Incremental value of a multidomain epidemiological questionnaire for discriminating thyroid cancer case status: a case–control study

Abstract Background Multidomain epidemiological questionnaires capture socioeconomic, lifestyle and environmental information, but the contribution of each domain to thyroid cancer case discrimination remains unclear. We evaluated the incremental value of a seven-domain questionnaire beyond age, sex and a limited set of literature-supported variables, and compared contributions across domains. Methods This single-centre case–control study included 660 participants recruited from October 2023 to September 2024, comprising 330 cases and 330 controls matched by age and sex. We used 10 repeats of five-fold nested cross-validation, with preprocessing, variable selection and tuning confined to the outer training folds. We compared elastic net logistic regression, L2-penalised logistic regression, random forest, XGBoost and support vector machines. The area under the receiver operating characteristic curve (AUROC) was calculated from each participant's mean across 10 outer-test predictions. We estimated 95% confidence intervals using 2000 bootstrap resamples. Domain addition and leave-one-domain-out ablation analyses assessed the contribution of each domain. Results The support vector machine had the numerically highest AUROC, at 0.933 (95% confidence interval, 0.915–0.950). Its full model exceeded the restricted model by 0.289 AUROC units (95% confidence interval, 0.247–0.333). Domain addition and ablation analyses both ranked salt use and diet, digital device use, and demographic and socioeconomic characteristics as the three leading contributors. These domains also ranked in the top three in the elastic net sensitivity analysis. Conclusions The full multidomain questionnaire provided additional discriminatory information beyond the restricted variable set in this case–control sample. These findings may inform questionnaire refinement for future epidemiological surveys, but require validation in independent samples.

Authors

Institutions

Publication Details

Journal
Research Square
Published
2026-11-16
DOI
https://doi.org/10.21203/rs.3.rs-11195324/v1
Primary Topic
Thyroid Cancer Diagnosis and Treatment
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Incremental value of a multidomain epidemiological questionnaire for discriminating thyroid cancer case status: a case–control study

Hailin Sun, Dehao Wang, Zhen Wu, Jinchao Wang et al.
1 links
Research Square
Thyroid Cancer Diagnosis and Treatment
preprint

Incremental value of a multidomain epidemiological questionnaire for discriminating thyroid cancer case status: a case–control study

Hailin Sun, Dehao Wang, Zhen Wu, Jinchao Wang, Zhaoqing Cui, Li Liu
preprint en

Abstract

Abstract Background Multidomain epidemiological questionnaires capture socioeconomic, lifestyle and environmental information, but the contribution of each domain to thyroid cancer case discrimination remains unclear. We evaluated the incremental value of a seven-domain questionnaire beyond age, sex and a limited set of literature-supported variables, and compared contributions across domains. Methods This single-centre case–control study included 660 participants recruited from October 2023 to September 2024, comprising 330 cases and 330 controls matched by age and sex. We used 10 repeats of five-fold nested cross-validation, with preprocessing, variable selection and tuning confined to the outer training folds. We compared elastic net logistic regression, L2-penalised logistic regression, random forest, XGBoost and support vector machines. The area under the receiver operating characteristic curve (AUROC) was calculated from each participant's mean across 10 outer-test predictions. We estimated 95% confidence intervals using 2000 bootstrap resamples. Domain addition and leave-one-domain-out ablation analyses assessed the contribution of each domain. Results The support vector machine had the numerically highest AUROC, at 0.933 (95% confidence interval, 0.915–0.950). Its full model exceeded the restricted model by 0.289 AUROC units (95% confidence interval, 0.247–0.333). Domain addition and ablation analyses both ranked salt use and diet, digital device use, and demographic and socioeconomic characteristics as the three leading contributors. These domains also ranked in the top three in the elastic net sensitivity analysis. Conclusions The full multidomain questionnaire provided additional discriminatory information beyond the restricted variable set in this case–control sample. These findings may inform questionnaire refinement for future epidemiological surveys, but require validation in independent samples.

Research Square
Liaocheng People's Hospital (CN), Shandong First Medical University (CN), Liaocheng Center for Disease Control and Prevention (CN)
Good health and well-being
Thyroid Cancer Diagnosis and Treatment
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon