Advances in Temporal Data Mining for Clustering, Classification, and Prediction Purposes
Monitoring, interpretation, and analysis of large amounts of time-stamped data are subtasks that are at the core of tasks such as the detection of malware in communication data, the integration of homeland security information from multiple sources, the management of chronic patients using clinical guidelines, the retrospective assessment of the quality of the application of such a guideline, and the learning of new knowledge from analyzing the data regarding repeating patterns of process-related actions, of measured data, and of meaningful abstractions derivable from these data.
I will briefly describe several conceptual and computational architectures developed over the past 20 years, mostly by my research teams at Stanford and Ben Gurion universities, for knowledge-based performance of these tasks, and will highlight the complex and interesting relationships amongst them. I will highlight the differences and similarities between interactive visual-exploration frameworks for single and multiple longitudinal records, and data-driven frameworks for temporal data mining, and how both can be used and even integrated for clustering, classification, and prediction. I will focus on the medical domain.
I will also point out the progression, exemplified through the case of the medical domain, from individual-subject monitoring, diagnosis, and therapy, to multiple-subject aggregate analysis and research, and finally to the learning of new knowledge from large numbers of longitudinal records.