INFERNO: a system for early outbreak detection and signature forecasting

Area of Science:

• Public Health
• Epidemiology
• Infectious Disease Surveillance

Background:

• Public health surveillance systems are crucial for monitoring disease incidence and enabling rapid detection of enteric outbreaks.
• Existing systems provide valuable information but can be enhanced with advanced forecasting and detection algorithms.

Purpose of the Study:

• To describe the INtegrated Forecasts and EaRly eNteric Outbreak (INFERNO) detection system.
• To detail its algorithms for enhanced outbreak detection and forecasting of infectious diseases.

Main Methods:

• INFERNO integrates infectious disease epidemiology knowledge into adaptive forecasts.
• It utilizes an outbreak signature concept, defined as a composite of disease epidemic curves.
• The system employs loess-type smoothing, derivative estimation for near-term forecasting, and a gamma-based signature curve for outbreak size prediction.

Main Results:

• The system comprises four components: training, warning/flagging, signature forecasting, and evaluation.
• A five-level, color-coded warning index quantifies the level of concern, balancing Type I and Type II errors.
• The system demonstrated potential through its application to a 1993 cryptosporidiosis outbreak in Milwaukee.

Conclusions:

• The INFERNO system shows promise for improving public health surveillance and outbreak detection.
• Further development is planned, including adjustments for seasonality and reporting delays.
• The system's ability to forecast outbreak size and provide early warnings is a significant advancement.