Privacy aware K-means clustering with high utility

Privacy-preserving data mining aims to keep data safe, yet useful. But algorithms providing strong guarantees often end up with low utility. We propose a novel privacy preserving framework that thwarts an adversary from inferring an unknown data point by ensuring that the estimation error is almost invariant to the inclusion/exclusion of the data point. By focusing directly on the estimation error of the data point, our framework is able to significantly lower the perturbation required. We use this framework to propose a new privacy aware K-means clustering algorithm. Using both synthetic and real datasets, we demonstrate that the utility of this algorithm is almost equal to that of the unperturbed K-means, and at strict privacy levels, almost twice as good as compared to the differential privacy counterpart.

The effectiveness of a brief intervention using a pedometer and step-recording diary in promoting physical activity in people diagnosed with type 2 diabetes or impaired glucose tolerance

Issue addressed: To evaluate the effectiveness of a brief intervention using a pedometer and step-recording diary on promoting physical activity in people with type 2 diabetes or impaired glucose tolerance (IGT). Methods: People with type 2 diabetes or IGT who attended the Illawarra Diabetes Service were invited to participate. Participants in the intervention group received a pedometer and a diary to record their daily steps for a two-week period. Both the intervention and comparison group received advice on physical activity. Physical activity levels were measured using the Active Australia Survey at baseline, and at two and 20 weeks. Results: A total of 226 participants were recruited. At two-week follow-up the mean self-reported minutes of walking was significantly higher in the intervention group than the comparison group (223 minutes versus 164 minutes; p=0.01), as was the percentage of intervention participants achieving recommended levels of moderate-intensity physical activity (63.5% versus 41.8%, p=0.02) and the percentage of intervention participants achieving adequate levels of total physical activity (68.9% versus 48.0%, p=0.04). There were no differences between study groups for any physical activity measure at 20-week follow-up. Conclusions: A pedometer and a step-recording diary were useful tools to promote short-term increase in physical activity in people diagnosed with type 2 diabetes or IGT. Future studies need to examine whether a longer intervention, individualised physical activity counselling and support for achieving step goals could result in increasing physical activity over the long term.