Evaluations of methods for statistical surveillance

Area of Science:

• Medical Statistics
• Public Health Surveillance
• Time Series Analysis

Background:

• Effective statistical surveillance is crucial across various medical disciplines.
• Evaluating the performance of alarm procedures is essential for timely medical interventions.
• Continual surveillance of medical data requires robust methodologies.

Purpose of the Study:

• To describe the needs for statistical surveillance in medicine.
• To suggest measures, including predictive value of an alarm, for evaluating alarm procedures.
• To assess common methods for continual surveillance of time series data.

Main Methods:

• Evaluation of statistical surveillance methods using suggested measures.
• Analysis of common continual surveillance techniques for time series data.
• Assessment of alarm procedures and their predictive values.

Main Results:

• Demonstration that some surveillance methods maintain consistent properties throughout the observation period.
• Identification of methods where performance varies significantly between the start and later stages of surveillance.
• Highlighting that the implications of early versus late alarms differ for certain methods.

Conclusions:

• The choice of statistical surveillance method impacts its reliability over time.
• Alarm procedures require careful evaluation, considering the timing of alerts.
• Methodological differences in time series surveillance necessitate tailored approaches for medical monitoring.