A high performance k-NN approach using binary neural networks.

This paper evaluates a novel k-nearest neighbour (k-NN) classifier built from binary neural networks. The binary neural approach uses robust encoding to map standard ordinal, categorical and numeric data sets onto a binary neural network. The binary neural network uses high speed pattern matching to recall a candidate set of matching records, which are then processed by a conventional k-NN approach to determine the k-best matches. We compare various configurations of the binary approach to a conventional approach for memory overheads, training speed, retrieval speed and retrieval accuracy. We demonstrate the superior performance with respect to speed and memory requirements of the binary approach compared to the standard approach and we pinpoint the optimal configurations.

Health care resource use and stroke outcome. Multinational comparisons within the GAIN International trial.

BACKGROUND AND PURPOSE: Outcome in patients hospitalized for acute stroke varies considerably between populations. Within the framework of the GAIN International trial, a large multicenter trial of a neuroprotective agent (gavestinel, glycine antagonist), stroke outcome in relation to health care resource use has been compared in a large number of countries, allowing for differences in case mix. METHODS: This substudy includes 1,422 patients in 19 countries grouped into 10 regions. Data on prognostic variables on admission to hospital, resource use, and outcome were analyzed by regression models. RESULTS: All results were adjusted for differences in prognostic factors on admission (NIH Stroke Scale, age, comorbidity). There were threefold variations in the average number of days in hospital/institutional care (from 20 to 60 days). The proportion of patients who met with professional rehabilitation staff also varied greatly. Three-month case fatality ranged from 11% to 28%, and mean Barthel ADL score at three months varied between 64 and 73. There was no relationship between health care resource use and outcome in terms of survival and ADL function at three months. The proportion of patients living at home at three months did not show any relationship to ADL function across countries. CONCLUSIONS: There are wide variations in health care resource use between countries, unexplained by differences in case mix. Across countries, there is no obvious relationship between resource use and clinical outcome after stroke. Differences in health care traditions (treatment pathways) and social context seem to be major determinants of resource use. In making comparisons between countries, great care should be exercised in using outcome variables as indicators of quality of stroke care.