ML支援による嗅覚疫学調査：都市化する中国、塩城市の人口ベース研究

Urbanization-related air pollution may be associated with olfactory dysfunction (OD) in China, yet population studies are lacking.

We conducted a cross-sectional study of 1500 participants in urbanizing Yancheng, China (2023-2025). Olfactory function was assessed via Sniffin' Sticks. Machine learning (XGBoost, k-means clustering) was used to analyze risk factors and phenotypes.

Head trauma (OR = 163.04) and chronic rhinosinusitis (CRS, OR = 161.39) were the dominant risk factors for OD (p < 0.001). Rural residence was associated with a 99% lower OD risk compared to urban residence (OR = 0.011, p < 0.0001). Four OD subtypes emerged: (1) sinusitis-linked generalized impairment; (2) age-related threshold deficit; (3) neurological disorder-associated identification loss; (4) child-specific mild dysfunction. Olfactory ability peaked at the age of 23 years, remained stable among individuals aged 40-60 years, and declined sharply in those aged over 63 years. Anosmia quadrupled from childhood to elderly (7.1% → 28.2%, p < 0.0001).

In urbanizing China, head trauma and sinusitis are the most potent risk factors for olfactory dysfunction (OD). Additionally, long-term exposure to air pollution-proxied by urban residence-and the natural aging process are significant contributing factors. Phenotypic stratification enables targeted interventions, thereby urging reforms in environmental policies and enhanced clinical management of high-risk conditions (e.g., head trauma, sinusitis).