<b>Prediction of risk of hearing loss by industry noise from cross-sectional and longitudinal data</b>

Noise-induced hearing loss (NIHL) at speech frequencies is a critical occupational health issue, impairing communication, reducing quality of life, and contributing to social isolation, depression, and cognitive decline. We developed and validated multimodal prediction models for speech-frequency NIHL among shipyard workers, integrating epidemiological, audiological, and genetic data. The training set included 5,053 workers, and the test set 2,086 workers recruited between 2012 and 2024. Cumulative noise exposure (CNE) was quantified from job history and workplace measurements. Participants completed structured questionnaires, audiological evaluations, and provided blood samples for genotyping. Sex-specific and standard-specific models were constructed, addressing Chinese (0.5, 1, 2 kHz) and WHO (0.5, 1, 2, 4 kHz) definitions of speech-frequency hearing loss.言语频率噪声性听力损失（Noise-induced hearing loss, NIHL）是一类重要的职业健康问题，可损害沟通能力、降低生活质量，并引发社交隔离、抑郁及认知功能衰退。本研究针对造船厂作业工人群体构建并验证了言语频率NIHL的多模态预测模型，整合了流行病学、听力学与遗传学数据。本研究的训练集包含5053名工人，测试集包含2086名于2012年至2024年间招募的研究受试者。累积噪声暴露量（Cumulative noise exposure, CNE）通过受试者职业史及工作场所噪声测量数据进行量化。所有受试者均完成了结构化问卷调查、听力学评估，并提供血液样本用于基因分型检测。本研究构建了性别特异性及标准特异性模型，分别适配中国标准（0.5、1、2 kHz）与世界卫生组织（World Health Organization, WHO）定义的言语频率听力损失判定标准。