Local and global mortality experience: A novel hierarchical model for regional mortality risk

Asmik Nalmpatian,Christian Heumann,Levent Alkaya,William Jackson

DOI: https://doi.org/10.1101/2024.10.17.24315673

2024-10-18

Abstract:Accurate mortality risk assessment is critical for decision-making in life insurance, healthcare, and public policy. Regional variability in mortality, driven by diverse local factors and inconsistent data availability, presents significant modeling challenges. This study introduces a novel hierarchical mortality risk model that integrates global and local data, enhancing regional mortality estimation across diverse regions. The proposed approach employs a two-stage process: first, a global Light Gradient Boosting Machine model is trained on globally shared features; second, region-specific models are developed to incorporate local characteristics. This framework outperforms both purely local models and standard imputation techniques, particularly in data-scarce regions, by leveraging global patterns to improve generalization. The model is computationally efficient, scalable, and robust in handling missing values, making it adaptable for other domains requiring integration of multi-regional data. This method enhances predictive accuracy across various regions and provides a more reliable approach for mortality risk estimation in data-scarce environments.

Health Informatics

What problem does this paper attempt to address?

### Problems the Paper Attempts to Solve

The paper aims to address the critical issue of accurately assessing mortality risk in fields such as life insurance, healthcare, and public policy. Specifically, the estimation of regional mortality rates faces significant challenges due to the notable differences in data availability and factors influencing mortality across different regions. These issues include:

1. **Data Scarcity**: Insufficient mortality data in certain regions makes it difficult for models to predict accurately.
2. **Data Inconsistency**: The varying formats and quality of data across different regions increase the complexity of modeling.
3. **Balance Between Global and Local Features**: Existing models often struggle to balance global trends and local specificities, leading to models that are either overly generalized or unable to capture the nuances of specific regions.

To address these challenges, the authors propose a new hierarchical mortality risk model that improves the accuracy of regional mortality rate estimates by integrating both global and local data, especially in data-scarce regions. The model employs a two-stage process: first, a global Light Gradient Boosting Machine (LightGBM) model is trained based on globally shared features; second, region-specific models are developed to incorporate local features. This framework not only outperforms purely local models and standard imputation techniques in terms of prediction accuracy but also excels in handling missing values, offering high computational efficiency and strong scalability.