运用电子病历识别系统监测新冠肺炎相关症状的探索

目的 分析基于自然语言的电子病历识别系统EMRRS识别门急诊病历中与新冠肺炎相关的主要症状指标的准确性,探索其在疾病监测中的应用.方法 随机抽取2020年4月1日-4月10日期间在某三甲综合医院门急诊就诊的4 802例患者电子病历,以临床医生判定的指标为金标准,对比EMRRS识别指标和患者自填指标,分别计算EMRRS和...

Exploration of COVID-19 Related Symptoms Surveillance by Electronic Medical Record Recognition System

Objective Electronic medical record recognition system(EMRRS)based on natural language was adopted to extract COVID-19 related symptoms from electronic medical record of the outpatients and the emergency patients.The accuracy of Electronic medical record recognition system(EMRRS)was analyzed and compared to explore the application of EMRRS in disease monitoring.Methods The 4802 electronic medical records of the outpatients and the emergency patients were randomly selected from a tertiary comprehensive hospital during April 1st-April 10th,2020.Using the symptoms judged by the clinicians as the gold standard,the EMRRS identification symptoms and the patients'self-reported;symptoms were compared.The positive predictive value,sensitivity,specificity and Youden index of EMRRS and the self-reported symptoms of the patients were calculated and consistency test was conducted with the golden standard.Results All of the evaluation indexes of EMRRS were better than the results of the patients'self-reported,and EMRRS were highly consistent with the gold standard.For the two key indicators,fever and cough,the sensitivities were over 93%,the positive prediction rates were more than 95%,and the Kappa values were more than 0.94.Conclusion EMRRS has high consistency with target gold target and can be used for the identification and monitoring of COVID-19 pneumonia related symptoms.