人工智能辅助鼻细胞学全定量检测在慢性鼻炎中的临床效度分析OA

OBJECTIVE To systematically evaluate the clinical validity of AI-assisted fully quantitative nasal cytology(hereinafter referred to as quantitative)in chronic rhinitis,including its consistency with traditional semi-quantitative methods,the strength of association with clinical symptoms and serological indicators,and its explanatory power for nasal symptoms.METHODS Nasal smears were collected to obtain traditional semi-quantitative grades(0-4+)and fully quantitative cytological parameters(Eos%,absolute Eos count,degranulated Eos ratio).Serum total IgE,aeroallergen-specific IgE screening,total nasal symptom score(TNSS,0-12),total ocular symptom score(TOSS,0-6),and individual visual analog scale(VAS)scores were simultaneously recorded.Spearman correlation was used to assess correlations,Mann-Whitney U test to compare clinical phenotypic differences,ordinal logistic regression to compare symptom explanatory power,and restricted cubic splines(RCS)to delineate dose-response relationships.RESULTS The fully quantitative Eos count was highly correlated with the semi-quantitative method(Spearman ρ=0.791,P<0.001),while Eos%showed a moderate correlation(ρ=0.575).Fully quantitative Eos%had the strongest correlation with TNSS(ρ=0.254),whereas blood Eos%correlated most strongly with total IgE and aeroallergen-specific IgE screening(ρ=0.19-0.22).In asthma stratification,only blood Eos%showed a significant inter-group difference(P<0.05).Ordinal logistic regression showed that the fully quantitative Eos%model(AIC=1578.2,Nagelkerke R²=0.039)outperformed the semi-quantitative model(AIC=1587.0,R²=0.033)and the blood Eos%model(AIC=1603.1,R²=0.023).RCS revealed a linear dose-response relationship between Eos%and TNSS.CONCLUSION The fully quantitative Eos count is highly consistent with the semi-quantitative method.AI-assisted fully quantitative testing can replace the semi-quantitative approach by providing continuous values and a wider dynamic range.Fully quantitative nasal cytology outperforms traditional semi-quantitative methods in symptom explanatory power,and a stable linear relationship exists between Eos%and TNSS,which can continuously reflect the severity of nasal symptoms.