Xue Rong, Zhang Hao, Pu Yu, Kong Xinru
Department of Otorhinolaryngology Head and Neck Surgery, Air Force Medical Center, Air Force Medical University, Beijing, China.
Department of Vertigo Center, Air Force Medical Center, PLA, Air Force Medical University, Beijing, China.
Noise Health. 2025;27(125):112-122. doi: 10.4103/nah.nah_188_24. Epub 2025 Apr 29.
The extended high-frequency (EHF; 0.9-16 kHz) region is sensitive to noise exposure and can indicate early noise-induced hearing loss. EHF hearing loss (EHFHL; >20 dB HL for EHF averages) may affect pilots' noise perception, impacting communication and response in flight. Early identification and monitoring of EHFHL are crucial for pilots' hearing health and flight safety. However, EHF is not included in routine medical assessments for pilots in China. This study aimed to develop a nomogram to predict EHFHL in pilots with normal audiograms (≤20 dB HL at each standard frequency), providing an early intervention tool.
A total of 1091 pilots were randomly assigned to the training set (763) and validation set (328). Set characteristics were compared using univariate analysis. In the training set, least absolute shrinkage and selection operator regression identified key predictors, followed by multivariable binary logistic regression to construct a nomogram. The nomogram's performance was evaluated in both sets, assessing calibration, discrimination and clinical utility.
The nomogram incorporated four factors as follows: left-ear high-frequency audiometry threshold averages (HFAs: 3, 4, 6 and 8 kHz; odds ratio [OR] = 1.144; 95% confidence interval [CI] = 1.083-1.210), right-ear HFAs (OR = 1.186, 95% CI = 1.115-1.263), flight time (OR = 1.001, 95% CI = 1-1.001) and triglyceride (OR = 1.393, 95% CI = 1.038-1.885). The model's area under the curve was 0.819 (95% CI = 0.790-0.850) and 0.771 (95% CI = 0.712-0.830) during validation. The predictive model was well calibrated (Hosmer-Lemeshow test, χ2 = 10.77; P = 0.292). Decision curve analysis showed a net benefit for the training set between 4% and 88%, with similar benefits observed for the validation set from 12% to 100%.
This study developed and validated the first prediction model for EHFHL in Chinese pilots, demonstrating its reliability and clinical utility. The findings support early detection and personalised monitoring, with potential applications in hearing protection strategies and flight safety.
超高频（EHF；0.9 - 16千赫兹）区域对噪声暴露敏感，可提示早期噪声性听力损失。EHF听力损失（EHFHL；EHF平均听阈>20分贝听力级）可能影响飞行员对噪声的感知，进而影响飞行中的通信和反应。早期识别和监测EHFHL对飞行员的听力健康和飞行安全至关重要。然而，中国飞行员的常规医学评估中未包括EHF。本研究旨在开发一种列线图，用于预测听力图正常（各标准频率听阈≤20分贝听力级）的飞行员的EHFHL，提供一种早期干预工具。
总共1091名飞行员被随机分配到训练集（763名）和验证集（328名）。使用单因素分析比较两组特征。在训练集中，最小绝对收缩和选择算子回归确定关键预测因素，随后进行多变量二元逻辑回归以构建列线图。在两组中评估列线图的性能，评估校准、区分度和临床实用性。
列线图纳入了以下四个因素：左耳高频听力测定阈值平均值（HFAs：3、4、6和8千赫兹；比值比[OR]=1.144；95%置信区间[CI]=1.083 - 1.210）、右耳HFAs（OR = 1.186，95% CI = 1.115 - 1.263）、飞行时间（OR = 1.001，95% CI = 1 - 1.001）和甘油三酯（OR = 1.393，95% CI = 1.038 - 1.885）。验证期间，模型的曲线下面积分别为0.819（95% CI = 0.790 - 0.850）和0.771（95% CI = 0.712 - 0.830）。预测模型校准良好（Hosmer - Lemeshow检验，χ2 = 10.77；P = 0.292）。决策曲线分析显示训练集的净效益在4%至88%之间，验证集的净效益在12%至100%之间，情况类似。
本研究开发并验证了首个针对中国飞行员EHFHL的预测模型，证明了其可靠性和临床实用性。研究结果支持早期检测和个性化监测，在听力保护策略和飞行安全方面具有潜在应用价值。
