Dataset Information

Modeling disease severity in multiple sclerosis using electronic health records.

ABSTRACT:

Objective

To optimally leverage the scalability and unique features of the electronic health records (EHR) for research that would ultimately improve patient care, we need to accurately identify patients and extract clinically meaningful measures. Using multiple sclerosis (MS) as a proof of principle, we showcased how to leverage routinely collected EHR data to identify patients with a complex neurological disorder and derive an important surrogate measure of disease severity heretofore only available in research settings.

Methods

In a cross-sectional observational study, 5,495 MS patients were identified from the EHR systems of two major referral hospitals using an algorithm that includes codified and narrative information extracted using natural language processing. In the subset of patients who receive neurological care at a MS Center where disease measures have been collected, we used routinely collected EHR data to extract two aggregate indicators of MS severity of clinical relevance multiple sclerosis severity score (MSSS) and brain parenchymal fraction (BPF, a measure of whole brain volume).

Results

The EHR algorithm that identifies MS patients has an area under the curve of 0.958, 83% sensitivity, 92% positive predictive value, and 89% negative predictive value when a 95% specificity threshold is used. The correlation between EHR-derived and true MSSS has a mean R(2) = 0.38±0.05, and that between EHR-derived and true BPF has a mean R(2) = 0.22±0.08. To illustrate its clinical relevance, derived MSSS captures the expected difference in disease severity between relapsing-remitting and progressive MS patients after adjusting for sex, age of symptom onset and disease duration (p = 1.56×10(-12)).

Conclusion

Incorporation of sophisticated codified and narrative EHR data accurately identifies MS patients and provides estimation of a well-accepted indicator of MS severity that is widely used in research settings but not part of the routine medical records. Similar approaches could be applied to other complex neurological disorders.

SUBMITTER: Xia Z

PROVIDER: S-EPMC3823928 | biostudies-literature |

REPOSITORIES: biostudies-literature

ACCESS DATA

Similar Datasets

Project description:BackgroundMulti-center electronic health records (EHR) can support quality improvement initiatives and comparative effectiveness research in stroke care. However, limitations of EHR-based research include challenges in abstracting key clinical variables from non-structured data at scale. This is further compounded by missing data. Here we develop a natural language processing (NLP) model that automatically reads EHR notes to determine the NIH stroke scale (NIHSS) score of patients with acute stroke.MethodsThe study included notes from acute stroke patients (>= 18 years) admitted to the Massachusetts General Hospital (MGH) (2015-2022). The MGH data were divided into training (70%) and hold-out test (30%) sets. A two-stage model was developed to predict the admission NIHSS. A linear model with the least absolute shrinkage and selection operator (LASSO) was trained within the training set. For notes in the test set where the NIHSS was documented, the scores were extracted using regular expressions (stage 1), for notes where NIHSS was not documented, LASSO was used for prediction (stage 2). The reference standard for NIHSS was obtained from Get With The Guidelines Stroke Registry. The two-stage model was tested on the hold-out test set and validated in the MIMIC-III dataset (Medical Information Mart for Intensive Care-MIMIC III 2001-2012) v1.4, using root mean squared error (RMSE) and Spearman correlation (SC).ResultsWe included 4,163 patients (MGH = 3,876; MIMIC = 287); average age of 69 [SD 15] years; 53% male, and 72% white. 90% patients had ischemic stroke and 10% hemorrhagic stroke. The two-stage model achieved a RMSE [95% CI] of 3.13 [2.86-3.41] (SC = 0.90 [0.88-0. 91]) in the MGH hold-out test set and 2.01 [1.58-2.38] (SC = 0.96 [0.94-0.97]) in the MIMIC validation set.ConclusionsThe automatic NLP-based model can enable large-scale stroke severity phenotyping from EHR and therefore support real-world quality improvement and comparative effectiveness studies in stroke.

Project description:BackgroundPassive reporting and laboratory testing delays may limit gastrointestinal (GI) disease outbreak detection. Healthcare systems routinely collect clinical data in electronic medical records (EMRs) that could be used for surveillance. This study's primary objective was to identify data streams from EMRs that may perform well for GI outbreak detection.MethodsZip code-specific daily episode counts in 2009 were generated for 22 syndromic and laboratory-based data streams from Kaiser Permanente Northern California EMRs, covering 3.3 million members. Data streams included outpatient and inpatient diagnosis codes, antidiarrheal medication dispensings, stool culture orders, and positive microbiology tests for six GI pathogens. Prospective daily surveillance was mimicked using the space-time permutation scan statistic in single and multi-stream analyses, and space-time clusters were identified. Serotype relatedness was assessed for isolates in two Salmonella clusters.ResultsPotential outbreaks included a cluster of 18 stool cultures ordered over 5 days in one zip code and a Salmonella cluster in three zip codes over 9 days, in which at least five of six cases had the same rare serotype. In all, 28 potential outbreaks were identified using single stream analyses, with signals in outpatient diagnosis codes most common. Multi-stream analyses identified additional potential outbreaks and in one example, improved the timeliness of detection.ConclusionsGI disease-related data streams can be used to identify potential outbreaks when generated from EMRs with extensive regional coverage. This process can supplement traditional GI outbreak reports to health departments, which frequently consist of outbreaks in well-defined settings (e.g., day care centers and restaurants) with no laboratory-confirmed pathogen. Data streams most promising for surveillance included microbiology test results, stool culture orders, and outpatient diagnoses. In particular, clusters of microbiology tests positive for specific pathogens could be identified in EMRs and used to prioritize further testing at state health departments, potentially improving outbreak detection.

Project description:BackgroundLittle is known about risk factors for early (e.g., erythema migrans) and disseminated Lyme disease manifestations, such as arthritis, neurological complications, and carditis. No study has used both diagnoses and free text to classify Lyme disease by disease stage and manifestation.MethodsWe identified Lyme disease cases in 2012-2016 in the electronic health record (EHR) of a large, integrated health system in Pennsylvania. We developed a rule-based text-matching algorithm using regular expressions to extract clinical data from free text. Lyme disease cases were then classified by stage and manifestation using data from both diagnoses and free text. Among cases classified by stage, we evaluated individual, community, and health care variables as predictors of disseminated stage (vs. early) disease using Poisson regression models with robust errors. Final models adjusted for sociodemographic factors, receipt of Medical Assistance (i.e., Medicaid, a proxy for low socioeconomic status), primary care contact, setting of diagnosis, season of diagnosis, and urban/rural status.ResultsAmong 7310 cases of Lyme disease, we classified 62% by stage. Overall, 23% were classified using both diagnoses and text, 26% were classified using diagnoses only, and 13% were classified using text only. Among the staged diagnoses (n = 4530), 30% were disseminated stage (762 arthritis, 426 neurological manifestations, 76 carditis, 95 secondary erythema migrans, and 76 other manifestations). In adjusted models, we found that persons on Medical Assistance at least 50% of time under observation, compared to never users, had a higher risk (risk ratio [95% confidence interval]) of disseminated Lyme disease (1.20 [1.05, 1.37]). Primary care contact (0.59 [0.54, 0.64]) and diagnosis in the urgent care (0.22 [0.17, 0.29]), compared to the outpatient setting, were associated with lower risk of disseminated Lyme disease.ConclusionsThe associations between insurance payor, primary care status, and diagnostic setting with disseminated Lyme disease suggest that lower socioeconomic status and less health care access could be linked with disseminated stage Lyme disease. Intervening on these factors could reduce the individual and health care burden of disseminated Lyme disease. Our findings demonstrate the value of both diagnostic and narrative text data to identify Lyme disease manifestations in the EHR.

Project description:Determining the etiology of an acute ischemic stroke (AIS) is fundamental to secondary stroke prevention efforts but can be diagnostically challenging. We trained and validated an automated classification machine intelligence tool, StrokeClassifier, using electronic health record (EHR) text data from 2,039 non-cryptogenic AIS patients at 2 academic hospitals to predict the 4-level outcome of stroke etiology determined by agreement of at least 2 board-certified vascular neurologists' review of the stroke hospitalization EHR. StrokeClassifier is an ensemble consensus meta-model of 9 machine learning classifiers applied to features extracted from discharge summary texts by natural language processing. StrokeClassifier was externally validated in 406 discharge summaries from the MIMIC-III dataset reviewed by a vascular neurologist to ascertain stroke etiology. Compared with stroke etiologies adjudicated by vascular neurologists, StrokeClassifier achieved the mean cross-validated accuracy of 0.74 (±0.01) and weighted F1 of 0.74 (±0.01). In the MIMIC-III cohort, the accuracy and weighted F1 of StrokeClassifier were 0.70 and 0.71, respectively. SHapley Additive exPlanation analysis elucidated that the top 5 features contributing to stroke etiology prediction were atrial fibrillation, age, middle cerebral artery occlusion, internal carotid artery occlusion, and frontal stroke location. We then designed a certainty heuristic to deem a StrokeClassifier diagnosis as confidently non-cryptogenic by the degree of consensus among the 9 classifiers, and applied it to 788 cryptogenic patients. This reduced the percentage of the cryptogenic strokes from 25.2% to 7.2% of all ischemic strokes. StrokeClassifier is a validated artificial intelligence tool that rivals the performance of vascular neurologists in classifying ischemic stroke etiology for individual patients. With further training, StrokeClassifier may have downstream applications including its use as a clinical decision support system.