2002

Empowering the patient: How to provide patients with personalized medical information

Empowering the Patient: How to provide patients with personalized medical information

.G.Bellika, Norwegian Centre for Telemedicine E.Bønes, Department of Computer Science, University of Tromsø G.Hartvigsen, Norwegian Centre for Telemedicine/Department of Computer Science, University of Tromsø J.Norum, Norwegian Centre for Telemedicine S.Skorpen, Norwegian Centre for Telemedicine

Abstract

Through the Internet, we may access an enormous amount of medical information. By accessing some of this information, and learning about their medical situation and treatment alternatives, patients are expected to play a more active role in their medical care.

PaSent is a personal health advisor that regularly locates relevant medical information on the Internet. Through the use of neural networks the systems matches the contents of a patient's medical record or other medical data with potential relevant information located on the Internet. The most relevant information is then presented to the user. In this way, PaSent can be described as an intelligent system aimed at providing patients with personalised healthcare and information.

Given the huge amount of data available on the Internet, it is necessary to provide a tool that extract relevant medical information. Today, patients can use search engines like Google and Alltheweb, or they may simply visit health portals like doctor online (www.doctoronline.no). In the first case, you often get an large amount of mostly irrelevant data. In your attempt to restrict the search, important information could be excluded. In the latter case, narrowing your search to one source could exclude lots of relevant and important information.

The major goal of the PaSent project is to construct a personal health adviser - a virtual secretary. As a patient you want to tell your secretary to get all relevant information as regards your own EPR. Or you want to say to your secretary: "This is the information I found on doctor online. Could you please find more information on this matter?" Or: "This news article describing disease X is very relevant for my own illness. Could you please find out whether there is more information available?"

In this presentation, we outline the system architecture, demonstrate the potential use of PaSent and present a detailed analysis of the accuracy of the search system. The tests have been conducted by PaSent. It looked for specific information based on extracts of a patient record and compared the results with similar results obtained by manual search. We will also locate the most relevant data from the manual search and compare the result with PaSent's ranking.

---