目的：探讨人工智能宫颈癌筛查系统与阴道镜在宫颈鳞状上皮内病变筛查中的临床效果。方法：我院收治的1 726例行宫颈鳞状上皮内病变筛查的女性,采用人工智能宫颈癌筛查系统与阴道镜联合筛查宫颈鳞状上皮内病变,并与病理检查结果比较,分析单独检测与联合检测的一致性、诊断价值。结果：1 726例筛检者中,活检病理结果异常5.27%（91/1 726）,其中LSIL、HSIL、鳞癌分别占63.7%（58/91）、28.6%（26/91）、7.7%（7/91）。TS检测结果异常4.81%（83/1 726）（95%CI：3.62%~6.68%）。TS检测的特异度为99.8%（1 631/1 635）（95%CI：85.74%~99.99%）,灵敏度为91.2%（83/91）（95%CI：84.21%~93.42%）。阴道镜检结果异常4.17%（72/1 726）（95%CI：87.39%~99.28%）。阴道镜检的特异度为99.6%（1 629/1 635）（95%CI：87.84%~99.99%）,灵敏度为79.1%（72/91）（95%CI：70.27%~86.29%）。联合检测结果异常5.00%（86/1 726）（95%CI：1.27%~9.38%）。联合检测的特异度为99.9%（1 634/1 635）（95%CI： 90.23%~99.99%）,灵敏度为94.5%（86/91）（95%CI：88.45%~98.75%）。Logistic多因素回归分析显示宫颈癌家族史（OR=1.805,95%CI：1.217~2.675）、生殖道HPV感染（OR=1.842,95%CI：1.263~2.681）、性伴侣生殖系统恶性肿瘤（OR=1.364,95%CI：1.096~1.874）是影响宫颈癌发生的危险因素。结论：宫颈癌家族史、生殖道HPV感染、性伴侣生殖系统恶性肿瘤是影响宫颈癌发生的危险因素,为临床预防和治疗提供了依据。采用人工智能宫颈癌筛查系统与阴道镜联合检测宫颈鳞状上皮内病变,监测的灵敏度、特异度高,与病理检测一致性好,诊断价值高,适于在临床推广应用。

Objective To explore the clinical effects of artificial intelligence cervical cancer screening system and colposcopy in cervical lesion screening. Methods 1 726 women admitted to our hospital for cervical lesion screening were screened for cervical lesions using artificial intelligence cervical cancer screening system in combination with colposcopy, and the results were compared with those of pathological examination to analyze the consistency and diagnostic value of separate detection and combined detection. Results Among 1 726 screened patients, 5.27% (91/1 726) had abnormal biopsy pathology results, of which LSIL, HSIL, and squamous carcinoma accounted for 63.7% (58/91), 28.6% (26/91), and 7.7% (7/91), respectively. TS test results were abnormal in 4.81% (83/1 726) (95% CI: 3.62%-6.68%). The specificity of the TS test was 99.8% (1 631/1 635) (95% CI: 85.74%-99.99%) and the sensitivity was 91.2% (83/91) (95% CI: 84.21%-93.42%). Colposcopy findings were abnormal in 4.17% (72/1 726) (95% CI: 87.39%-99.28%). The specificity of colposcopy was 99.6% (1 629/1 635) (95% CI: 87.84%-99.99%) and the sensitivity was 79.1% (72/91) (95% CI: 70.27%-86.29%). The combined assay was abnormal in 5.00% (86/1 726) (95% CI: 1.27%-9.38%). The specificity of the combined test was 99.9% (1 634/1 635) (95% CI: 90.23%-99.99%) and the sensitivity was 94.5% (86/91) (95% CI: 88.45%-98.75%). Logistic multivariate regression analysis revealed that family history of cervical cancer (OR=1.805, 95%CI: 1.217-2.675), genital HPV infection (OR=1.842, 95%CI: 1.263-2.681), and reproductive system malignancy in sexual partners (OR=1.364, 95%CI: 1.096-1.874) are risk factors for the occurrence of cervical cancer. Conclusion Family history of cervical cancer, genital HPV infection, and reproductive system malignancy in sexual partners are risk factors for the occurrence of cervical cancer, providing a basis for clinical prevention and treatment. The adoption of an artificial intelligence cervical cancer screening system combined with colposcopy for detecting cervical lesions results in high sensitivity and specificity, good consistency with pathological testing, and high diagnostic value, making it suitable for clinical application and promotion.