SOMA: A Proposed Framework for Trend Mining in Large UK Diabetic Retinopathy Temporal Databases
In this paper, we present SOMA, a new trend mining framework; and Aretaeus, the associated trend mining algorithm. The proposed framework is able to detect different kinds of trends within longitudinal datasets. The prototype trends are defined mathematically so that they can be mapped onto the temporal patterns. Trends are defined and generated in terms of the frequency of occurrence of pattern changes over time. To evaluate the proposed framework the process was applied to a large collection of medical records, forming part of the diabetic retinopathy screening programme at the Royal Liverpool University Hospital.
KeywordsDiabetic Retinopathy Frequent Pattern Time Stamp Logic Rule Support Threshold
Unable to display preview. Download preview PDF.
- 1.Somaraki, V., Broadbent, D., Coenen, F. and Harding, S.: Finding Temporal Patterns in Noisy Longitudinal Data: A Study in Diabetic Retinopathy. Proc. 10th Ind. Conf. on Data Mining, Springer LNAI 6171, pp 418-431 (2010).Google Scholar
- 4.Kanski, J.: Clinical Ophthalmology: A systematic Approach. Butterworth-Heinemann/Elsevier (2007).Google Scholar