Development and Validation of a Deep Learning Model for Earlier Detection of Cognitive Decline From Clinical Notes in Electronic Health Records.
Development and Validation of a Deep Learning Model for Earlier Detection of Cognitive Decline From Clinical Notes in Electronic Health Records.
复制标题
深度学习模型的开发和验证，用于早期检测电子健康记录中的临床笔记中的认知下降。
DOI：
10.1001/jamanetworkopen.2021.35174
复制
发表时间：
2021-11-01
影响因子：
13.8
通讯作者：
Zhou L
中科院分区：
文献类型：
--
作者：
Wang L;Laurentiev J;Yang J;Lo YC;Amariglio RE;Blacker D;Sperling RA;Marshall GA;Zhou L
应用于临床笔记的深度学习算法能否在轻度认知障碍（MCI）诊断之前检测到认知下降的证据？在这项诊断研究中，使用2166名MCI诊断前患者的临床记录，对深度学习算法进行了训练和验证，以使用具有和不具有关键字过滤的数据集检测认知下降。在具有关键字过滤的数据集中训练的模型在没有关键字过滤的数据集中表现令人满意。这项研究的结果表明，深度学习模型可以从MCI诊断前的笔记中检测到认知下降的证据，这可能有助于早期检测电子健康记录中的认知下降。早期检测老年人的认知能力下降可以促进临床试验和早期干预的登记。纵向电子健康记录（EHR）中的临床记录提供了比结构化EHR字段中作为正式诊断更早检测认知衰退的机会。开发和验证深度学习模型，以从EHR中的临床记录中检测认知下降的证据。在最初的轻度认知功能障碍（MCI）诊断前4年记录的笔记是从Mass General Brigham的企业数据仓库中提取的，适用于年龄在50岁或以上的患者，并在2019年期间进行了初步MCI诊断。研究于2020年3月1日至2021年6月30日进行。手动标记认知下降的笔记部分，并创建2个参考数据集。数据集I包含4950个音符部分的随机样本，通过与认知功能相关的关键字列表进行过滤，并用于模型训练和测试。数据集II包含2000个随机选择的部分，没有关键词过滤，用于评估模型性能是否依赖于特定的关键词。开发了一个深度学习模型和4个基线模型，并使用受试者操作特征曲线下面积（AUROC）和精确召回曲线下面积（AUPRC）比较了它们的性能。数据集I代表1969例患者（1046例[53.1%]女性;平均[SD]年龄，76.0 [13.3]岁）。数据集II包括1161例患者（619例[53.3%]女性;平均[SD]年龄，76.5 [10.2]岁）。删除一些重叠患者后，唯一人群为2166人。在数据集I的1453个切片（29.4%）和数据集II的69个切片（3.45%）中观察到认知下降。与4个基线模型相比，深度学习模型在两个数据集中都取得了最佳性能，AUROC为0.971（95% CI，0.967-0.976）和AUPRC为0.933（95% CI，0.921-0.944），数据集II的AUROC为0.997（95% CI，0.994-0.999），AUPRC为0.929（95% CI，0.870-0.969）。在这项诊断研究中，深度学习模型在MCI诊断之前准确地检测到了临床记录中的认知下降，并且比基于关键字的搜索和其他机器学习模型具有更好的性能。这些结果表明，深度学习模型可以用于早期检测EHR中的认知下降。这项诊断研究评估了深度学习模型在使用电子健康记录中的临床记录诊断轻度认知障碍之前检测认知下降证据的能力。
Can a deep learning algorithm applied to clinical notes detect evidence of cognitive decline before a mild cognitive impairment (MCI) diagnosis? In this diagnostic study, using clinical notes on 2166 patients preceding an MCI diagnosis, a deep learning algorithm was trained and validated for detecting cognitive decline using data sets with and without keyword filtering. The model trained in the data set with keyword filtering performed satisfactorily in the data sets without keyword filtering. The results of this study suggest that a deep learning model can detect evidence of cognitive decline from notes preceding an MCI diagnosis, potentially facilitating earlier detection of cognitive decline in electronic health records. Detecting cognitive decline earlier among older adults can facilitate enrollment in clinical trials and early interventions. Clinical notes in longitudinal electronic health records (EHRs) provide opportunities to detect cognitive decline earlier than it is noted in structured EHR fields as formal diagnoses. To develop and validate a deep learning model to detect evidence of cognitive decline from clinical notes in the EHR. Notes documented 4 years preceding the initial mild cognitive impairment (MCI) diagnosis were extracted from Mass General Brigham’s Enterprise Data Warehouse for patients aged 50 years or older and with initial MCI diagnosis during 2019. The study was conducted from March 1, 2020, to June 30, 2021. Sections of notes for cognitive decline were labeled manually and 2 reference data sets were created. Data set I contained a random sample of 4950 note sections filtered by a list of keywords related to cognitive functions and was used for model training and testing. Data set II contained 2000 randomly selected sections without keyword filtering for assessing whether the model performance was dependent on specific keywords. A deep learning model and 4 baseline models were developed and their performance was compared using the area under the receiver operating characteristic curve (AUROC) and area under the precision recall curve (AUPRC). Data set I represented 1969 patients (1046 [53.1%] women; mean [SD] age, 76.0 [13.3] years). Data set II comprised 1161 patients (619 [53.3%] women; mean [SD] age, 76.5 [10.2] years). With some overlap of patients deleted, the unique population was 2166. Cognitive decline was noted in 1453 sections (29.4%) in data set I and 69 sections (3.45%) in data set II. Compared with the 4 baseline models, the deep learning model achieved the best performance in both data sets, with AUROC of 0.971 (95% CI, 0.967-0.976) and AUPRC of 0.933 (95% CI, 0.921-0.944) for data set I and AUROC of 0.997 (95% CI, 0.994-0.999) and AUPRC of 0.929 (95% CI, 0.870-0.969) for data set II. In this diagnostic study, a deep learning model accurately detected cognitive decline from clinical notes preceding MCI diagnosis and had better performance than keyword-based search and other machine learning models. These results suggest that a deep learning model could be used for earlier detection of cognitive decline in the EHRs. This diagnostic study assesses the ability of a deep learning model to detect evidence of cognitive decline before a diagnosis of mild cognitive impairment using clinical notes from electronic health records.
登录
查看更多内容
影响因子：
2.1
作者：
Bradford A;Kunik ME;Schulz P;Williams SP;Singh H
通讯作者：
Singh H
影响因子：
6.3
作者：
Moura LMVR;Festa N;Price M;Volya M;Benson NM;Zafar S;Weiss M;Blacker D;Normand SL;Newhouse JP;Hsu J
通讯作者：
Hsu J
DOI：
10.3390/s20247292
发表时间：
2020-12-18
期刊：
Sensors (Basel, Switzerland)
影响因子：
--
作者：
Fernández Montenegro JM;Villarini B;Angelopoulou A;Kapetanios E;Garcia-Rodriguez J;Argyriou V
通讯作者：
Argyriou V
影响因子：
--
作者：
Petersen, Ronald C.;Roberts, Rosebud O.;Knopman, David S.;Boeve, Bradley F.;Geda, Yonas E.;Ivnik, Robcrt J.;Smith, Glenn E.;Jack, Clifford R., Jr.
通讯作者：
Jack, Clifford R., Jr.
影响因子：
13.8
作者：
Wang, Liqin;Sha, Long;Zhou, Li
通讯作者：
Zhou, Li