Improving Mortality Forecasts with Empirical Prediction Intervals
Author Information
Author(s): Duerst Ricarda, Schöley Jonas
Primary Institution: Max Planck Institute for Demographic Research
Hypothesis
Can empirical prediction intervals provide a more accurate assessment of excess deaths compared to conventional parametric intervals?
Conclusion
Empirical prediction intervals offer a better method for estimating forecast uncertainty in excess death analyses than traditional parametric intervals.
Supporting Evidence
- The empirical prediction intervals were better calibrated than conventional parametric intervals.
- Seasonal variations significantly affect the likelihood of exceeding a 10% threshold of excess deaths.
- Empirical prediction intervals account for the seasonality of mortality forecasting errors.
Takeaway
This study shows that using past data to predict future deaths can help us understand how many more people might die than expected, especially during different seasons.
Methodology
The study used weekly death data from the Short-term Mortality Database for 23 countries to derive empirical prediction intervals based on past forecasting errors.
Potential Biases
Potential biases may arise from the selection of countries and the reliance on historical data for forecasting.
Limitations
The study's findings may not be generalizable to countries outside the 23 studied, and the methodology relies on the assumption that past errors will reflect future errors.
Participant Demographics
Data was sourced from 23 countries, but specific demographic details of participants were not provided.
Statistical Information
P-Value
p<0.05
Confidence Interval
95%
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website