Data-Enhancement Strategies in Weather-Related Health Studies

Area of Science:

• Environmental Health
• Biostatistics
• Time Series Analysis

Background:

• The link between weather and health is established but has knowledge gaps.
• Existing studies often overlook nuanced temporal aspects of weather-health relationships.

Purpose of the Study:

• To propose and evaluate novel data transformation strategies for analyzing weather-health dynamics.
• To explore non-conventional aspects of temperature-mortality relationships, including temporal scales and event timing.

Main Methods:

• Four data transformation strategies: temporal aggregation, empirical mode decomposition, daily temperature curve disaggregation, and annual data as a continuous function.
• Application of functional regression for analyzing complex exposure-response relationships.
• Cross-validation for comparing model performance.

Main Results:

• The proposed strategies successfully retrieve non-dominant time scales from data.
• Functional regression models (strategies iii and iv) demonstrated a good fit for temperature-related cardiovascular mortality.
• Case study in Montreal highlighted the utility of these methods.

Conclusions:

• Data transformation offers a powerful approach to address gaps in weather-health research.
• Advanced methods like functional regression enhance the accuracy of climate-related health impact assessments.
• These strategies provide deeper insights into the complex interplay between weather and public health.