The Journal on Information Technology in Healthcare 2005; 3(5): 307 313 A Telematic System for Diabetes Management, Reporting and Patient Advice Dimitra Iliopoulou*, Kostas Giokas*, Stavroula Mougiakakou, John Stoitsis, Andriana Prentza*, Konstantina Nikita * Biomedical Engineering Laboratory, Biomedical Simulations and Medical Imaging Laboratory, National Technical University of Athens, Greece. ABSTRACT This paper presents a telematic system for providing advice and assisting in the management and reporting of patients with Type 1 diabetes. The system is comprised of four distinct entities: devices for measuring blood glucose levels; an insulin advisory system; a mobile phone application and a diabetes patient management and reporting system. The system integrates wireless personal area networks with the use of mobile and Internet technologies and commercially available measurement devices, along with online analytical processing techniques and reporting tools. In addition to its clinical applications, the proposed system can be used by healthcare managers to evaluate healthcare provision, and for demographic analysis to estimate diabetes management costs. INTRODUCTION Diabetes mellitus is a common chronic disease, characterised by blood glucose levels outside the physiological range. Diabetic patients can be classified as Type 1 where there is an absence of insulin secretion, or Type 2 where the underlying defect is insulin resistance. It is estimated that there are currently more than 150 million diabetic patients world-wide and this figure is predicted to double to over 300 million by the year 2025. Diabetes is associated with both short- and long-term complications, and prevention of these by strict glycaemic control is a major challenge. The use of telemedicine offers significant potential for improving management of the disease 1 4. In this paper we present a system for continuous and remote monitoring and management of diabetes. The system, which includes increased management, advisory, and reporting capabilities, is comprised of four distinct entities: (i) Devices for measuring blood glucose levels (ii) Insulin Advisory System (IAS) Correspondence and reprint requests: D. Iliopoulou, 9 Iroon Polytechniou Str., Zografou Campus, Athens Greece, 15773. E-mail: dilio@biomed.ntua.gr. 307
Iliopoulou, Giokas, Mougiakakou, Stoitsis, Prentza & Nikita (iii) Mobile phone application (iv) Diabetes Patient Management and Reporting System (DPMRS) SYSTEM DESCRIPTION The architecture of the proposed system is shown in Figure 1. The four distinct entities are as follows: A. Devices for Measuring Blood Glucose Levels The system is designed to utilise commercially available blood glucose meters supporting radio frequency (RF), infrared (IR), and/or wired connection methods for data transmission. Values obtained from the meters are inputted into the Insulin Advisory Service (IAS) and sent to the DPMRS for storage and further analysis. B. Insulin Advisory System (IAS) The IAS is designed for patients that use insulin pumps or multiple insulin injection protocols. It consists of a physiological model that aims to simulate glucose-insulin metabolism in Type 1 diabetes patients. The IAS utilises compartmental models, mathematical models, and artificial intelligence techniques 5. In addition to the patient s weight, other inputs into the IAS are: Time and measured blood glucose levels Time and amount of previous insulin injection Time and amount of carbohydrate intake The outputs of the IAS are the appropriate insulin intake: For insulin injections types and doses of insulin For insulin pumps basal infusion rate and bolus doses Figure 1. Architecture of the proposed system DBMS = Database Management System, DPMRS = Diabetes Patient Management and Reporting System, GPRS = General Packet Radio Service, http = Hypertext Transfer Protocol, OLAP = On Line Analytical Processing 308 The Journal on Information Technology in Healthcare 2005; 3(5): 307 313
A Telematic System for Diabetes Management, Reporting and Patient Advice In the event that the output results are outside pre-assigned limits set by the physician, an alarm signal is sent to the patient s physician to approve or disapprove the recommended changes. In both cases the physician sends an SMS (Short Message Service) text to the patient s mobile with his recommendation. C. Mobile Phone Application The mobile phone application is responsible for establishing communication between the mobile phone and a service provider using General Packet Radio Service (GPRS), and between the mobile phone and the glucose measurement devices, using wired or wireless, e.g. radio frequency (RF) connections. Measured glucose levels are transmitted, through the mobile phone, in predefined time intervals to the DPMRS. The mobile phone application also allows patients to submit other relevant information to the DPMRS, e.g. time and amount of injected insulin, diet, weight, physical activity, special events (stress, illness) etc. D. Diabetes Patient Management and Reporting System (DPMRS) The DPMRS is the core of the system and contains a distributed Database Management System (DBMS) along with advanced tools for intelligent processing of available data from diabetic patients (Figure 2). The system is installed on a Web server that can be located in a medical centre, and/or home-care provider, and/or diabetes institute. Access to the server is granted to both diabetes patients and the health professionals caring for them. Important considerations to be taken into account in the construction of the DBMS include: the source and the type of diabetes data; the security of the system; the tools for data processing and the type of data distribution. Figure 2. An analysis tool such as Business Objects uses multimode OLAP techniques The Journal on Information Technology in Healthcare 2005; 3(5): 307 313 309
Iliopoulou, Giokas, Mougiakakou, Stoitsis, Prentza & Nikita Figure 3. An intelligent decision support system Measured blood glucose levels are transmitted automatically from the measurement device to the DPMRS, while additional patient information (sex, age, years with diabetes, weight, other diseases, pregnancy, unusual stress etc), and habits (time and insulin dosage, time and description of food intake, duration and type of physical exercise) are manually inputted by the patient using a mobile phone that supports Java applications. Data is downloaded regularly from the mobile phone to the DPMRS. Other data stored in the DPMRS includes health professional records, laboratory examination results and comments about the patient s state of health. In order to draw helpful results for the large amount of data sent to the DPMRS on a daily basis, an underlying reporting and analysis sub-system is used. As shown in Figure 3, this system also acts as an interactive decision support making tool for patients, physicians and healthcare managers. Generally, systems that involve storage and retrieval of large volumes of data are known as data warehouses 6. The purpose of a data warehouse is to enable rapid access to information for analysis and reporting for both medical experts (physicians and healthcare managers) and diabetic patients. In this specific application the data warehouse is an online repository of data from diabetic patients that is necessary for the decision making process. The data is analysed using On Line Analytical Processing (OLAP) 7 9. OLAP takes a snapshot of the relational database and restructures it into dimensional data. Queries can then be run against this permitting Fast Analysis of Shared Multidimensional Information. It has been claimed that for complex queries OLAP can produce an answer in around 0.1% of the time for the same query on relational data. An OLAP structure created from the operational data is called an OLAP cube. The cube is created from a star schema of tables. At the centre is the fact table that lists the core facts which make up the query. Numerous dimension tables are linked to the fact tables. These tables indicate how the aggregations of relational data can be analysed. The number of possible aggregations is determined by every possible manner in which the original data can be hierarchically linked. 310 The Journal on Information Technology in Healthcare 2005; 3(5): 307 313
A Telematic System for Diabetes Management, Reporting and Patient Advice Table 1. Explanation of OLAP Terms Measures, Dimensions and Cubes Schema Measures are key performance indicators that require evaluation. Numbers that would make sense if aggregated are usually measures (number of patients, amount of injected insulin, time oriented aggregates). Dimensions are the categories of the data analysis and are arranged in hierarchical levels with unique positions that in turn are made up by members (patient sex, smoking habits, etc). Putting measures and multiple dimensions together we can create a data cube that can be rotated according to the wish of the user. The dimensions and measures are physically represented by a star schema. The most basic star schema arranges the dimension tables around a central fact table that contains the measures. In the case of DPMRS, OLAP permits the physician to rapidly make decisions related to the health status of their Type 1 diabetes patients. Using OLAP it is possible to leverage existing data from a relational schema or data warehouse (data source) by placing key performance indicators (measures) into context (dimensions) as explained in Table 1. Measured results submitted by diabetic patients, e.g. time date, number of meals, number of insulin injections, etc., can be estimated. Dimensions include aggregations by time, such as weekly, monthly, quarterly or yearly aggregates. Once processed into a multidimensional database (cube), as shown in Figure 4, all of the measures are pre-aggregated to make data retrieval significantly faster. The processed cube can be made available to business users of a health institution or physicians themselves, who can browse the data using a variety of tools, and make ad hoc analysis an interactive and analytical process rather than a development effort. Figure 4. Multidimensional cube The Journal on Information Technology in Healthcare 2005; 3(5): 307 313 311
Iliopoulou, Giokas, Mougiakakou, Stoitsis, Prentza & Nikita a. OLAP Query Types Slice-and-dice queries make selections to reduce a cube. For example, the cube in Figure 4 can be sliced to consider only diabetic patients in Athens, and then further sliced to consider only patients for the year 2003. Selecting a single dimension value reduces the cube s dimensionality, but more general selections are also possible. Drill-down and roll-up queries are inverse operations that use dimension hierarchies and measures to perform aggregations. Rolling up to a top value corresponds with omitting the dimension. For example, the relevant values for the city aggregates Athens, Salonica, Patras, etc. can be rolled into a single value Greece. These values can indicate overall patient numbers, number of insulin injections or any other relevant measure. Drill-across queries combine cubes that share one or more dimensions. In relational algebraic terms, this operation performs a join. Ranking or top/bottom queries can return only those cells that appear at the top or bottom of the specified order, for example, the most active insulin intake patient group in 2004. Rotating a cube allows users to see the data grouped by other dimensions. b. Achieving Fast Results Healthcare professionals must be able to obtain fast and accurate results in order to make decisions. To this end, essential performance-enhancing techniques, such as pre-computation and pre-aggregation can be used for multidimensional databases. These techniques permit response times to be fast enough to allow interactive data analysis for queries involving potentially huge amounts of data. Modern OLAP systems adopt the practical pre-aggregation approach of materialising only selected combinations of aggregates, and then reusing these to efficiently compute other aggregates. Reusing aggregates requires a well-organised multidimensional data structure. c. Software The proposed DPMRS is built using Microsoft s SQL (Structured Query Language) Server as the storage database along with SQL Server 2000 Analysis Services. In order to create and present detailed analysis reports the system would greatly benefit from applications like Business Objects V11 for query and analysis and Crystal Reports V11 for reporting. DISCUSSION This paper presents a system for the monitoring, analysis and management of Type 1 diabetes mellitus. The system is based on state-of-the-art technologies for data communication using wired, wireless and mobile network infrastructures. 312 The Journal on Information Technology in Healthcare 2005; 3(5): 307 313
A Telematic System for Diabetes Management, Reporting and Patient Advice It ensures portability and flexibility and should help increase the quality of life of diabetes patients through seamless monitoring of blood glucose levels. Furthermore, the use of state of the art storing and retrieving database technologies, along with the ability to create complex ad hoc reports, maximise the benefits of information management and handling for both the patients and the physicians. It is envisaged that the proposed system will permit optimisation of the treatment of Type 1 diabetes patients through intensive monitoring of blood glucose levels, and improvement of healthcare provision and healthcare organisation through advanced analysis of medical data. REFERENCES 1 Belazzi R, Larizza C, Montani S, et al. A telemedicine support for diabetes management: the T-IDDM Project. Computers Methods and Programs in Biomedicine 2002; 69: 147 61. 2 Bellazzi R, Montani S, Riva A, Stefanelli M. Web-based telemedicine systems for home-care: technical issues and experiences. Computers Methods and Programs in Biomedicine 2001; 64: 175 87. 3 Liao YT, Tang ST, Chen TC, et al. A communication platform for diabetes surveillance. Proceeding of the 26th Annual International Conference of the IEEE EMBS, San Francisco, CA, USA, 2004. 4 Gómez J, Hernando ME, García A, et al. Telemedicine as a tool for intensive management of diabetes: the DIABTel experience. Computers Methods and Programs in Biomedicine 2002; 69: 163 77. 5 Mougiakakou SG, Nikita KS. Blood glucose profile prediction for Type 1 diabetes patients using a hybrid approach. Proceedings of the 2nd European Medical and Biological Engineering Conference, Vienna, Austria, 2002, pp. 1160 61. 6 Kimball R. The Data Warehouse Toolkit: Practical Techniques for Building Dimensional Datawarehouses. New York: John Wiley & Sons. Inc, 1996 7 Codd EF, Codd SB, Salley CT. Providing OLAP (On-Line Analytical Processing) to User- Analysts: An IT Mandate. http://www.hyperion.com/solutions/whitepapers.cfm. 8 Chauduri S, Dayal U. An overview of data warehousing and OLAP technology. SIGMOD Record 1997; 26: 65 74. 9 Winter R. Databases: back in the OLAP game. Intelligent Enterprise Magazine 1998; 1: 60 64. The Journal on Information Technology in Healthcare 2005; 3(5): 307 313 313