Data Elements In Electronic Health Record Research Articles

Feasibility of structuring electronic health record data to facilitate real-world data research: ICAREdata methods applied to multicenter cancer clinical trials.

The use of electronic health record (EHR) data for research is limited by a lack of structure and a standard data model. The objective of the ICAREdata (Integrating Clinical Trials and Real-World Endpoints Data) project was to structure key research data elements in EHRs using a minimal Common Oncology Data Elements (mCODE) data model to extract and transmit data. The ICAREdata project captured two EHR data elements essential to clinical trials: cancer disease status and treatment plan change. The project was implemented in clinical sites participating in Alliance for Clinical Trials in Oncology trials. Data were extracted from EHRs and sent by secure Fast Healthcare Interoperability Resource messaging (a standard for exchanging EHRs) to a database. Selected elements were compared with corresponding data from the trial's electronic data capture (EDC) system, Medidata Rave. By December 2023, data were extracted and transmitted from 10 sites for 35 patients, involving 367 clinical encounters across 15 clinical trials. Data through March 2023 demonstrated that concordance for the elements treatment plan change and cancer disease status was 79% and 34%, respectively. When disease evaluation was reported by both EHR and EDC (n=15), there was 87% agreement on cancer disease status. Documentation, extraction, and aggregation of structured data elements in EHRs using mCODE and ICAREdata methods is feasible in multi-institutional cancer clinical trials. EDC as a reference data setallowed assessment of the completeness of EHR data capture. Future initiatives will focus on elements with shared definitions in clinical and research environments and efficient workflows. Clinical trials use electronic case report forms to report data, and data must be manually entered on these forms, which is costly and time consuming. ICAREdata methods use structured, organized data from clinical trials that can be more easily shared instead having to enter free text into electronic health records.

Use of Electronic Health Record-Based Measures to Assess Quality of Care for Pediatric Agitation.

Acute agitation during pediatric mental health emergency department (ED) visits presents safety risks to patients and staff. We previously convened multidisciplinary stakeholders who prioritized 20 proposed quality measures for pediatric acute agitation management. Our objectives were to assess feasibility of evaluating performance on these quality measures using electronic health record (EHR) data and to examine performance variation across 3 EDs. At a children's hospital and 2 nonchildren's hospitals, we assessed feasibility of evaluating quality measures for pediatric acute agitation management using structured EHR data elements. We retrospectively evaluated measure performance during ED visits by children 5 to 17 years old who presented for a mental health condition, received medication for agitation, or received physical restraints from July 2020 to June 2021. Bivariate and multivariable regression were used to examine measure performance by patient characteristics and hospital. We identified 2785 mental health ED visits, 275 visits with medication given for agitation, and 35 visits with physical restraints. Performance was feasible to measure using EHR data for 10 measures. Nine measures varied by patient characteristics, including 4.87 times higher adjusted odds (95% confidence interval 1.28-18.54) of physical restraint use among children with versus without autism spectrum disorder. Four measures varied by hospital, with physical restraint use varying from 0.5% to 3.3% of mental health ED visits across hospitals. Quality of care for pediatric acute agitation management was feasible to evaluate using EHR-derived quality measures. Variation in performance across patient characteristics and hospitals highlights opportunities to improve care quality.

Design and evaluation of a web-based electronic health record for amblyopia.

Amblyopia, or lazy eye, is a type of visual impairment in which the eyesight is not complete, even with the use of glasses. For the treatment of this disease, accurate and continuous examinations are needed. Nowadays, patient-centered care, by relying on web-based electronic records for amblyopia, has the potential to reduce treatment costs, increase the quality of care, and improve the safety and effectiveness of treatment. Therefore, the purpose of this study is to design and evaluate an Electronic Health Record (EHR) for patients with amblyopia. The present study is applied developmental research. Using a Morgan table as a sampling tool, a straightforward random sampling technique selected 150 records from 1,500 records that were free of flaws. The design of the electronic version proceeded in a cascading manner so that after the design of each part, it was presented to the amblyopia experts, and if approved, the next part was designed. To design this EHR, the C# programming language and MySQL database were used. A system evaluation was performed by entering and recording patient information. For this purpose, the standard Questionnaire of User Interaction Satisfaction (QUIS), consisting of 18 questions, was used. According to the amblyopia EHR data elements, the data of physician and patient, examinations, website members, and members' roles were determined. After defining the fields and classes that explain the tables, the EHR was designed. The usability evaluation of the system showed that the mean selection of very good and good options by the users of EHRs was over 90%, indicating the patients' acceptance of web-based EHRs. The design of an EHR for amblyopia is an effective step toward integrating and improving the information management of these patients. It will also enable the storage and retrieval of patients' information to reduce and facilitate the control of amblyopia complications.

Development and Validation of Case-Finding Algorithms to Identify Pancreatic Cancer in the Veterans Health Administration.

Survival in pancreatic ductal adenocarcinoma (PDAC) remains poor due to late diagnosis. Electronic Health Records (EHRs) can be used to study this rare disease, but validated algorithms to identify PDAC in the United States EHRs do not currently exist. To develop and validate an algorithm using Veterans Health Administration (VHA) EHR data for the identification of patients with PDAC. We developed two algorithms to identify patients with PDAC in the VHA from 2002 to 2023. The algorithms required diagnosis of exocrine pancreatic cancer in either ≥ 1 or ≥ 2 of the following domains: (i) the VA national cancer registry, (ii) an inpatient encounter, or (iii) an outpatient encounter in an oncology setting. Among individuals identified with ≥ 1 of the above criteria, a random sample of 100 were reviewed by three gastroenterologists to adjudicate PDAC status. We also adjudicated fifty patients not qualifying for either algorithm. These patients died as inpatients and had alkaline phosphatase values within the interquartile range of patients who met ≥ 2 of the above criteria for PDAC. These expert adjudications allowed us to calculate the positive and negative predictive value of the algorithms. Of 10.8 million individuals, 25,533 met ≥ 1 criteria (PPV 83.0%, kappa statistic 0.93) and 13,693 individuals met ≥ 2 criteria (PPV 95.2%, kappa statistic 1.00). The NPV for PDAC was 100%. An algorithm incorporating readily available EHR data elements to identify patients with PDAC achieved excellent PPV and NPV. This algorithm is likely to enable future epidemiologic studies of PDAC.

Identifying virtual care modality in electronic health record data.

Virtual care increased dramatically during the COVID-19 pandemic. The specific modality of virtual care (video, audio, eVisits, eConsults, and remote patient monitoring) has important implications for the accessibility and quality of care, but rates of use are relatively unknown. Methods for identifying virtual care modalities, especially in electronic health records (EHR) are inconsistent. This study (a) developed a method to identify virtual care modalities using EHR data and (b) described the distribution of these modalities over a 3-year study period. EHR data from 316 primary care safety net clinics throughout the study period (4/1/2020-3/31/2023) were included. Visit type (in-person vs virtual) by adults >18 years old were classified. Expert consultation informed the development of two algorithms to classify virtual care visit modalities; these algorithms prioritized different EHR data elements. We conducted descriptive analyses comparing algorithms and the frequency of virtual care modalities. Agreement between the algorithms was 96.5% for all visits and 89.3% for virtual care visits. The majority of disagreement between the algorithms was among encounters scheduled as audio-only but billed as a video visit. Restricting to visits where the algorithms agreed on visit modality, there were 2-fold more audio-only than video visits. Visit modality classification varies depending upon which data in the EHR are prioritized. Regardless of which algorithm is utilized, safety net clinics rely on audio-only and video visits to provide care in virtual visits. Elimination of reimbursement for audio visits may exacerbate existing inequities in care for low-income patients.

Developing an Algorithm to Better Identify Pregnant Women with Inherited Bleeding Disorders within Electronic Health Records

Background: When using electronic health records (EHRs) to conduct population-based studies on inherited bleeding disorders (IBDs), multiple studies demonstrated that using diagnoses codes alone result in a high number of false positive identifications (J Blood Med. 2017 Aug 30;8:131-139). To improve data integrity, in-depth chart reviews are often required which typically can only be done from a local institutional level. This limits the potential to use national research infrastructure such as the National Patient-Centered Clinical Research Network (PCORnet) which draws its data from millions of EHRs across healthcare institutions in the United States. Objective: The objective of this study was to test the diagnostic accuracy of an algorithm that utilized common data definitions and multiple data elements in EHR to identify pregnant women with IBDs. Methods: We used data for pregnant women with IBDs who delivered at our institution from January 1, 2016 to December 31, 2020 to test the accuracy of the algorithm. Women with IBD included those with the following diagnoses: hemophilia and hemophilia carriers, von Willebrand disease and rarer bleeding disorders including factor II, V, VII, X, XI and XIII deficiencies. We did not include the following diagnoses: Platelet function disorder, fibrinogen disorder or hereditary hemorrhagic telangiectasia. We used EHR data elements from EPIC Clarity data model that conformed to those from the PCORnet Common Data Model (CDM). The PCORnet CDM can help ensure consistent data definitions and formats across multiple sites to facilitate large-scale patient-centered research. The original version of our algorithm included the following: (i) ICD-9/10 codes for an inherited bleeding disorder (Table 1) based on discharge diagnosis of any patient encounter visit types including inpatient, outpatient and emergency visits (criterion 1); (ii) Medications used for management of IBD (criterion 2); and (iii) Coagulation factor test and results (criterion 3). A validated diagnosis of IBD required fulfilling either criteria (1) and (2) or criteria (1) and (3). To develop our algorithm in an iterative fashion, we ran sequential queries (queries 1.0-1.1) with revisions to our initial criteria. The results of our retrospective queries were confirmed to have an IBD (true positive) using our local registry and by manual chart review. The local registry is a clinical database that is housed with EPIC and includes accurate demographic, phenotypic, and laboratory data of all persons with a confirmed diagnosis of IBD seen at our federally-funded adult hemophilia treatment center. Sensitivity and the positive predictive value of the algorithm to identify pregnant women with IBDs were calculated. Results: Using our original algorithm (query 1.0), there were 301 pregnant women who fulfilled criterion 1 and had at least one live birth or fetal death (>20 weeks gestation) at our institution during the study period. Of these, only 25 fulfilled criteria 2 and/or criteria 3. For the 276 cases that did not fulfill criteria 2 or 3, we verified the diagnosis through our local registry or through a manual review of the charts. We found that the following ICD-diagnosis codes for IBD - Z14.8; 286.9/D68.8/D68.9; 286.3/D68.2 - resulted in contamination and could not be used to diagnose carriers of IBD or the rarer bleeding disorders. These diagnoses codes were removed and we added a fourth criteria in our revised algorithm (Table 2). For cases that did not fulfill criteria 2 or 3, a case was considered as having a diagnosis of IBD if they had at least two identical ICD diagnosis code for an IBD in at least 2 separate patient encounter visit types (criterion 4). The revised algorithm (query 1.1) identified 35 pregnant women with IBD, of which 32 were confirmed to be true positive. The 3 women incorrectly identified (false positive) by our revised algorithm had seen a hematologist for work-up for a possible IBD which was eventually ruled out. From the original cohort of 301 pregnant women who had an ICD diagnosis code of an IBD, one patient who was a carrier for hemophilia A was missed (false negative) by our revised algorithm. The sensitivity of the revised algorithm was 97.0% and the PPV was 91.4% Conclusion: This study demonstrates the feasibility of an algorithm to accurately identify pregnant women with specific types of IBD within an EHR.

Comparing Natural Language Processing and Structured Medical Data to Develop a Computable Phenotype for Patients Hospitalized Due to COVID-19: Retrospective Analysis.

Throughout the COVID-19 pandemic, many hospitals conducted routine testing of hospitalized patients for SARS-CoV-2 infection upon admission. Some of these patients are admitted for reasons unrelated to COVID-19 and incidentally test positive for the virus. Because COVID-19-related hospitalizations have become a critical public health indicator, it is important to identify patients who are hospitalized because of COVID-19 as opposed to those who are admitted for other indications. We compared the performance of different computable phenotype definitions for COVID-19 hospitalizations that use different types of data from electronic health records (EHRs), including structured EHR data elements, clinical notes, or a combination of both data types. We conducted a retrospective data analysis, using clinician chart review-based validation at a large academic medical center. We reviewed and analyzed the charts of 586 hospitalized individuals who tested positive for SARS-CoV-2 in January 2022. We used LASSO (least absolute shrinkage and selection operator) regression and random forests to fit classification algorithms that incorporated structured EHR data elements, clinical notes, or a combination of structured data and clinical notes. We used natural language processing to incorporate data from clinical notes. The performance of each model was evaluated based on the area under the receiver operator characteristic curve (AUROC) and an associated decision rule based on sensitivity and positive predictive value. We also identified top words and clinical indicators of COVID-19-specific hospitalization and assessed the impact of different phenotyping strategies on estimated hospital outcome metrics. Based on a chart review, 38.2% (224/586) of patients were determined to have been hospitalized for reasons other than COVID-19, despite having tested positive for SARS-CoV-2. A computable phenotype that used clinical notes had significantly better discrimination than one that used structured EHR data elements (AUROC: 0.894 vs 0.841; P<.001) and performed similarly to a model that combined clinical notes with structured data elements (AUROC: 0.894 vs 0.893; P=.91). Assessments of hospital outcome metrics significantly differed based on whether the population included all hospitalized patients who tested positive for SARS-CoV-2 or those who were determined to have been hospitalized due to COVID-19. These findings highlight the importance of cause-specific phenotyping for COVID-19 hospitalizations. More generally, this work demonstrates the utility of natural language processing approaches for deriving information related to patient hospitalizations in cases where there may be multiple conditions that could serve as the primary indication for hospitalization.

Development and validation of a model predicting mild stroke severity on admission using electronic health record data

ObjectiveInitial stroke severity is a potent modifier of stroke outcomes but this information is difficult to obtain from electronic health record (EHR) data. This limits the ability to risk-adjust for evaluations of stroke care and outcomes at a population level. The purpose of this analysis was to develop and validate a predictive model of initial stroke severity using EHR data elements. MethodsThis observational cohort included individuals admitted to a US Department of Veterans Affairs hospital with an ischemic stroke. We extracted 65 independent predictors from the EHR. The primary analysis modeled mild (NIHSS score 0-3) versus moderate/severe stroke (NIHSS score ≥4) using multiple logistic regression. Model validation included: (1) splitting the cohort into derivation (65%) and validation (35%) samples and (2) evaluating how the predicted stroke severity performed in regard to 30-day mortality risk stratification. ResultsThe sample comprised 15,346 individuals with ischemic stroke (n = 10,000 derivation; n = 5,346 validation). The final model included 15 variables and correctly classified 70.4% derivation sample patients and 69.4% validation sample patients. The areas under the curve (AUC) were 0.76 (derivation) and 0.76 (validation). In the validation sample, the model performed similarly to the observed NIHSS in terms of the association with 30-day mortality (AUC: 0.72 observed NIHSS, 0.70 predicted NIHSS). ConclusionsEHR data can be used to construct a surrogate measure of initial stroke severity. Further research is needed to better differentiate moderate and severe strokes, enhance stroke severity classification, and how to incorporate these measures in evaluations of stroke care and outcomes.

Use of Electronic Health Record Data for Drug Safety Signal Identification: A Scoping Review.

Pharmacovigilance programs protect patient health and safety by identifying adverse event signals through postmarketing surveillance of claims data and spontaneous reports. Electronic health records (EHRs) provide new opportunities to address limitations of traditional approaches and promote discovery-oriented pharmacovigilance. To evaluate the current state of EHR-based medication safety signal identification, we conducted a scoping literature review of studies aimed at identifying safety signals from routinely collected patient-level EHR data. We extracted information on study design, EHR data elements utilized, analytic methods employed, drugs and outcomes evaluated, and key statistical and data analysis choices. We identified 81 eligible studies. Disproportionality methods were the predominant analytic approach, followed by data mining and regression. Variability in study design makes direct comparisons difficult. Studies varied widely in terms of data, confounding adjustment, and statistical considerations. Despite broad interest in utilizing EHRs for safety signal identification, current efforts fail to leverage the full breadth and depth of available data or to rigorously control for confounding. The development of best practices and application of common data models would promote the expansion of EHR-based pharmacovigilance.

Use of Structured Electronic Health Records Data Elements for the Development of Computable Phenotypes to Identify Potential Adverse Events Associated with Intravenous Immunoglobulin Infusion.

Detection of adverse reactions to drugs and biologic agents is an important component of regulatory approval and post-market safety evaluation. Real-world data, including insurance claims and electronic health records data, are increasingly used for the evaluation of potential safety outcomes; however, there are different types of data elements available within these data resources, impacting the development and performance of computable phenotypes for the identification of adverse events (AEs) associated with a given therapy. To evaluate the utility of different types of data elements to the performance of computable phenotypes for AEs. We used intravenous immunoglobulin (IVIG) as a model therapeutic agent and conducted a single-center, retrospective study of 3897 individuals who had at least one IVIG administration between 1 January 2014 and 31 December 2019. We identified the potential occurrence of four different AEs, including two proximal AEs (anaphylaxis and heart rate alterations) and two distal AEs (thrombosis and hemolysis). We considered three different computable phenotypes: (1) an International Classification of Disease (ICD)-based phenotype; (2) a phenotype-based on EHR-derived contextual information based on structured data elements, including laboratory values, medication administrations, or vital signs; and (3) a compound phenotype that required both an ICD code for the AE in combination with additional EHR-derived structured data elements. We evaluated the performance of each of these computable phenotypes compared with chart review-based identification of AEs, assessing the positive predictive value (PPV), specificity, and estimated sensitivity of each computable phenotype method. Compound computable phenotypes had a high positive predictive value for acute AEs such as anaphylaxis and bradycardia or tachycardia; however, few patients had both ICD codes and the relevant contextual data, which decreased the sensitivity of these computable phenotypes. In contrast, computable phenotypes for distal AEs (i.e., thrombotic events or hemolysis) frequently had ICD codes for these conditions in the absence of an AE due to a prior history of such events, suggesting that patient medical history of AEs negatively impacted the PPV of computable phenotypes based on ICD codes. These data provide evidence for the utility of different structured data elements in computable phenotypes for AEs. Such computable phenotypes can be used across different data sources for the detection of infusion-related adverse events.

Data quality considerations for evaluating COVID-19 treatments using real world data: learnings from the National COVID Cohort Collaborative (N3C)

BackgroundMulti-institution electronic health records (EHR) are a rich source of real world data (RWD) for generating real world evidence (RWE) regarding the utilization, benefits and harms of medical interventions. They provide access to clinical data from large pooled patient populations in addition to laboratory measurements unavailable in insurance claims-based data. However, secondary use of these data for research requires specialized knowledge and careful evaluation of data quality and completeness. We discuss data quality assessments undertaken during the conduct of prep-to-research, focusing on the investigation of treatment safety and effectiveness.MethodsUsing the National COVID Cohort Collaborative (N3C) enclave, we defined a patient population using criteria typical in non-interventional inpatient drug effectiveness studies. We present the challenges encountered when constructing this dataset, beginning with an examination of data quality across data partners. We then discuss the methods and best practices used to operationalize several important study elements: exposure to treatment, baseline health comorbidities, and key outcomes of interest.ResultsWe share our experiences and lessons learned when working with heterogeneous EHR data from over 65 healthcare institutions and 4 common data models. We discuss six key areas of data variability and quality. (1) The specific EHR data elements captured from a site can vary depending on source data model and practice. (2) Data missingness remains a significant issue. (3) Drug exposures can be recorded at different levels and may not contain route of administration or dosage information. (4) Reconstruction of continuous drug exposure intervals may not always be possible. (5) EHR discontinuity is a major concern for capturing history of prior treatment and comorbidities. Lastly, (6) access to EHR data alone limits the potential outcomes which can be used in studies.ConclusionsThe creation of large scale centralized multi-site EHR databases such as N3C enables a wide range of research aimed at better understanding treatments and health impacts of many conditions including COVID-19. As with all observational research, it is important that research teams engage with appropriate domain experts to understand the data in order to define research questions that are both clinically important and feasible to address using these real world data.

Will clinical standards not be part of the choir? Harmonization between the HL7 gender harmony project model and the NASEM measuring sex, gender identity, and sexual orientation report in the United States.

To propose an approach for semantic and functional data harmonization related to sex and gender constructs in electronic health records (EHRs) and other clinical systems for implementors, as outlined in the National Academies of Sciences, Engineering, and Medicine (NASEM) report Measuring Sex, Gender Identity, and Sexual Orientation and the Health Level 7 (HL7) Gender Harmony Project (GHP) product brief "Gender Harmony-Modeling Sex and Gender Representation, Release 1." Authors from both publications contributed to a plan for data harmonization based upon fundamental principles in informatics, including privacy, openness, access, legitimate infringement, least intrusive alternatives, and accountability. We propose construct entities and value sets that best align with both publications to allow the implementation of EHR data elements on gender identity, recorded sex or gender, and sex for clinical use in the United States. We include usability- and interoperability-focused reasoning for each of these decisions, as well as suggestions for cross-tabulation for populations. Both publications agree on core approaches to conceptualization and measurement of sex- and gender-related constructs. However, some clarifications could improve our ability to assess gender modality, alignment (or lack thereof) between gender identity and assigned gender at birth, and address both individual-level and population-level health inequities. By bridging the GHP and NASEM recommendations, we provide a path forward for implementation of sex- and gender-related EHR elements. Suggestions for implementation of gender identity, recorded sex or gender, and sex for clinical use are provided, along with semantic and functional justifications.

An ordinal severity scale for COVID-19 retrospective studies using Electronic Health Record data.

ObjectivesAlthough the World Health Organization (WHO) Clinical Progression Scale for COVID-19 is useful in prospective clinical trials, it cannot be effectively used with retrospective Electronic Health Record (EHR) datasets. Modifying the existing WHO Clinical Progression Scale, we developed an ordinal severity scale (OS) and assessed its usefulness in the analyses of COVID-19 patient outcomes using retrospective EHR data.Materials and MethodsAn OS was developed to assign COVID-19 disease severity using the Observational Medical Outcomes Partnership common data model within the National COVID Cohort Collaborative (N3C) data enclave. We then evaluated usefulness of the developed OS using heterogenous EHR data from January 2020 to October 2021 submitted to N3C by 63 healthcare organizations across the United States. Principal component analysis (PCA) was employed to characterize changes in disease severity among patients during the 28-day period following COVID-19 diagnosis.ResultsThe data set used in this analysis consists of 2 880 456 patients. PCA of the day-to-day variation in OS levels over the totality of the 28-day period revealed contrasting patterns of variation in disease severity within the first and second 14 days and illustrated the importance of evaluation over the full 28-day period.DiscussionAn OS with well-defined, robust features, based on discrete EHR data elements, is useful for assessments of COVID-19 patient outcomes, providing insights on the progression of COVID-19 disease severity over time.ConclusionsThe OS provides a framework that can facilitate better understanding of the course of acute COVID-19, informing clinical decision-making and resource allocation.

Availability of Structured Data Elements in Electronic Health Records for Supporting Patient Recruitment in Clinical Trials.

Automated identification of eligible patients for clinical trials is an evident secondary use for electronic health records (EHR) data accumulated during routine care. This task requires relevant data elements to be both available in the EHR and in a structured form. This work analyzes these data quality dimensions of EHR data elements corresponding to a selection of frequent eligibility criteria over a total of 436 patient records at 10 university hospitals within the MIRACUM consortium. Data elements from demographics, diagnosis and laboratory findings are typically structured with a completeness of 73 % to 88 % while medication as well as procedures are rather untructured with a completeness of only 44 %. The results can be used to derive suggestions for data quality improvement measures with respect to patient recruitment as well as to serve as a baseline to quantify future developments.

Comparing Machine Learning to Regression Methods for Mortality Prediction Using Veterans Affairs Electronic Health Record Clinical Data.

It is unclear whether machine learning methods yield more accurate electronic health record (EHR) prediction models compared with traditional regression methods. The objective of this study was to compare machine learning and traditional regression models for 10-year mortality prediction using EHR data. This was a cohort study. Veterans Affairs (VA) EHR data. Veterans age above 50 with a primary care visit in 2005, divided into separate training and testing cohorts (n= 124,360 each). The primary outcome was 10-year all-cause mortality. We considered 924 potential predictors across a wide range of EHR data elements including demographics (3), vital signs (9), medication classes (399), disease diagnoses (293), laboratory results (71), and health care utilization (149). We compared discrimination (c-statistics), calibration metrics, and diagnostic test characteristics (sensitivity, specificity, and positive and negative predictive values) of machine learning and regression models. Our cohort mean age (SD) was 68.2 (10.5), 93.9% were male; 39.4% died within 10 years. Models yielded testing cohort c-statistics between 0.827 and 0.837. Utilizing all 924 predictors, the Gradient Boosting model yielded the highest c-statistic [0.837, 95% confidence interval (CI): 0.835-0.839]. The full (unselected) logistic regression model had the highest c-statistic of regression models (0.833, 95% CI: 0.830-0.835) but showed evidence of overfitting. The discrimination of the stepwise selection logistic model (101 predictors) was similar (0.832, 95% CI: 0.830-0.834) with minimal overfitting. All models were well-calibrated and had similar diagnostic test characteristics. Our results should be confirmed in non-VA EHRs. The differences in c-statistic between the best machine learning model (924-predictor Gradient Boosting) and 101-predictor stepwise logistic models for 10-year mortality prediction were modest, suggesting stepwise regression methods continue to be a reasonable method for VA EHR mortality prediction model development.

Automated Pulmonary Embolism Risk Assessment Using the Wells Criteria: Validation Study.

BackgroundComputed tomography pulmonary angiography (CTPA) is frequently used in the emergency department (ED) for the diagnosis of pulmonary embolism (PE), while posing risk for contrast-induced nephropathy and radiation-induced malignancy.ObjectiveWe aimed to create an automated process to calculate the Wells score for pulmonary embolism for patients in the ED, which could potentially reduce unnecessary CTPA testing.MethodsWe designed an automated process using electronic health records data elements, including using a combinatorial keyword search method to query free-text fields, and calculated automated Wells scores for a sample of all adult ED encounters that resulted in a CTPA study for PE at 2 tertiary care hospitals in New York, over a 2-month period. To validate the automated process, the scores were compared to those derived from a 2-clinician chart review.ResultsA total of 202 ED encounters resulted in a completed CTPA to form the retrospective study cohort. Patients classified as “PE likely” by the automated process (126/202, 62%) had a PE prevalence of 15.9%, whereas those classified as “PE unlikely” (76/202, 38%; Wells score >4) had a PE prevalence of 7.9%. With respect to classification of the patient as “PE likely,” the automated process achieved an accuracy of 92.1% when compared with the chart review, with sensitivity, specificity, positive predictive value, and negative predictive value of 93%, 90.5%, 94.4%, and 88.2%, respectively.ConclusionsThis was a successful development and validation of an automated process using electronic health records data elements, including free-text fields, to classify risk for PE in ED visits.

Dental EHR-infused Persona Ontologies to Enrich Dental Dialogue Interaction of Agents.

The quality of patient-provider communication can predict the healthcare outcomes in patients, and therefore, training dental providers to handle the communication effort with patients is crucial. In our previous work, we developed an ontology model that can standardize and represent patient-provider communication, which can later be integrated in conversational agents as tools for dental communication training. In this study, we embark on enriching our previous model with an ontology of patient personas to portray and express types of dental patient archetypes. The Ontology of Patient Personas that we developed was rooted in terminologies from an OBO Foundry ontology and dental electronic health record data elements. We discuss how this ontology aims to enhance the aforementioned dialogue ontology and future direction in executing our model in software agents to train dental students.

Predicting Life Expectancy to Target Cancer Screening Using Electronic Health Record Clinical Data.

Guidelines recommend breast and colorectal cancer screening for older adults with a life expectancy >10 years. Most mortality indexes require clinician data entry, presenting a barrier for routine use in care. Electronic health records (EHR) are a rich clinical data source that could be used to create individualized life expectancy predictions to identify patients for cancer screening without data entry. To develop and internally validate a life expectancy calculator from structured EHR data. Retrospective cohort study using national Veteran's Affairs (VA) EHR databases. Veterans aged 50+ with a primary care visit during 2005. We assessed demographics, diseases, medications, laboratory results, healthcare utilization, and vital signs 1 year prior to the index visit. Mortality follow-up was complete through 2017. Using the development cohort (80% sample), we used LASSO Cox regression to select ~100 predictors from 913 EHR data elements. In the validation cohort (remaining 20% sample), we calculated the integrated area under the curve (iAUC) and evaluated calibration. In 3,705,122 patients, the mean age was 68 years and the majority were male (97%) and white (85%); nearly half (49%) died. The life expectancy calculator included 93 predictors; age and gender most strongly contributed to discrimination; diseases also contributed significantly while vital signs were negligible. The iAUC was 0.816 (95% confidence interval, 0.815, 0.817) with good calibration. We developed a life expectancy calculator using VA EHR data with excellent discrimination and calibration. Automated life expectancy prediction using EHR data may improve guideline-concordant breast and colorectal cancer screening by identifying patients with a life expectancy >10 years.

Leveraging electronic health record data to inform hospital resource management : A systematic data mining approach.

Early identification of resource needs is instrumental in promoting efficient hospital resource management. Hospital information systems, and electronic health records (EHR) in particular, collect valuable demographic and clinical patient data from the moment patients are admitted, which can help predict expected resource needs in early stages of patient episodes. To this end, this article proposes a data mining methodology to systematically obtain predictions for relevant managerial variables by leveraging structured EHR data. Specifically, these managerial variables are: i) Diagnosis categories, ii) procedure codes, iii) diagnosis-related groups (DRGs), iv) outlier episodes and v) length of stay (LOS). The proposed methodology approaches the problem in four stages: Feature set construction, feature selection, prediction model development, and model performance evaluation. We tested this approach with an EHR dataset of 5,089 inpatient episodes and compared different classification and regression models (for categorical and continuous variables, respectively), performed temporal analysis of model performance, analyzed the impact of training set homogeneity on performance and assessed the contribution of different EHR data elements for model predictive power. Overall, our results indicate that inpatient EHR data can effectively be leveraged to inform resource management on multiple perspectives. Logistic regression (combined with minimal redundancy maximum relevance feature selection) and bagged decision trees yielded best results for predicting categorical and numerical managerial variables, respectively. Furthermore, our temporal analysis indicated that, while DRG classes are more difficult to predict, several diagnosis categories, procedure codes and LOS amongst shorter-stay patients can be predicted with higher confidence in early stages of patient stay. Lastly, value of information analysis indicated that diagnoses, medication and structured assessment forms were the most valuable EHR data elements in predicting managerial variables of interest through a data mining approach.

Establishing the common oncology data model for the national biobank consortium of Taiwan.

e19283 Background: Taiwan has 32 biobanks under Government’ governance. The Ministry of Health and Welfare have established a National Biobank Consortium of Taiwan to unify the specimen quality and the medical record database. The total recruited participants exceed 350,000. The National Health Research Institutes in Taiwan hold the responsibility of establish a common data model for aggregating data elements from electronic health records (EHRs) of institutes through direct feeds. The goals are to assemble a set of common oncology data elements and to facilitate cancer data interoperability for patient care and research across institutes of Biobank Consortium. Methods: We first conduct a thorough review of available EHR data elements for patient characteristics, diagnosis/staging, treatments, laboratory results, vital signs and outcomes. The data dictionary was organized based on HL7 FHIR and also included data elements from Taiwan Cancer Registry (TCR) and National Health Insurance (NHI) Program, which the common definition has already been established and implemented for years. Data elements suggested by ASCO CancerLinQ and minimal Common Oncology Data Elements (mCODE) are also referenced during planning. The final common model was then reviewed by a panel of experts consisting oncologists as well as data science specialists. Results: There are finally 9 data tables with 281 data elements, in which 248 of them are from the routinely uploaded data elements to government agencies (TCR &amp; NHI) and 33 elements are collected with partial common definition among institutes. There are 164 data elements which are to be collected one observation per case, while 117 elements will be accumulated periodically. Conclusions: A comprehensive understanding of genetics, phenotypes, disease variation as well as treatment responses is crucial to fulfill the needs of real-world studies, which potentially would lead to personalized treatment and drug development. At the first stage of this project, we aim to accumulate available EHR structured data elements and to maintain sufficient cancer data quality. Consequently, the database can provide real-world evidence to promote evidence-based &amp; data-driven cancer care.