but the carbohydrate and insulin sensitivity factors themselves are based on statistical analysis of a patient’s glucose readings, diet, and insulin dosages, and therefore require significant amounts of accurate data for proper adjustment. Although exercise is not explicitly accounted for in the above formula, it is the most complicated factor to both measure and understand. While some activities, such as jogging, require a reduction in basal dosage, bolus dosage, or both, activities that elicit adrenaline release may actually require a slight increase in bolus dosage. To compound this complexity, the amount of increase or decrease in dosage required, as well as the duration of the specific exercise’s effects, is highly variable from patient to patient.

The most effective tool that doctors have in preparing effective plans of treatment for diabetic patients is

accurate and comprehensive data. Several advancements have been made over the past few decades to assist in the acquisition of such data: the replacement of urine test strips with digital glucometers; the inclusion of nutrition facts on packaged food; the use of continuous glucose monitoring systems; and the improved access to HbA1c testing. Unfortunately, it is still difficult and often extremely inconvenient for a patient to record all food consumption, exercise, insulin/medication intake, and glucose levels for a single day, let alone every day.

I.B. Existing Solutions

There are currently several methods of data acquisition and recording used by diabetic patients, often in combination. These include paper logs, glucometer/pump memories, desktop management software, and integrated computer solutions. A brief description of each method with examples is provided below. Paper Logs: Paper-based data logs are typically paper booklets with small spaces for times, blood glucose levels, and insulin dosages, with some also having space for food and exercise records. This is the most commonly used “comprehensive” method due to its minimal price and ease of use. With diligent recordkeeping, patients can keep up-to-date logs and then share them with their doctors in-person or via fax/mail. However, There are several issues with handwritten logs. They require carrying the log on person at all times (or remembering events/meals to record later). They provide no analysis of trends or correlations, leaving this instead to a patient’s or doctor’s ability to “see” commonalities and anomalies in a table of numbers. Since sharing logs with a doctor is done either at appointments (every 3-to-6 months) or via scanning and faxing, mailing, or e-mailing, professional feedback is very limited. In addition, paper logs do not allow for dynamic entries––that is, non-standard days are tough to record because log is set up only for some predetermined entry type. Appendix A shows a typical log book setup. Glucometer/Insulin Pump Memories: Glucometers, or glucose meters, are able to store a limited number of previous blood glucose readings in an internal memory. The number of readings ranges from 10 to 3000, with an average of around 400 [7]. A patient can access these memories through a menu on the meter itself, going back through the readings one at a time. Most meters also provide some basic statistics such as 14and 30-day averages. Some advanced meters, such as LifeScan’s OneTouch® UltraSmart®, allow for entry of basic food, exercise, and insulin data and provide some trend analysis through graphs [8]. This is performed on a low-resolution, black-and-white screen with a limited number of input options. Memories can be downloaded to a personal computer via a proprietary USB cable, which is not provided with the monitor. Doctors often use this capability in their offices at the beginning of appointments (especially for patients who do not keep good paper logs), but again, this does not help the patient in between visits. Similar logs are kept by insulin pumps, with the same interface and connectivity issues prevailing. Because they are stand-alone devices, time discrepancies often arise in the memories when patients replace batteries, change time zones, or improperly set their glucometer’s settings, rendering the data log asynchronous with any other records that may have been kept.


3


Desktop Management Software: Diabetes management software is gaining popularity because of the prevalence of personal computers in the home as well as the integration of glucose monitors with said software. Unfortunately, current software suites suffer from some common shortcomings, related both to the platforms upon which they are built as well as the habits they engender. Many glucometer-specific software packages are operating system-specific, rendering Apple and Linux users (as well as those with incompatible versions of Windows) helpless. While other software packages may be platform-independent, they still do not solve the problem of recording data other than blood glucose readings; recording exercise, insulin dosages, and diet still requires manual entry and essentially the use of a paper log (unless the patient has their computer on-hand). Connectivity is still not optimal either, with patients only receiving feedback through analysis when they physically link their monitor to their computer. Even if the patient has nearconstant access to a computer, it is still highly inconvenient to reconnect every time a new glucose check has been completed. The fact that records are kept on a computer’s hard drive instead of on the internet makes remote access nearly impossible. Integrated Computer Solutions: These solutions come the closest to providing patients and doctors with the access, analysis, and convenience needed to effectively record and use patient data. An example is Sinovo’s SiDiary, which combines glucometer input, web-based records, and the ability to access records via a smart phone into one suite [9]. Although it has powerful analysis tools and enables the patient to do more selfassessment than with any other single method, it is still Windows-only and lacks some additional functionality that will be discussed in Sections II and III. WellDoc and t+ Medical also have mobile solutions for diabetes data management, but neither is as sophisticated or feature-rich as SiDiary [10], [11].

Though each solution has its own strengths, there are several common shortcomings that continue to make effective diabetes management a serious challenge. None of the above methods make recording diet convenient, requiring the patient to somehow deduce the nutrition facts of their food (which is nontrivial when eating at restaurants or where nutrition labels are not readily available). The patient’s only options are to either carry around a nutritional values book or write down everything they consume to check against books and labels later. Recording exercise is often neglected because paper log books are not on-hand and remembering to record the activity––let alone the specific time, duration, and intensity––is difficult. Instant feedback regarding trends and common reactions is missing in all cases, even though adjustment is a key factor in managing glucose levels. Even if all of the data were recorded by a patient, the doctor’s ability to make use of it is limited by the lack of a functional, intuitive interface that provides access to all of the data in a meaningful way. An effective system must address these deficiencies and focus on the most important facets of diabetes management: ease of data acquisition and recording, provision of instant feedback, and improvement of analysis capabilities.

II. Goals

As with any design, understanding the overarching goals before developing the actual components is critical. An improved diabetes management system should address the following objectives.

II.A. Improve Completeness of Data

The proposed system should increase the amount and improve the accuracy of patient data that is actually recorded. This includes, but is not necessarily limited to, the patient's glucose levels, insulin and other medication


4


dosages, exercise types, exercise durations, dietary intake, and ketone2 measurements (taken when glucose levels are too high), as well as the times at which these events take place. In addition to having the capability to record this data, the system should implement the easiest and most intuitive recording methods available, including avoiding manual input whenever possible. For glucose readings and insulin dosages (when a pump is used), this means automatic transmission of data from glucometers and insulin pumps to a records database. For data such as food intake and exercise, this means enabling users to input their data using minimal external resources (i.e. nutrition labels, nutrition factbooks, and exercise charts). Emphasis should be placed on enabling and encouraging the patient to record as much data as they have access to without impelling them to exert significantly more effort than before.

II.B. Provide Instant & Helpful Feedback

The proposed system should assist in the patient’s day-to-day management of diabetes, as well as improve the overall patient self-management education, through the delivery of instantaneous feedback. Beyond simply recording the data, the system should be usable as a tool to access otherwise unavailable or inconveniently accessed information. For example, the system should be able to assist the patient in determining the proper insulin bolus based on current blood glucose level, food intake, and level of activity. Additionally, use of the system should improve the patient’s understanding of how their own treatment plan is working and what is particular to their situation. The more a patient sees with regards to his or her treatment, the better prepared he or she will be to make adjustments in the future. An example of this is the learning of nutritional information. By making it easy for a patient to lookup and record the food they eat, they will more frequently see gradually internalize the nutrition facts of the foods they commonly consume. Such repetition is important for being able to make intelligent management decisions when resources are not readily available.

II.C. Facilitate Improved Analysis

The proposed system should allow both the patient and his or her doctor to perform both high-level and indepth analysis of the recorded data. This includes providing simple analysis metrics such as averages, highs and lows, and variance, as well as allowing the user to initiate more advanced investigations of trends and correlations. For example, a patient may desire to know the effect of a particular food or type of exercise on his or her glucose levels. By searching his or her database for instances involving this factor, the patient can easily isolate potential cause-effect trends and adjust accordingly. Doctors can similarly examine particular times of the day, reactions to a new insulin dosage, or any other trend they wish to observe. With the sheer volume of data being recorded, developing an intuitive and pragmatic interface for analyzing it is crucial.

II.D. Enable Large-Scale Medical Studies

As a peripheral goal to the improvement of patient management, the proposed solution has the potential to facilitate future medical studies involving diabetics. The technology allows for the passive acquisition of enormous amounts of data, covering a large, dispersed sample and promoting the examination of numerous variable interactions––essentially, an ongoing, real-time survey. With permission, any user’s data could be compiled anonymously into a database that would serve as an easily searchable sampling of diabetes patient data. Although a

2

Ketone bodies (more commonly referred to as simply ketones) are by-products produced when fat is broken down for energy. This

occurs in the absence of insulin, which normally allows cells to utilize glucose for energy. If ketones are generated in large amounts, a state called ketoacidosis––where the body’s pH decreases to a dangerously acidic level––may ensue, possibly leading to coma or even death. This is particularly dangerous for Type 1 diabetics, though Type 2 diabetics are also sometimes susceptible. Patients are typically instructed to test for ketones using a simple urine dipstick whenever blood glucose levels persist above 250 mg/dL.


5


secondary goal of this proposal, and obviously subject to several hurdles governed by the Health Portability & Accountability Act of 1996 (HIPAA), this is still a potentially powerful consequence of the technology.

III. Proposed Solution III.A. System Overview

The proposed integrated diabetes management system is comprised of several components: a glucometer with embedded wireless connectivity; an (optional) insulin pump with embedded wireless connectivity; a mobile communication device, such as a cell phone; a personal computer; and a database system, including patient records and other necessary data. Some of these components, such as the glucometer and the insulin pump, require only minor software and hardware upgrades to existing products. Others, such as the mobile device and computer, rely on the development of new software packages rather than alterations to the devices themselves. The centerpiece of the system is the Internet-accessible patient record database, which aggregates and makes available all of the desired data previously discussed. The other components serve to gather, disseminate, or analyze patient data, or to provide the user with necessary tools to assist in his or her diabetes management. The interactions between these devices are shown in Appendix B, and the component details are elucidated in Sections III.B to III.F below.

The basic operation of the system is as follows. The patient uses the glucometer to test his or her blood glucose level just as before. When the glucometer produces a reading, the reading is immediately sent via a wireless connection to the patient's mobile device as well as stored in an internal memory (to be transmitted automatically at a later time if the mobile device is not currently present). The mobile device, which is also connected to the patient's record database through a wireless Internet connection, simultaneously uploads the new data to the database and allows the user to access other database information, such as nutrition facts for food they may be consuming, exercise charts for their current activity, and records they have previously uploaded. The patient is prompted to input their meal and exercise data with the assistance of the software, and feedback is delivered instantly as to how much insulin is recommended (based on their custom treatment plan and current factors). If an insulin pump is in use, it automatically transmits the amount of insulin being delivered back to the mobile device. If injections are in use, the patient can manually enter the dosage. At any point, the patient can use his or her mobile device to access the record database for a quick analysis or a deeper look into previous activity. If the patient or doctor wishes to perform a more detailed analysis, including long-term trends and food- or exercise-specific queries, any personal computer with Internet access can be used. Because the database and analysis software is entirely web-based, no local software installation is necessary, and record sharing is automatic between patient and doctor. There are several other features of the system, which are elaborated upon in the following sections. Section III.G presents a variety of typical scenarios where the proposed system can provide dramatic improvements over the traditional management system.

III.B. Glucometer The glucometer (also referred to as a glucose meter or blood glucose monitor) used in the proposed system is similar to existing glucometers except it contains an embedded wireless communication module. Some meters, such as the BD Paradigm® Link, already use wireless technology to enable communication between the meter and an insulin pump [12]. These devices use either radio frequency (RF) or infrared (IR) technology to transmit data over a short range, typically less than 5 meters. The proposed system would instead use a new Bluetooth® specification called Wibree. Bluetooth® is an industry specification for short-range RF data transmission. It is ubiquitous in mobile phones, personal computers, and numerous peripherals, and allows for the easy "pairing" of devices and subsequent transfer of data. A common use for Bluetooth® is the syncing of contact information between mobile phones and PCs. After pairing the two devices, contact information is automatically synchronized between them whenever they are within range of each other. Wibree is a new wireless technology based on the Bluetooth® standard that is ultra-low power and has medical devices as a specific target application [13]. It compliments existing Bluetooth® modules,


6


which can interact with Wibree using a dual-mode chipset. By using Wibree (or even a standard Bluetooth® module), a glucometer could be paired with any mobile device or personal computer that Bluetooth®-enabled. Whenever a glucose reading is taken, for example, the data can automatically be sent to the user's cell phone without any physical (or line-of-sight, as required by IR) connection. The glucometer would still maintain an internal memory, so that the patient can still review basic records without computer or mobile device access and transmit data automatically when the device is back in range.

III.C. Insulin Pump If applicable, a patient's insulin pump would also be integrated into the management system. Although the majority of Type 1 diabetics still use daily injections to deliver insulin, pump use is becoming increasingly common. Many insurance companies are now recognizing pumps as an improvement over injections and are therefore covering the devices. As of 2008, there are an estimated 367,000 insulin pump users in the United States [14]. As mentioned above, some insulin pumps currently have the ability to receive wireless data transfer from glucometers. The proposed system would outfit pumps with a Bluetooth®/Wibree module so that it too could communicate with a mobile communications device. The insulin pump would be capable of transmitting data such as basal and bolus dosages, date and time of infusion set changes, and status information such as current battery and insulin levels. Warnings that are currently displayed on the pump's screen (such as "low reservoir" and "no delivery") could also be announced on the user's mobile device with customized vibrations and/or tones--a request that many pump users have asked for [15]. More advanced features of the system could incorporate the use of a mobile device as a remote control, allowing for the control of pump functions straight from the mobile phone platform described in Section III.D. Current pumps have low-resolution screens, with most being black and white. Accessing pump data and controls from a high-resolution, color screen on a mobile phone could lead to much easier operation.

III.D. Mobile Communication Device

The mobile communication device serves as the fulcrum of the management system, acting as both the

conduit of data transfer between the medical devices and databases and as the primary reference tool for patients. The key to maximizing the phone's utility is extending it beyond a basic record-keeping device. Acquisition of glucometer and insulin pump data through Bluetooth®/Wibree transmission is the first step in this development, as it automates tedious procedures. The next step is providing the user with otherwise unavailable services, such as immediate access to previous records and easy recording/look-up of food and exercise information, which must be accomplished using an intuitive mobile software application. For the purpose of this proposal, software development for use on a smartphone will be assumed. Development for entry-level cell phones is also important, however, and would involve the simplification or removal of some of the features described henceforth. Discussion of the smartphone market can be found in Section IV.B.

The application is accessed using a single click (or finger tap) of the icon on the phone's home screen. The

main menu of the application is simple and emphasizes the most commonly used features. Recording new entries and viewing the record database are anticipated to be the most frequently accessed functions; the other options on the main menu could easily be rearranged or replaced based on further user interaction studies. What matters most at this point is that the user is able to perform the desired tasks with as few clicks as possible. For now, only the record entry scenario will be graphically elucidated.

A mock-up of the data entry screen from this application can be found in Appendix C. Although the mockup is shown on an Apple iPhone this is only used as example hardware, and the demonstrated features are not proprietary to the specific platform (though input methods may vary based on hardware). The time is displayed at the top, followed by the current blood glucose reading (either automatically transmitted via Bluetooth®/Wibree or manually entered by the patient). To further streamline the recording process, the application can be set up to


7


automatically go to the recording screen whenever the phone detects a new glucometer reading, immediately displaying the current blood glucose level and prompting the user for other relevant data. It is believed that this would encourage patients to record more data due to the immediate reminder and minimization of input time required. Below the glucose level is the food entry section. This portion is spatially dynamic in that it can expand or contract with the amount of food being entered. When the user clicks on this region, they are able to type in the food they are eating. As they are typing, the application is accessing an online database of potential food matches, which the user can then select from and indicate portion sizes. This portion of the technology is key because it allows the patient to interface directly with extensive lists of foods without having to carry a book or visit a separate program or website. Some of these databases, such as Family Health Network's free Calorie King site, also include popular packaged food and meals found at chain restaurants [16]. Technology using direct access to such online databases has been demonstrated, for example in the iPhone web-based dietary diary application found in [17]. Food options are displayed as the user inputs the text, using pseudo real-time searching that mimics the predictive text completion found in other applications. This expedites the entry process, further encouraging its use. In addition, the repetition induced in seeing what is actually in the food will help the patient learn to judge meals when technology is not available. Again, a better flow of information to the patient can assist in long-term diabetes management education, and subsequently, better health decisions. Other input options, such as manual entry of new foods and quick access to a personalized list of commonly consumed items, could easily be implemented.

Below the food entry field is the exercise section, allowing the user to record an exercise event and duration. This can be done using a similar type of entry as the food, where the user begins to enter in an activity and the program instantly searches an online database (in this case, the database is relatively small and could actually be stored locally for quicker access), or using a list-based system that can dynamically present the most commonly selected exercise types first. Ann Doherty, Registered Nurse and Certified Diabetes Educator at the Alta Bates Summit Diabetes Center in Berkeley, CA, suggested the addition of a stress level indicator to the data form [18]. Stress has been widely reported to affect glucose levels due to the rapid release of hormones, and so keeping even a qualitative record of stress levels––particularly to indicate when stress is abnormally high––could provide useful insight to the patient or doctor during analysis [19]. A simple sliding scale is suggested as a preliminary input method. The suggested insulin bolus dosage is then displayed at the bottom. This dosage, based on the formula found in Equation (1), uses the data entered on the screen as well as sensitivity factors that the patient (or doctor) has previously entered into the system. This is sometimes called a “bolus wizard,” and is prominently featured on insulin pumps. For patients who use daily injections, a built-in bolus wizard that automatically calculates the prescribed dosage is a highly desired feature [20]. The actual dosage taken can be recorded either manually or via automatic transmission from an insulin pump. Finally, a section for comments is used to record anything that may need to be clarified or expounded upon (not shown in mock-up).

The mobile application will also have the ability to do record lookups and analysis, though it is believe that more in-depth examinations will be performed on a personal computer. As mentioned previously, the potential exists for using the mobile device as a remote control for insulin pumps, adding even more convenience. Integrating the device with a continuous glucose monitoring system––an in-situ device used to check glucose levels semicontinuously––would allow for even more data collection and the use of the phone as an alarm system if glucose levels are out of a safe range.

III.E. Personal Computer

For those without access to a mobile communication device, as well as to facilitate in-depth analysis, an application for use on personal computers will be developed. Like the mobile application, this will be kept up-to-date with records using the online patient record database. Additionally, to ensure that patients and doctors can access it from any computer with Internet access, the application will be web-based instead of locally installed. This is advantageous because it (a) frees the user from his or her own computer if not available, (b) enables use by


8


individuals who use public computers and do not have their own, and (c) allows for the creation of an operating system-independent application, eliminating the proprietary nature of existing solutions. Web-based applications that mimic desktop applications are already available and highly successful. Appendix D shows one of these applications––the Spreadsheets application in the Google Docs suite––being used for diabetes data management. In that particular instance, the user created his own method on a generic spreadsheet for easily recording and accessing data, as well as for sharing it with his doctor. The proposed system would instead build a custom interface geared specifically towards managing and analyzing diabetes data. The patient and doctor would be able to add new records, review and edit old records, and run numerous statistical analyses on any of the available data. One problem that doctors experience with current computerized systems is an overwhelming amount of data that is difficult to analyze. A more general overview of trends, shown graphically, is much more useful. When more information is desired (i.e. for a particular date, type of food, et cetera), a “drill-down” tool has been requested [20]. User interface design is the most critical part of this application, as making something too complicated renders it useless to a large portion of potential users.

III.F. Patient Record Database

The record database itself is merely the pool of data that is accessible to the user via the mobile or web interfaces. The key technical issues associated with the database itself are security and uptime. HIPPA limitations and authorization requirements must be addressed before such a system goes online; the technology, however, is already well-developed and proven.

III.G. Example Scenarios

Table 2 shows a variety of common diabetes management scenarios and compares traditional solutions with the solution proposed in this paper.

Table 2. Possible usage scenarios comparing management with and without the proposed system.

SCENARIO

Patient eating a meal

New HbA1c reading (and additional lab work) is taken

Doctor curious as to what is causing abnormal glucose readings at a particular time of day

WITH TRADITIONAL

WITH PROPOSED MANAGEMENT

MANAGEMENT

SOLUTION

Glucose level must be manually recorded. Nutrition information must be acquired from a book or nutrition labels, if available at all. Insulin dosage is manually calculated.

Glucose level automatically recorded. Patient able to quickly search food database to discover and record nutrition. Insulin dosage automatically calculated and recorded, all on single device.

Patient is notified of results via phone, fax or mail. Comparing glucose levels to HbA1c reading done manually. Patient asked to record more data and mail or fax it to office, which is often inconvenient and difficult for the doctor to interpret.

Patient automatically receives results on phone and PC. Results are synced with corresponding glucose data for easy comparison (may assist in the identification of abnormalities in HbA1c due to anemia [18]). Doctor able to query data for specific instances of abnormal glucose levels, viewing overall trends and possible causes.


9


IV. Development Plan IV.A. Plan

The development of the integrated diabetes management system will proceed in three phases: (1) the development of the web-based software and records database; (2) the development of the mobile application; and (3) the integration of Bluetooth®/Wibree modules into the glucometer and insulin pump. This will allow for the creation of incremental technologies that are both testable and useful on their own (for example, manually entering data using stages (1) and (2) will still improve the recording and analysis of data). The first two stages will require a choice of platform on which to develop. Because each smartphone uses a different operating system (Symbian OS, Windows Mobile, Palm OS, OS X, BlackBerry, Android, et cetera), a universal approach is preferred. At the moment, Java is believed to be the best potential environment for mobile development. The team is also looking into leveraging the power of an online-offline tool, such as Google Gears, which permits devices to run web-based applications natively when not connected to the internet.

Extension of the project beyond the scope of diabetes management is possible and greatly desired. Although the authors’ personal experiences have led to the selection of diabetes as the initial application, this technology and information framework could very easily be applied to other medical IT scenarios: tracking drug efficacy, exercise programs, or other embedded medical devices, for example. Exploration of these applications will not be limited by the initial goals described in this paper, and the team plans on pursuing these avenues as the project progresses.

IV.B. Market Feasibility

One of the biggest improvements associated with this management system is the power of the mobile communication device as a platform to both acquire and deliver useful data using something the patient already has. The ubiquity of cell phones , the majority of which have Bluetooth® and the capability of Internet access, makes their integration into a management strategy an obvious choice. Smartphones, such as the Motorola Q, the Nokia Nseries, the Apple iPhone, Research In Motion's BlackBerry, and the Palm Treo, represent a new level of functionality for mobile phones (examples shown in Appendix E). Beyond the basic cell phone capabilities such as making calls and sending text messages, these phones support advanced software applications that transform the mobile platform into a powerful computing device. High-speed wireless Internet access, bright and high-contrast screens, and full QWERTY or touch-screen keyboards make smartphones an ideal development platform for this management solution. Penetration of smartphones beyond the enterprise community is growing rapidly. In the US, 20.9 million smartphones were shipped in 2007––a 100% increase over 2006 [21]. Gartner predicts that 173 million smartphones will be sold in 2008, and and that by 2010, the total sales will break the 1 billion unit mark [22]. This gradual replacement of traditional consumer cell phones with smartphones makes developing the application for use on smartphone platforms a reasonable decision. An application for entry-level phones can and should be developed in parallel, with only minor feature reductions necessary.

IV.C. Potential Collaborators

Although the team has extensive experience in design, computer science, and diabetes management,

professional and industrial collaboration has distinct advantages. Continued collaboration with the Alta Bates Summit Diabetes Center, as well as with endocrinologists in the area, will help ensure that the software is easy to use and powerful for both patients and doctors. Sarah Beth Eisinger's position within Google may foster a cooperative effort within the application development and database solution areas. A potential partnership with Medtronic MiniMed, the largest manufacturer of insulin pumps in the United States, is being sought. Additionally, the team would like to add a student from the UC Berkeley Department of Public Health to assist in the areas of health records management and public dissemination of medical technology.


10


References [1] Narayan, K. M. V., et al. “Diabetes –– a common, growing, serious, costly, and potentially preventable public health problem,” Diabetes Research and Clinical Practice, 50 Suppl. 2, 2000. [2] World Health Organization, “Diabetes,” Fact Sheet No. 312, September 2006. Available Online: http:// www.who.int/mediacentre/factsheets/fs312/en/index.html [3] United States Center for Disease Control and Prevention, “National Diabetes Fact Sheet,” 2005. Available Online: http://apps.nccd.cdc.gov/ddtstrs/template/ndfs_2005.pdf [4] Engelgau, M. W., et al. “The Evolving Diabetes Burden in the United States,” Annals of Internal Medicine, Vol. 140, No. 11, June 2004. [5] Vijan, S., Hofer, T. P., and Hayward, R. A., “Estimated Benefits of Glycemic Control in Microvascular Complications in Type 2 Diabetes,” Annals of Internal Medicine, Vol. 127, Issue 9, November 1997. [6] Nathan, D. M., et al. “Intensive Diabetes Treatment and Cardiovascular Disease in Patients with Type 1 Diabetes,” The New England Journal of Medicine, Vol. 353, No. 25, December 2005. [7] Glucose Meter Comparison Chart, Focus Pharmacy, Accessed April 7, 2008. Available Online: http:// www.focuspharmacy.com/acatalog/meter_comparison_chart.html [8] Johnson & Johnson’s LifeScan’s OneTouch® UltraSmart® product page, Accessed April 2, 2008. Available Online: http://www.lifescan.com/products/meters/ultrasmart/ [9] Sinovo’s SiDiary product page, Accessed March 4, 2008. Available Online: http://www.sidiary.org/ [10] WellDoc product page. Accessed March 24, 2008. Available Online: http://www.welldoc-communications.com/ index.html [11] t+ Medical’s t+ diabetes product page, Accessed March 24, 2008. Available Online: http:// www.tplusmedical.com/us/patientDiabetes.html [12] Medtronic MiniMed’s BD Paradigm® Link product page. Accessed April 2, 2008. Available Online: http:// www.minimed.com/products/insulinpumps/components/paradigmlink.html [13] Nokia’s Wibree Technology Datasheet. Accessed April 1, 2008. Available Online: http://www.wibree.com/ technology/Wibree_2Pager.pdf [14] Pham, M. “Medtronic Diabetes: Sizing the market for real-time, continuous blood glucose monitors from MDT, DXCM, and ABT,” HSBC Global Research, June, 2006. Available Online: http://www.research.hsbc.com/midas/Res/ RDV?p=pdf&key=ate2b8rygn&name=127057.PDF [15] von Wartburg, L., "Your Insulin Pump Proposals: What You Want the Manufacturers to Change," Diabetes Health Professional, Vol. 17, No. 1, February 2008. [16] Family Health Network’s Calorie King food database product page. Accessed April 2, 2008. Available Online: http://www.calorieking.com/ [17] 4Technologies My Net Diary iPhone product test drive page. Accessed April 4, 2008. Available Online: http:// mynetdiary.com/iphone.do [18] Doherty, A. (Registered Nurse & Certified Diabetes Educator, Alta Bates Summit Diabetes Center, Berkeley, CA). Personal Communication. April 2, 2008. [19] Cox, D., et al. “Effects of acute experimental Stressors on insulin-dependent diabetes mellitus,” Paper presented at the meetings of the American Diabetes Association, New Orleans, LA, 1988.


11


[20] Eng, G. (Endocrinologist, Associated Internal Medical Group, Oakland, CA). Personal Communication. March 6, 2008. [21] "Smart mobile device shipments hit 118 million in 2007, up 53% on 2006," Canalys Research Release 2008/021, February 2008. Avaiable Online: http://www.canalys.com/pr/2008/r2008021.pdf [22] Milanesi, C. "Key Issues for Mobile Devices, 2008," Gartner Research, ID No. G00156657, April 2008.

Member Profiles Christopher Hannemann is a graduate student in the Department of Mechanical Engineering at the University of California, Berkeley. He received his BS ME with highest honors from the Georgia Institute of Technology in 2006. His graduate research focuses on energy efficiency in data centers, and he has previously worked on projects in geothermal energy, offshore structure vibration analysis, and recyclable polymer cardboard coatings. Chris was diagnosed with Type 1 diabetes at the age of eight, and has been on insulin therapy ever since. Chris's younger sister, Victoria, also has Type 1 diabetes; his father, Kim, has Type 2.

Sarah Beth Eisinger is a graduate of the University of California, Berkeley, having received her BS in Electrical Engineering and Computer Science with an emphasis in computer graphics in 2007. She is presently a Software Engineer at Google Docs, specializing in web application development.

Acknowledgements

The authors would like to acknowledge several contributors for their assistance in the development of this project: Dr. Grace Eng, for her professional knowledge and guidance; Ann Doherty, for her input as an educator in diabetes management; Brendan Hannemann, for his help with mobile platforms and databases; and Victoria Hannemann, Kim Hannemann, and Janet Hannemann, for their first-hand experiences and advice on diabetes management.

Proposed Plan for Competition Award Money

Award money will be used for the purchase of necessary materials to develop a prototype of the proposed

system according to the phases outlined in Section IV.A. This includes a smartphone, Bluetooth®/Wibree modules, software for developing the mobile and web-based interfaces, and server space. During the initial development phase, a phone contract will not be necessary to test the interface; a used smartphone may therefore be considered in an effort to maximize resources. The Alta Bates Summit Diabetes Center has been generous enough to donate a variety of glucometers, which can be used for research and prototyping purposes. Integration into insulin pumps will not occur at this stage, as it will require cooperation with a pump manufacturer.

Contact Information & Authorization Primary Contact: Christopher R. Hannemann phone: (703) 402 - 7064 [email protected] The authors consent to public, online dissemination of this paper.


12


This sample log book from OneTouch® accompanies the OneTouch® UltraMini glucometer and is representative of typical paper log books.

Appendix A. OneTouch® Log Book


13


Appendix B. Schematic of Integrated Diabetes Management System


14


Appendix C. Mobile Device Application Mock-Up The mock-up below shows the functionality of the mobile communication device software associated with the proposed integrated diabetes management system. Although the actual interface is not represented here, the mockup does highlight the data that can be acquired through this system. Although the design is shown on an Apple iPhone, this is merely one of many possibilities and should not influence development.


15



16


and doctors (i.e. more drill-down” tools, better organization options, several more graphs showing daily, monthly, and yearly trends, et cetera).

factors. The interface uses only basic spreadsheet functionality, whereas as more advanced system would be more specific to the needs of patients

The data was useful in determining time-of-day trends, problematic activities and foods, and appropriate insulin and carbohydrate sensitivity

insulin dosages, food intake, and activities as he transitioned from NPH to Lantus insulin, and then from injection to the use of an insulin pump.

This Google Docs spreadsheet was created and used by team member Chris Hannemann during the summer of 2007 to record his glucose levels,

Appendix D. Example of Web-Based Diabetes Data Tracking


17

Symbian OS, OS X, BlackBerry, and Palm OS systems, respectively.

From left to right, the Motorola Q, Nokia N95, Apple iPhone, RIM BlackBerry Pearl, and Palm Treo, running Windows Mobile,

Appendix E. Smartphones
