第一部分：基于电子住院病历信息化建立的子宫内膜异位症单中心数据库: 目的: 子宫内膜异位症（内异症）是妇科最为常见的疾病之一。而电子病历包含大量临床信息，本研究拟通过信息化手段处理北京协和医院电子住院病历，建立一个子宫内膜异位症单中心数据库。方法: 检索 2015 年 4 月到 2019 年 8 月于北京协和医院接受手术治疗、出院诊断或病理报告符合子宫内膜异位症的住院患者，并获得相应电子病历。 研究者（临床医生）设计拟获取的数据项及数据项的提取位置、提取逻辑。睿医（北京）数据技术有限公司（信息工程师）将临床医学逻辑转化成信息化程序语言，进行电子病历数据挖掘，实现患者信息抽取。结果: 共纳入 5770 例内异症女性，年龄多集中在 21 岁到 50 岁（占 94.25%）。2.67%的患者已自然绝经。 61.56%报告有疼痛症状， 57.90%报告有盆腔包块表现， 21.56%报告合并不孕。1.82%合并生殖道畸形，34.07%合并子宫肌瘤，20.47%合并子宫腺肌症，13.21%合并子宫内膜病变（11.11%合并子宫内膜息肉，1.40%合并子宫内膜癌， 1.37%合并子宫内膜增生） 。 81.99%经病理确诊，内异症见于卵巢、骶子宫韧带、腹壁、直肠阴道隔和子宫直肠窝、 输卵管、 子宫（包括子宫剖宫产瘢痕） 、 输尿管、直肠、 膀胱腹膜反折、 阔韧带、 会阴、 宫颈、 膀胱、 阴道、 结肠、 大网膜、 圆韧带、阑尾、腹股沟、固有韧带、小肠、回肠、盲肠（按发生率由大到小排列） 。结论: 本研究使用数据技术对电子病历进行临床信息抽取，成功建立了一个单中心内异症数据库，可用于深入总结与进一步探索。第二部分：基于所建数据库的 45 岁以上内异症相关卵巢癌的危险因素分析: 目的: 内异症虽然是良性疾病，但可以发生恶变，80%的内异症恶变发生在卵巢，为内异症相关卵巢癌（Endometriosis Associated Ovarian Carcinoma, EAOC）。本研究基于所建的北京协和医院 2015.4-2019.8 住院病人子宫内膜异位症数据库，对 45 岁以上经手术病理验证的卵巢型内异症患者和合并盆腔内异症的卵巢癌患者进行回顾性分析，以寻找 EAOC 危险因素。方法: 共纳入 64 例 EAOC、438 例卵巢型内异症。使用 IBM SPSS Statistics 22.0 进行统计描述及 Binary Logistic 回归分析。使用受试者工作特征曲线（ROC 曲线）和约登指数判断有显著差异的数值变量的最佳临界值。本研究取 P<0.05 为有统计学意义。结果: 64 例 EAOC 患者中， 同侧卵巢癌和卵巢型内异症（符合或不严格符合 Sampson和 Scott 诊断标准） 占 43.75%，卵巢透明细胞癌和卵巢子宫内膜样癌占 78.13%， 早期（I 期或 II 期）卵巢癌占 76.56%。单因素分析显示，EAOC 组与卵巢型内异症组在年龄、 自然绝经状态、 术前血清 CA125 水平、 术前血清 CA199 水平、 肿瘤大小、痛经等方面差异显著。 多因素 Logistic 回归分析提示自然绝经状态、 术前血清 CA125 水平、肿瘤大小是 EAOC 独立危险因素。肿瘤大小最佳临界值为 8 cm。结论: 45 岁以上卵巢型内异症患者，卵巢肿瘤超过 8 cm、 已自然绝经、 术前血清 CA125升高，需警惕恶变。

Part I: A Single-center Database of Endometriosis Established by Information Processing of Inpatient Electronic Medical Records: Objective: Endometriosis (EM) is one of the most common diseases in gynecology. The electronic medical record contains a large amount of clinical information. This study intends to process the inpatient electronic medical records of Peking Union Medical College Hospital (PUMCH) through information technology and establish a single-center database of endometriosis. Methods: The data of inpatients who received surgical treatment and were diagnosed or had pathologically confirmed endometriosis at PUMCH from April 2015 to August 2019 were retrieved and the corresponding electronic medical records were obtained. The researcher (clinician) determined the data items to be extracted by pinpointing the localization and extraction logic. Ruiyi (Beijing) Data Technology Co., Ltd. (information engineer) converted clinical medical logic into an information-based programming language, conducted electronic medical record data mining, and realized patient information extraction. Results: A total of 5770 women with endometriosis were included, mostly aged from 21 to 50 years (accounting for 94.25%). 2.67% had postmenopausal endometriosis. 61.56% reported pain, 57.90% reported pelvic masses, and 21.56% reported infertility. 1.82% had genital tract malformations, 34.07% had uterine fibroids, 20.47% had adenomyosis, 13.21% had endometrial lesions (11.11% with endometrial polyps, 1.40% with endometrial cancer, 1.37% with uterus Intimal hyperplasia). 81.99% were diagnosed with pathology, endometriosis was found in the ovaries, sacral uterine ligament, abdominal wall, rectal vaginal septum and uterine rectum fossa, fallopian tube, uterus (including cesarean scar), ureters, rectum, bladder peritoneal reflex, broad ligaments, perineum, cervix, bladder, vagina, colon, omentum, round ligament, appendix, groins, intrinsic ligament, small intestine, ileum, and cecum (ranked from highest to lowest incidence). Conclusion: This study used data technology to extract clinical information from electronic medical records, and successfully established a single-center inpatient disease database, which can be used for in-depth summary and further exploration. Part II: Risk Factors of Endometriosis Associated Ovarian Carcinoma in Women over 45 Years Old: Objective: Endometriosis, an overall benign disease, can undergo malignant transformation. About 80% of the cases are found in the ovary, which is termed endometriosis associated ovarian carcinoma, or EAOC. By utilizing the PUMCH database of inpatients between April 2015 and August 2019, the study retrospectively analyzed characteristics of those over 45 years old with pathologically confirmed ovarian endometriosis or ovarian carcinoma with pelvic endometriosis to shed light on potential risk factors of EAOC. Methods: 64 patients with EAOC and 438 patients with ovarian endometriosis were included. IBM SPSS Statistics 22.0 was used for statistical description and binary logistic regression analysis. ROC curve and Youden index were used to determine the best cutoff points for the continuous variables that were statistically significant (p<0.05).Results: Among the 64 patients with EAOC, 43.75% had ovarian carcinoma and ovarian endometriosis in the ipsilateral ovary; 78.13% had ovarian clear cell carcinoma or ovarian endometrioid carcinoma; 76.56% had early-stage ovarian cancer (stage I or II). Univariate analysis documented statistically significant differences in their ages, menopausal states, pre-surgery serum levels of CA125 and CA199, tumor sizes and occurrence of dysmenorrhoea. Multivariate Logistic regression showed that menopausal states, presurgery levels of CA125 and tumor sizes were independent risk factors for EAOC. The best cutoff value was 8cm for tumor sizes. Conclusion: For patients over 45 years old with ovarian endometriosis, the presence of an ovarian tumor larger than 8cm, an elevated pre-surgery serum CA125 level and the menopausal state would signal an increased risk of malignant transformation.