抄録

○梶栗芳夫・甲斐瑛子（九大）・Ahmed Ashir（九大/Grameen Communications）・Islam Rafiqul（Grameen Communications）・金子邦彦・福田　晃（九大）

Kyushu University and Grameen Communication have developed a portable health clinic to serve the unreached community with low resource settings. A local server collects the health status of a patient by using multiple medical sensors and send them to the remote urban doctor for clinical decision. In this process, the measured data is manually inserted in the local server by local health assistants. Any errors in the data, will produce wrong diagnosis of a disease. It is a huge manual task to check thousands of data manually and validate them. In this work, we statistically analyzed the data to discover errors from outliers. Errors can contain in inliers too. We are now finding appropriate techniques to discover errors from inliers too.