Meeting the challenges of effective pandemic detection and response requires addressing several key threads of data mining research: constructing models of a system in terms of time-varying spatial-social networks, mining spatial-social-temporal structures capturing significant properties of the disease spread, and actively sampling/controlling to observe and respond. The SIAM-DM 2007 Workshop on Data Mining for Pandemic Preparedness aims to provide a forum for such an exchange. Example issues that are of topical interest to the workshop include:

• Algorithms for fast computations of multi-level network properties induced by spatial-social data; in particular, provable approximations for estimating expansion, between-ness, and community structure, and tracking such properties across time-indexed snapshots.
• Integrating model-driven methods with spatial mining, e.g., combining a model for disease spread with a method for detecting critical individuals and locations; or using data mining to derive a social distancing policy or to formulate quarantine procedures.
• Disaggregation on demand, i.e., determining a small set of multi-level aggregates (among the multitude of possibilities) to be stored as sufficient statistics, thus allowing other microscopic parameters to be re-generated on demand.
• Co-evolving epidemic policy, simulation, and mining; unlike passive observation of data to derive targeted sampling policies that model a static dataset, implementing an intervention policy can fundamentally change the course of future simulation runs, thus making data mining an integral part of the simulation-based model.
• New objective functions for active data mining, that mimic targeted detection and targeted vaccination goals in epidemiological modeling, and, in this manner, close the monitor-simulate-mine loop.
• Support the view of simulation models as procedural representations of large datasets; thus allowing the rich modeling literature to be harnessed for data mining goals.

Paper Submission and Review

We are soliciting papers that respond to the three general themes discussed above (modeling, mining, and active sampling) in data mining for pandemic preparedness. While we encourage the use of the provided pandemic dataset, in order to unify the discussion at workshop, we also welcome contributions that are responsive and complementary, but are demonstrated in other ways. By bringing together researchers from diverse perspectives to tackles these and related challenges, the workshop will provide a forum for exchanging ideas and fostering novel collaborations.

Papers should be at most 6 pages in SIAM-DM conference format (data mining file; general website). A PDF file for review should be emailed to cbk@cs.dartmouth.edu by January 8, 2007. The proceedings will be published as an on-line collection of working papers. Following discussion at the workshop, participants will be invited to contribute to a summary report that will be communicated to a premier data mining magazine or journal.

Important Dates

• Submission: extended: January 31, 2007
• Notification: February 19, 2007
• Workshop: April 28, 2007 (half a day at SIAM-DM 2007)

Organization

Program chairs:

• Chris Bailey-Kellogg, Dartmouth
• Naren Ramakrishnan, Virginia Tech
• Madhav Marathe, Virginia Bioinformatics Institute

Program committee:

• Keith Bisset, Virginia Bioinformatics Institute
• Jiawei Han, University of Illinois at Urbana-Champaign
• Ken Kleinman, Harvard Medical School
• Srinivasan Parthasarathy, The Ohio State University