Natural language processing improves identification of colorectal cancer testing in the electronic medical record.
Natural language processing improves identification of colorectal cancer testing in the electronic medical record.
复制标题
DOI：
10.1177/0272989x11400418
复制
发表时间：
2012-01
影响因子：
3.6
通讯作者：
Peterson, Neeraja B.
中科院分区：
文献类型：
--
作者：
Denny, Joshua C.;Choma, Neesha N.;Peterson, Josh F.;Miller, Randolph A.;Bastarache, Lisa;Li, Ming;Peterson, Neeraja B.
关键词：
及时识别需要结直肠癌（CRC）筛查的患者的困难导致全国观察到的总体筛查率较低。使用电子健康记录（EHR）中的数据来识别既往接受过CRC检测的患者。我们修改了本地开发的临床自然语言处理（NLP）系统，以确定电子临床文件中的四个CRC测试（结肠镜检查，乙状结肠镜检查，粪便潜血测试和双对比钡灌肠）。系统对临床记录和手术报告中包含CRC检测参考的文本短语进行解释，以确定是否计划或完成检测，并估计完成检测的日期。大型学术医疗中心。200例≥ 50岁的患者，在1年内完成了至少2次非急性初级保健门诊访视。我们使用所有可用信息源的人工审查参考标准，比较了NLP系统、账单记录和人工审查电子记录的召回率（敏感性）和精确度（阳性预测值）。对于所有CRC测试的识别，召回率和精确度如下：NLP系统（召回率93%，精确度94%），手动图表审查（74%，98%）和账单记录审查（44%，83%）。识别需要筛查的患者的召回率和准确率为：NLP系统（召回率95%，准确率88%），手动图表审查（99%，82%）和账单记录审查（99%，67%）。这项研究是在一个医疗中心对有限的患者进行的，需要一个强大的EHR来实现。将NLP应用于EHR记录检测到的CRC测试比手动图表审查或单独的计费记录审查更多。与账单记录审查相比，NLP在识别应进行CRC筛查的患者方面具有更好的精确度，但召回率略低。
Difficulties in the timely identification of patients in need of colorectal cancer (CRC) screening contribute to the low overall screening rates observed nationally. To use data within Electronic Health Record (EHR) to identify patients with prior CRC testing. We modified a locally-developed clinical natural language processing (NLP) system to identify four CRC tests (colonoscopy, flexible sigmoidoscopy, fecal occult blood testing, and double contrast barium enema) within electronic clinical documentation. Text phrases in clinical notes and procedure reports which included references to CRC tests were interpreted by the system to determine whether testing was planned or completed, and to estimate the date of completed tests. Large academic medical center. 200 patients ≥ 50 years old who had completed at least two non-acute primary care outpatient visits within a one-year period. We compared the recall (sensitivity) and precision (positive predictive value) of the NLP system, billing records, and manual review of electronic records, using a reference standard of human review of all available information sources. For identification of all CRC tests, recall and precision were as follows: NLP system (recall 93%, precision 94%), manual chart review (74%, 98%), and billing records review (44%, 83%). Recall and precision for identification of patients in need of screening were: NLP system (recall 95%, precision 88%), manual chart review (99%, 82%), and billing records review (99%, 67%). This study was performed in one medical center on a limited set of patients, and requires a robust EHR for implementation. Applying NLP to EHR records detected more CRC tests than either manual chart review or billing records review alone. NLP had better precision but marginally lower recall to identify patients who were due for CRC screening than billing record review.