This document is about the candidate's Master thesis concerning a handheld medical Decision Support System (DSS) for early diagnosis of heart attack. The purpose of such a DSS is to provide faster identification of patients with a critical coronary condition as a heart attack is, as well as obtaining better patient selection for admittance to CCU-units. Fast selection means faster treatment. Another benefit is the possible elimination of unnecessary treatment of patients not actually suffering from a heart disease. The advantage of a handheld system is that it is able to operate almost everywhere, outside of a medical environment such as a hospital. The system can easily be brought to the patient, instead of the other way around. In addition, it may seem less intimidating for a patient to be diagnosed by a doctor using a small, discrete device, compared with that of being questioned by one stationed at a desk, or sitting by a laptop computer.
The object of this thesis was to port an existing DSS from being PC-based, to being able to be run on a handheld device. For this, some decisions concerning platform, programming language and environment had to be made. It had to be tested on a physical device, to see that it operated as expected. In the end, the system was programmed in the Java API (Application Programming Interface) collection called J2ME. It was implemented on an emulator called the Wireless Toolkit, provided by Sun Microsystems. In addition to programming the application, I provide some medical background for heart disease, as well as the role of a DSS in diagnosing it. I will describe the DSS, and give some insight into how the original algorithm, based on regression analysis, performs. I will dip shortly into the realm of regression analysis, and attempt to explain how the system works. I will also refer to and describe some of the previous studies, which are the forerunners leading up to this work. There will be some information about the technical issues involved with implementing the handheld system. I will describe the result, its graphical user interface and the underlying system, the nuts and bolts holding it all together. Towards the end I will relate to some of the problems I encountered, and how these were (or weren t) solved. Towards the end there will be some notes concerning testing, and a short evaluation of the application. Lastly, I suggest some concluding remarks about what further work can be done.