Scholarlink

<p>This study aims to model Disability Adjusted Life Years (DALYs) for cardiovascular diseases and neurological disorders using ARIMA models, exploring regional disparities and evaluating model performance across countries. This is a quantitative, exploratory time series study using secondary data from the Global Burden of Disease (GBD) database, covering annual DALYs across 48 European countries, from 1990 until 2019. ARIMA models were applied using an automated algorithm. Model performance was assessed using MAE, RMSE, MSE, and MAPE. Time series were compared using Piccolo distances, and countries were clustered with hierarchical average linkage. Cluster quality was evaluated using Silhouette, Dunn, McClain, and C-index. Simpler ARIMA models often yielded better forecasts than more complex ones, particularly in smaller countries such as San Marino and Monaco. Cardiovascular disease time series were grouped into 2 clusters, while neurological disorders formed 15 clusters, reflecting diverse epidemiological and reporting patterns. The mean MAPE was 6.8% for cardiovascular diseases and 0.95% for neurological disorders, indicating greater predictive accuracy and stability for the latter. ARIMA modelling is effective for capturing temporal dynamics in DALY data, but should be tailored to avoid overfitting. Clustering countries based on time series similarities reveals insights into health system differences and regional disease patterns. These findings support the use of data-driven approaches to improve forecasting and inform global public health planning.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Forecasting and clustering the burden of non communicable diseases in Europe to support public health policy

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年6月20日

查看全文
  • Cláudia Vinhal,
  • Alexandra Oliveira,
  • Brígida Mónica Faria,
  • Rui Pimenta,
  • Ana Paula Nascimento

摘要

This study aims to model Disability Adjusted Life Years (DALYs) for cardiovascular diseases and neurological disorders using ARIMA models, exploring regional disparities and evaluating model performance across countries. This is a quantitative, exploratory time series study using secondary data from the Global Burden of Disease (GBD) database, covering annual DALYs across 48 European countries, from 1990 until 2019. ARIMA models were applied using an automated algorithm. Model performance was assessed using MAE, RMSE, MSE, and MAPE. Time series were compared using Piccolo distances, and countries were clustered with hierarchical average linkage. Cluster quality was evaluated using Silhouette, Dunn, McClain, and C-index. Simpler ARIMA models often yielded better forecasts than more complex ones, particularly in smaller countries such as San Marino and Monaco. Cardiovascular disease time series were grouped into 2 clusters, while neurological disorders formed 15 clusters, reflecting diverse epidemiological and reporting patterns. The mean MAPE was 6.8% for cardiovascular diseases and 0.95% for neurological disorders, indicating greater predictive accuracy and stability for the latter. ARIMA modelling is effective for capturing temporal dynamics in DALY data, but should be tailored to avoid overfitting. Clustering countries based on time series similarities reveals insights into health system differences and regional disease patterns. These findings support the use of data-driven approaches to improve forecasting and inform global public health planning.