Tele-Medika: a data mining empowered web portal for online health

Abstract

The Swinfen Charitable Trust (SCT) provides an e-referral service for health practitioners in developing countries. Referral messages and responses are passed by email. We have developed a Web-based system to facilitate information retrieval from these messages and multimedia data mining. The system contains a search engine which implements full text searching, together with nearness-based retrieval. In pilot testing, cases with relatively high nearness ranking were reasonably near to the query cases. In addition, a Support Vector Machine (SVM) was used to classify cases into categories of gender and age. False positives did not increase significantly with the number of true positives, i.e. this represented very good classification performance. The system we have developed appears to be a promising way of storing and retrieving email-based medical correspondence.