Estimation of the importance of survey items in verbal autopsy for cause of death

2023 Fiscal Year Final Research Report

• Project/Area Number: 21K13274
• Research Category: Grant-in-Aid for Early-Career Scientists
• Allocation Type: Multi-year Fund
• Review Section: Basic Section 07030:Economic statistics-related
• Research Institution: Kwansei Gakuin University
• Principal Investigator: Kunihama Tsuyoshi 関西学院大学, 経済学部, 准教授 (40779716)
• Project Period (FY): 2021-04-01 – 2024-03-31
• Keywords: ベイズ統計学

Outline of Final Research Achievements

This research project developed a statistical method to quantify the importance of survey items in the analysis of distributions of cause-of-death and created a computational algorithm to estimate the importance of each item from the data. Additionally, we applied the proposed method to empirical analysis using actual cause-of-death verbal autopsy survey data and estimated the importance of survey items such as symptoms and medical history. The analysis revealed that, in addition to factors like prematurity and low birth weight, demographic information such as age and sex has a relatively strong association with the cause of death.

Free Research Field

ベイズ統計学

Academic Significance and Societal Importance of the Research Achievements

先行研究において個々の質問項目と死因の関連性の強さについて議論が行われてきたが、本研究課題では、他の質問との情報の重複を除外した上で、個々の質問項目が死因に関して持つ固有の情報量を計算した点に独自性がある。提案手法を活用することで、今後行われる死因の聞き取り調査において、質問表から重要度の低い質問項目を減らす、逆に重要度の高い症状に関しては新たな項目を加えることにより、死因分布の分析精度が向上する可能性がある。