Benchmark Data for Evaluating Space-Time Disease Outbreak Detection Methods

Martin Kulldorff of Harvard University created simulated benchmark datasets for evaluating statistical power in disease outbreak detection. The data is based on the geography and population of New York City and includes scenarios with and without simulated outbreaks of varying size and location. These datasets were used to compare the performance of prospective purely temporal and space-time scan statistics.

Use Cases

• Evaluating the statistical power of space-time early detection methods based on simulated outbreak scenarios.
• Comparing the performance of purely temporal versus space-time scan statistics for outbreak detection.
• Testing anomaly detection algorithms for geographically localized disease outbreaks as described in the study.
• Benchmarking new surveillance methods against established scan statistics using the provided simulated data.

Strengths

• Data is specifically designed for formal statistical power evaluations and comparisons, as stated in the study conclusions.
• Includes simulated outbreaks of varying size and location, as well as data with no outbreak effects, providing controlled test scenarios.
• Based on the real-world geography and population of New York City, lending realism to the simulations.

Limitations

• Data is simulated, not real-world observational data, which may limit generalizability to actual outbreak patterns.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count, file formats, and sample data are unknown, which may limit suitability assessment.

Provenance

Source

Martin Kulldorff, Harvard University

Collection Method

Simulated data generation based on New York City geography and population.

Geography

New York City