Analysis Of Electronic Health Record Data Research Articles

A Multimodal Intervention to Improve Guideline-Based Screening for Alpha-1 Antitrypsin Deficiency in a Community Health Setting.

Evidence-based guidelines recommend screening all individuals with chronic obstructive pulmonary disease (COPD) for the genetic disorder alpha-1 antitrypsin deficiency (AATD). However, it is estimated that only 5% of people with COPD have been tested for AATD, and a large fraction of the estimated 70,000 to 100,000 Americans with AATD have not yet been diagnosed. Low familiarity with AATD and limited knowledge about diagnostic tests and available treatments contribute to suboptimal screening rates. To address barriers to and improve rates of guideline-based AATD diagnostic testing among racially and ethnically diverse patients with COPD at a large community health center. A quality improvement initiative consisting of educational sessions and electronic health record (EHR) system interventions was implemented to improve the adoption of guideline-based screening for AATD in patients with COPD. An analysis of EHR data demonstrated that of patients with a COPD diagnosis (N = 1,030), 22.2% (n = 229) were screened for AATD in the 12 months following the start of the quality improvement initiative compared with 1.3% (n = 13) of patients with a COPD diagnosis (N = 972) seen in the 12 months prior to the start of the quality improvement initiative (P < 0.001). A quality improvement initiative consisting of educational sessions and EHR system modifications was successful in increasing clinicians' knowledge and diagnostic screening rates for AATD in patients with COPD at a large community health center.

Favorable Treatment Responses of Recurrent Focal Segmental Glomerulosclerosis in Children after Kidney Transplantation: A Contemporary Multicenter Electronic Health Record Data Analysis

Favorable Treatment Responses of Recurrent Focal Segmental Glomerulosclerosis in Children after Kidney Transplantation: A Contemporary Multicenter Electronic Health Record Data Analysis

Abstract MP20: Leveraging Electronic Health Records to Assess Neighborhood Advantages and Risk of Cardiovascular Outcomes Among Hypertensive Patients

Introduction: Hypertension disproportionately impacts socially disadvantaged groups in the U.S. with increasing prevalence and suboptimal control rates. Our study aims to examine the association between neighborhood-level social vulnerability and hypertension outcomes using electronic health record (EHR) data from a large regional health system. Methods: We conducted a retrospective analysis of EHR data from the Sentara Health System, covering the period from January 1st, 2010, to December 31st, 2022. We assigned a Social Vulnerability Index (SVI) based on the residential census tract of each patient and established a longitudinal cohort of patients who consistently utilized healthcare services. To be included in the study, patients were required to have a minimum of five years of observational time and to have healthcare visits in at least 50% of the observed years. The study focused on patients diagnosed with hypertension—either through a hypertension diagnosis, a blood pressure measurement exceeding 140/90, or antihypertensive medication prescriptions. We utilized multivariate Cox Proportional Hazards models to analyze the associations of SVI with cardiovascular outcomes and blood pressure control (&lt;140/90 mmHg). Results: The longitudinal cohort consisted of 55,060 patients, with an average age of 54 years (Interquartile range: 42-65), 30.0% of whom were Black adults and 60.0% were females. Compared to patients residing in the most advantaged neighborhoods (quartile 1 of SVI), those in more disadvantaged neighborhoods (quartiles 2, 3, and 4) had hazard ratios for composite cardiovascular endpoint (myocardial infarction, heart failure, and stroke) of 1.22 (95% CI, 1.19-1.25), 1.40 (95% CI, 1.37-1.44), and 1.64 (95% CI, 1.60-1.68), respectively, after adjustment for age and sex. Additionally, blood pressure control was significantly lower in higher SVI quartiles, with control rates in quartiles 1 through 4 at 78.4%, 76.5%, 75.2%, and 72.8%, respectively (P&lt;0.001). Among the SVI themes, socioeconomic status (theme 1) and housing type and transportation (theme 4) were associated with the highest hazard ratios for cardiovascular outcomes. Conclusions: Neighborhood-level social vulnerability is significantly associated with worse hypertension control and adverse cardiovascular outcomes. Health systems can employ tailored interventions targeting high-risk neighborhoods to reduce disparities and enhance health outcomes for vulnerable populations.

Graph and text multi-modal representation learning with momentum distillation on Electronic Health Records

The emergence and widespread adoption of electronic health records (EHR) in modern healthcare systems has generated a large amount of data with the potential to significantly improve patient outcomes. However, extracting meaningful insights from EHR is challenging due to their extensive volume and inherent complexity. Leveraging diverse modalities of data enhances the exploration of patient data representation. Nevertheless, current research on multi-modal relationships between concealed graph structures within medical codes and unstructured text overlooks inherent disparities and inconsistencies. To address the significant gap, this study introduces a pretraining approach named graph and text multi-modal representation learning with momentum distillation on Electronic Health Records (GTMMRL). This approach tackles the challenge of noisy and unreliable labels using a large open-source EHR dataset, MIMIC-III, for pretraining. Five carefully selected proxy tasks are employed, each guiding the student model’s learning through pseudo-targets generated by the teacher model, which is an exponential moving average of the student model, thereby mitigating overfitting tendencies. We evaluate the performance of our model in five downstream tasks, each addressing real-world challenges in EHR data analysis. Results showcase that our method achieves state-of-the-art performance in various downstream tasks, highlighting the model’s critical role in enhancing the quality of insights extracted from complex EHR data. The code is available at https://github.com/deepEMR/GTMMRL.

Supplementing systematic review findings with healthcare system data: pilot projects from the Agency for Healthcare Research and Quality Evidence-based Practice Center program

ObjectivesThe US Agency for Healthcare Research and Quality, through the Evidence-based Practice Center (EPC) Program, aims to provide health system decision makers with the highest-quality evidence to inform clinical decisions. However, limitations in the literature may lead to inconclusive findings in EPC systematic reviews (SRs). The EPC Program conducted pilot projects to understand the feasibility, benefits, and challenges of utilizing health system data to augment SR findings to support confidence in healthcare decision-making based on real-world experiences. Study Design and SettingThree contractors (each an EPC located at a different health system) selected a recently completed SR conducted by their center and identified an evidence gap that electronic health record (EHR) data might address. All pilot project topics addressed clinical questions as opposed to care delivery, care organization, or care disparities topics that are common in EPC reports. Topic areas addressed by each EPC included infantile epilepsy, migraine, and hip fracture. EPCs also tracked additional resources needed to conduct supplemental analyses. The workgroup met monthly in 2022-2023 to discuss challenges and lessons learned from the pilot projects. ResultsTwo supplemental data analyses filled an evidence gap identified in the SRs (raised certainty of evidence, improved applicability) and the third filled a health system knowledge gap. Project challenges fell under three themes: regulatory and logistical issues, data collection and analysis, and interpretation and presentation of findings. Limited ability to capture key clinical variables given inconsistent or missing data within the EHR was a major limitation. The workgroup found that conducting supplemental data analysis alongside an SR was feasible but adds considerable time and resources to the review process (estimated total hours to complete pilot projects ranged from 283 to 595 across EPCs), and that the increased effort and resources added limited incremental value. ConclusionSupplementing existing SRs with analyses of EHR data is resource intensive and requires specialized skillsets throughout the process. While using EHR data for research has immense potential to generate real-world evidence and fill knowledge gaps, these data may not yet be ready for routine use alongside SRs.

Barriers to Optimal Clinician Guideline Adherence in Management of Markedly Elevated Blood Pressure

Hypertension poses a substantial public health challenge. Despite clinical practice guidelines for hypertension management, clinician adherence to these guidelines remains suboptimal. To develop a taxonomy of suboptimal adherence scenarios for severe hypertension and identify barriers to guideline adherence. This qualitative content analysis using electronic health records (EHRs) of Yale New Haven Health System included participants who had at least 2 consecutive visits with markedly elevated blood pressure (BP; defined as at least 2 consecutive readings of systolic BP ≥160 mm Hg and diastolic BP ≥100 mm Hg) between January 1, 2013, and December 31, 2021, and no prescription for antihypertensive medication within a 90 days of the second BP measurement. Data analysis was conducted from January to December 2023. The primary outcome was scenarios and influencing factors contributing to clinician nonadherence to the guidelines for hypertension management. A thematic analysis of EHR data was conducted to generate a pragmatic taxonomy of scenarios of suboptimal clinician guideline adherence in the management of severe hypertension. Of the 20 654 patients who met criteria, 200 were randomly selected and thematic saturation was reached after analyzing 100 patients (mean [SD] age at index visit, 66.5 [12.8] years; 50 female [50%]; 8 Black [8%]; 5 Hispanic or Latino [5%]; 85 White [85%]). Three content domains emerged: (1) clinician-related scenarios (defined as noninitiation or nonintensification of treatment due to issues relating to clinician intention, capability, or scope), which included 2 subcategories (did not address and diffusion of responsibility); (2) patient-related scenarios (defined as noninitiation or nonintensification of treatment due to patient behavioral considerations), which included 2 subcategories (patient nonadherence and patient preference); and (3) clinical complexity-related scenarios (defined as noninitiation or nonintensification of treatment due to clinical situational complexities), which included 3 subcategories (diagnostic uncertainty, maintenance of current intervention, and competing medical priorities). In this qualitative study of EHR data, a taxonomy of suboptimal adherence scenarios for severe hypertension was developed and barriers to guideline adherence were identified. This pragmatic taxonomy lays the foundation for developing targeted interventions to improve clinician adherence to guidelines and patient outcomes.

Identification of Time-Series Pattern Marker in Its Application to Mortality Analysis of Pneumonia Patients in Intensive Care Unit.

Electronic Health Records (EHRs) are a significant source of big data used to track health variables over time. The analysis of EHR data can uncover medical markers or risk factors, aiding in the diagnosis and monitoring of diseases. We introduce a novel method for identifying markers with various temporal trend patterns, including monotonic and fluctuating trends, using machine learning models such as Long Short-Term Memory (LSTM). By applying our method to pneumonia patients in the intensive care unit using the MIMIC-III dataset, we identified markers exhibiting both monotonic and fluctuating trends. Specifically, monotonic markers such as red cell distribution width, urea nitrogen, creatinine, calcium, morphine sulfate, bicarbonate, sodium, troponin T, albumin, and prothrombin time were more frequently observed in the mortality group compared to the recovery group throughout the 10-day period before discharge. Conversely, fluctuating trend markers such as dextrose in sterile water, polystyrene sulfonate, free calcium, and glucose were more frequently observed in the mortality group as the discharge date approached. Our study presents a method for detecting time-series pattern markers in EHR data that respond differently according to disease progression. These markers can contribute to monitoring disease progression and enable stage-specific treatment, thereby advancing precision medicine.

Applications of machine learning to electronic health record data in liver-related disease

Electronic Health Records (EHRs) has gained its increasing significance in modern healthcare as its promising prospects in the application of machine learning. The accumulation of vast clinical data holds potential for repurposing in clinical research such as prediction, diagnosis and prognosis. However, the collection and preparation of EHR data present challenges, primarily due to the inherent incompleteness of the data and the associated privacy and security concerns. To address the issue, text-mining tools based on domain-specific lexicons, data sharing and multiple de-identification methods have been suggested. In terms of methodologies, various machine learning models and algorithms used in EHR data analysis are analyzed, including logistic regression, decision trees, random forests, and natural language processing, each with its unique application scenarios in the healthcare domain. Liver related diseases, including HAV, HBV, HCV and especially Liver Cancer, has affected hundreds of millions of people around the world. The incidence and mortality rates for these diseases are still rising continually. With recent advancements of Machine Learning techniques, such as the attention mechanism and BERT-based embedding, which have shown exceptional results in EHR analysis when applying to liver diseases. While EHRs offer a treasure trove of data for clinical research, the challenges associated with their collection, processing, and analysis cannot be ignored. It underscores the need for robust methodologies and tools to harness the full potential of EHRs while ensuring data integrity and patient privacy. In this paper, we will gather and review the existing application in the realm of liver-related diseases.

HealthGAT: Node Classifications in Electronic Health Records using Graph Attention Networks.

While electronic health records (EHRs) are widely used across various applications in healthcare, most applications use the EHRs in their raw (tabular) format. Relying on raw or simple data pre-processing can greatly limit the performance or even applicability of downstream tasks using EHRs. To address this challenge, we present HealthGAT, a novel graph attention network framework that utilizes a hierarchical approach to generate embeddings from EHR, surpassing traditional graph-based methods. Our model iteratively refines the embeddings for medical codes, resulting in improved EHR data analysis. We also introduce customized EHR-centric auxiliary pre-training tasks to leverage the rich medical knowledge embedded within the data. This approach provides a comprehensive analysis of complex medical relationships and offers significant advancement over standard data representation techniques. HealthGAT has demonstrated its effectiveness in various healthcare scenarios through comprehensive evaluations against established methodologies. Specifically, our model shows outstanding performance in node classification and downstream tasks such as predicting readmissions and diagnosis classifications.

Variations in Using Diagnosis Codes for Defining Age-Related Macular Degeneration Cohorts

Data harmonization is vital for secondary electronic health record data analysis, especially when combining data from multiple sources. Currently, there is a gap in knowledge as to how studies identify cohorts of patients with age-related macular degeneration (AMD), a leading cause of blindness. We hypothesize that there is variation in using medical condition codes to define cohorts of AMD patients that can lead to either the under- or overrepresentation of such cohorts. This study identified articles studying AMD using the International Classification of Diseases (ICD-9, ICD-9-CM, ICD-10, and ICD-10-CM). The data elements reviewed included the year of publication; dataset origin (Veterans Affairs, registry, national or commercial claims database, and institutional EHR); total number of subjects; and ICD codes used. A total of thirty-seven articles were reviewed. Six (16%) articles used cohort definitions from two ICD terminologies. The Medicare database was the most used dataset (14, 38%), and there was a noted increase in the use of other datasets in the last few years. We identified substantial variation in the use of ICD codes for AMD. For the studies that used ICD-10 terminologies, 7 (out of 9, 78%) defined the AMD codes correctly, whereas, for the studies that used ICD-9 and 9-CM terminologies, only 2 (out of 30, 7%) defined and utilized the appropriate AMD codes (p = 0.0001). Of the 43 cohort definitions used from 37 articles, 31 (72%) had missing or incomplete AMD codes used, and only 9 (21%) used the exact codes. Additionally, 13 articles (35%) captured ICD codes that were not within the scope of AMD diagnosis. Efforts to standardize data are needed to provide a reproducible research output.

Reducing Firearm Access for Suicide Prevention: Implementation Evaluation of the Web-Based "Lock to Live" Decision Aid in Routine Health Care Encounters.

"Lock to Live" (L2L) is a novel web-based decision aid for helping people at risk of suicide reduce access to firearms. Researchers have demonstrated that L2L is feasible to use and acceptable to patients, but little is known about how to implement L2L during web-based mental health care and in-person contact with clinicians. The goal of this project was to support the implementation and evaluation of L2L during routine primary care and mental health specialty web-based and in-person encounters. The L2L implementation and evaluation took place at Kaiser Permanente Washington (KPWA)-a large, regional, nonprofit health care system. Three dimensions from the RE-AIM (Reach, Effectiveness, Adoption, Implementation, Maintenance) model-Reach, Adoption, and Implementation-were selected to inform and evaluate the implementation of L2L at KPWA (January 1, 2020, to December 31, 2021). Electronic health record (EHR) data were used to purposefully recruit adult patients, including firearm owners and patients reporting suicidality, to participate in semistructured interviews. Interview themes were used to facilitate L2L implementation and inform subsequent semistructured interviews with clinicians responsible for suicide risk mitigation. Audio-recorded interviews were conducted via the web, transcribed, and coded, using a rapid qualitative inquiry approach. A descriptive analysis of EHR data was performed to summarize L2L reach and adoption among patients identified at high risk of suicide. The initial implementation consisted of updates for clinicians to add a URL and QR code referencing L2L to the safety planning EHR templates. Recommendations about introducing L2L were subsequently derived from the thematic analysis of semistructured interviews with patients (n=36), which included (1) "have an open conversation," (2) "validate their situation," (3) "share what to expect," (4) "make it accessible and memorable," and (5) "walk through the tool." Clinicians' interviews (n=30) showed a strong preference to have L2L included by default in the EHR-based safety planning template (in contrast to adding it manually). During the 2-year observation period, 2739 patients reported prior-month suicide attempt planning or intent and had a documented safety plan during the study period, including 745 (27.2%) who also received L2L. Over four 6-month subperiods of the observation period, L2L adoption rates increased substantially from 2% to 29% among primary care clinicians and from <1% to 48% among mental health clinicians. Understanding the value of L2L from users' perspectives was essential for facilitating implementation and increasing patient reach and clinician adoption. Incorporating L2L into the existing system-level, EHR-based safety plan template reduced the effort to use L2L and was likely the most impactful implementation strategy. As rising suicide rates galvanize the urgency of prevention, the findings from this project, including L2L implementation tools and strategies, will support efforts to promote safety for suicide prevention in health care nationwide.

Call me Dr Ishmael: trends in electronic health record notes available at emergency department visits and admissions.

Numerous studies have identified information overload as a key issue for electronic health records (EHRs). This study describes the amount of text data across all notes available to emergency physicians in the EHR, trended over the time since EHR establishment. We conducted a retrospective analysis of EHR data from a large healthcare system, examining the number of notes and a corresponding number of total words and total tokens across all notes available to physicians during patient encounters in the emergency department (ED). We assessed the change in these metrics over a 17-year period between 2006 and 2023. The study cohort included 730968 ED visits made by 293559 unique patients and a total note count of 132574964. The median note count for all encounters in 2006 was 5 (IQR 1-16), accounting for 1735 (IQR 447-5521) words. By the last full year of the study period, 2022, the median number of notes had grown to 359 (IQR 84-943), representing 359 (IQR 84-943) words. Note and word counts were higher for admitted patients. The volume of notes available for review by providers has increased by over 30-fold in the 17 years since the implementation of the EHR at a large health system. The task of reviewing these notes has become commensurately more difficult. These data point to the critical need for new strategies and tools for filtering, synthesizing, and summarizing information to achieve the promise of the medical record.

Disparities in Eye Care Utilization Among Refugee and Migrant Populations.

In this cross-sectional study, we examined refugee/migrant participants' health and eye care utilization compared to controls in San Diego County. Data were collected from electronic health records (EHRs) at UCSD Health-affiliated medical centers. Through a manual review of EHRs, eligibility criteria to identify a cohort were developed. A total of 64 refugee/migrant participants and 95 control participants matched based on country of origin, age, and sex were included in the analysis. Demographic characteristics, insurance type, and vision/eye care utilization were compared between the two groups. A greater proportion of refugee/migrant participants were more likely to be enrolled in government-sponsored insurance programs, predominantly Medicaid when compared to controls (55% vs. 24%, P = < 0.01). When adjusting for age, history of ophthalmic procedure, and surgery, refugee status was associated with fewer encounters with ophthalmologists in a multivariable linear regression model (coefficient = -1.66 [95% confidence interval [CI] = -2.89 to -0.44], P = 0.009). This study highlights disparities in eye care utilization for refugee/migrant populations. When compared to controls, a larger proportion of refugees/migrants had government-funded insurance, and refugee status was associated with fewer encounters with ophthalmologists. These findings underscore the need for further research on this population to better understand potential healthcare barriers these individuals may encounter. This analysis of EHR data illustrates disparities in eye care experienced by refugees/migrants, highlighting potential gaps in care in a vulnerable population.

Maternal Tdap and influenza vaccination uptake 2017-2021 in the United States: Implications for maternal RSV vaccine uptake in the future

BackgroundAssessment of maternal vaccine coverage is important for understanding and quantifying the impact of currently recommended vaccines as well as modeling the potential impact of future vaccines. However, existing data lack detail regarding uptake according to week of gestational age (wGA). Such granularity is valuable for more accurate estimation of vaccine impact. ObjectiveTo summarize contemporary maternal Tdap vaccination uptake, overall, yearly, and by wGA, and maternal influenza vaccination uptake, overall, by influenza observation year, immunization month, and delivery month, in the US. MethodsFemale patients 18-49 years of age with a pregnancy resulting in a live born infant (i.e., delivery) between 2017 and 2021 were selected from the Optum electronic health records (EHRs) database. Recently published gestational age algorithms were utilized to estimate wGA. ResultsOf 1,021,260 deliveries among 886,660 women between 2017-2021, 55.1% had Tdap vaccination during pregnancy; vaccine coverage varied slightly by year (2017: 56.6%; 2018: 55.2%; 2019: 55.2%; 2020: 54.7%; 2021: 52.1%). Most (64.4%) maternal Tdap vaccinations occurred 27-32 wGA; 79.5% occurred during the entire 10-week recommended vaccination window (27-36 wGA). In the evaluation of influenza vaccination uptake (n=798,113 deliveries; 714,841 women), 33.5% of deliveries had influenza vaccination during influenza observation years 2017-2021, most (73.0%) of which occurred during influenza peak activity months (October-January) with approximately one-quarter (27.0%) of vaccinations having occurred during the off-peak months, mostly in September. ConclusionsIn this large contemporary analysis of EHR data, uptake of Tdap vaccination during pregnancy was consistent with previously published estimates; notably, most vaccination occurred early in the recommended 27-36 wGA window. Maternal influenza vaccination uptake largely correlated with peak influenza activity months and not gestational age. These study findings may have important implications for estimating the potential uptake and impact of future maternal vaccines.

Challenges in and Opportunities for Electronic Health Record-Based Data Analysis and Interpretation.

Electronic health records (EHRs) have been increasingly adopted in clinical practices across the United States, providing a primary source of data for clinical research, particularly observational cohort studies. EHRs are a high-yield, low-maintenance source of longitudinal real-world data for large patient populations and provide a wealth of information and clinical contexts that are useful for clinical research and translation into practice. Despite these strengths, it is important to recognize the multiple limitations and challenges related to the use of EHR data in clinical research. Missing data are a major source of error and biases and can affect the representativeness of the cohort of interest, as well as the accuracy of the outcomes and exposures. Here, we aim to provide a critical understanding of the types of data available in EHRs and describe the impact of data heterogeneity, quality, and generalizability, which should be evaluated prior to and during the analysis of EHR data. We also identify challenges pertaining to data quality, including errors and biases, and examine potential sources of such biases and errors. Finally, we discuss approaches to mitigate and remediate these limitations. A proactive approach to addressing these issues can help ensure the integrity and quality of EHR data and the appropriateness of their use in clinical studies.

A systematic review of artificial intelligence algorithms for predicting acute kidney injury.

Acute kidney injury (AKI) increases mortality and costs in hospitalized patients. New methods for early AKI identification have been developed with targeted biomarkers and electronic health records data analysis. Machine learning (ML) use in diagnostics and health data analysis has recently increased. We performed a systematic review to analyze the use of ML for AKI prediction in hospitalized adults. Tubmed, EMBASE, Cochrane, and Web of Science databases were searched until 31st March of 2023. English-language studies using ML in adults for AKI prediction were included using predetermined eligibility search terms such as acute kidney injury, machine learning, and artificial intelligence. Two reviewers evaluated the publications' titles, abstracts, and full texts separately and obtained appropriate data. The main outcome was an area under the curve (AUC) result of at least 0.70. Ten studies in 102 articles were included involving 242,251 patients. Deep learning (AUC 0.907 in critical care AKI; AUC 0.797 in hospitalized patients AKI) was similar to Logistic regression (AUC 0.877 in critical care AKI; AUC 0.789 in hospitalized patients). Decision tree constructions had similar AUC. In this review, most ML models analyzed fulfilled the main outcome. AKI is multifactorial; however, ML performed well with different etiologies, such as cardiac-related AKI, drug-related AKI, and critical care patients. Overfitting data and constructing black box models are limitations that might jeopardize the generalization and comprehension of the results. Most studies were single-center, and three manuscripts used the same database with a predominantly Caucasian population, resulting in a lack of diversity and reducing external generalization. In conclusion, ML could effectively predict AKI in hospitalized adults. Future directions rely on including a more diverse population and completing prospective and controlled trials.

Abstract 050: Leveraging Electronic Health Records To Assess Delay In Hypertension Diagnosis: Insights From A Large Regional Health System

Introduction: Timely hypertension diagnosis is vital for effective management and prevention of complications. This study assessed hypertension diagnosis delays in a large regional health system using electronic health records (EHRs). Methods: Retrospective analysis of EHR data (2010-2021) in Sentara Healthcare System was conducted. Hypertension diagnosis was defined as two outpatient blood pressure (BP) readings &gt;=140/90 mmHg within two years, with at least 30 days separation. The delay in diagnosis was determined as the time elapsed from the computable diagnosis to the structured diagnosis documented in the EHR. Results: Among 302,774 patients with computable hypertension diagnosis, 23.6% lacked a structured diagnosis. Furthermore, 23.8% received early structured diagnosis (after first BP elevation), and 18.6% received eventual structured diagnosis (after second BP elevation). Among those who received an eventual structured diagnosis, the median delay in hypertension diagnosis was 21.7 months (interquartile range: 7.5-45.5 months). Prescription rate for antihypertensive medication was significantly lower for patients with no structured diagnosis (44.2%) and eventual structured diagnosis (38.1%) than for those with early structured diagnosis (82.3%). Longer delays were associated with older age, male gender, and Black patients. Conclusions: The study highlights concerning delays in hypertension diagnosis, impacting timely intervention and complication prevention. By leveraging EHR data, health systems can identify areas for improvement and enhance the efficiency of hypertension diagnosis, leading to improved patient outcomes.

Capturing Semantic Relationships in Electronic Health Records Using Knowledge Graphs: An Implementation Using MIMIC III Dataset and GraphDB.

Electronic health records (EHRs) are an increasingly important source of information for healthcare professionals and researchers. However, EHRs are often fragmented, unstructured, and difficult to analyze due to the heterogeneity of the data sources and the sheer volume of information. Knowledge graphs have emerged as a powerful tool for capturing and representing complex relationships within large datasets. In this study, we explore the use of knowledge graphs to capture and represent complex relationships within EHRs. Specifically, we address the following research question: Can a knowledge graph created using the MIMIC III dataset and GraphDB effectively capture semantic relationships within EHRs and enable more efficient and accurate data analysis? We map the MIMIC III dataset to an ontology using text refinement and Protege; then, we create a knowledge graph using GraphDB and use SPARQL queries to retrieve and analyze information from the graph. Our results demonstrate that knowledge graphs can effectively capture semantic relationships within EHRs, enabling more efficient and accurate data analysis. We provide examples of how our implementation can be used to analyze patient outcomes and identify potential risk factors. Our results demonstrate that knowledge graphs are an effective tool for capturing semantic relationships within EHRs, enabling a more efficient and accurate data analysis. Our implementation provides valuable insights into patient outcomes and potential risk factors, contributing to the growing body of literature on the use of knowledge graphs in healthcare. In particular, our study highlights the potential of knowledge graphs to support decision-making and improve patient outcomes by enabling a more comprehensive and holistic analysis of EHR data. Overall, our research contributes to a better understanding of the value of knowledge graphs in healthcare and lays the foundation for further research in this area.

Deep learning approach for early prediction of COVID-19 mortality using chest X-ray and electronic health records

BackgroundAn artificial-intelligence (AI) model for predicting the prognosis or mortality of coronavirus disease 2019 (COVID-19) patients will allow efficient allocation of limited medical resources. We developed an early mortality prediction ensemble model for COVID-19 using AI models with initial chest X-ray and electronic health record (EHR) data.ResultsWe used convolutional neural network (CNN) models (Inception-ResNet-V2 and EfficientNet) for chest X-ray analysis and multilayer perceptron (MLP), Extreme Gradient Boosting (XGBoost), and random forest (RF) models for EHR data analysis. The Gradient-weighted Class Activation Mapping and Shapley Additive Explanations (SHAP) methods were used to determine the effects of these features on COVID-19. We developed an ensemble model (Area under the receiver operating characteristic curve of 0.8698) using a soft voting method with weight differences for CNN, XGBoost, MLP, and RF models. To resolve the data imbalance, we conducted F1-score optimization by adjusting the cutoff values to optimize the model performance (F1 score of 0.77).ConclusionsOur study is meaningful in that we developed an early mortality prediction model using only the initial chest X-ray and EHR data of COVID-19 patients. Early prediction of the clinical courses of patients is helpful for not only treatment but also bed management. Our results confirmed the performance improvement of the ensemble model achieved by combining AI models. Through the SHAP method, laboratory tests that indicate the factors affecting COVID-19 mortality were discovered, highlighting the importance of these tests in managing COVID-19 patients.

Variation in antibiotic resistance patterns for children and adults treated at 166 non-affiliated US facilities using EHR data.

Antibiotic resistance (AR) is a global public health threat. Surveillance of baseline AR and trends and emerging resistance among priority bacterial isolates with respect to the age of the patients and the type of healthcare setting are required due to differences in antimicrobial need and use in these populations. We performed a retrospective study using deidentified electronic health record (EHR) data in the Cerner Health Facts™ data warehouse. Antibiotic susceptibility data were extracted for all bacterial isolates of interest at 166 non-affiliated healthcare facilities reporting microbiology susceptibility results of the FDA recommended antibiotics between the years 2012 to 2017. We assessed and visualized the slope coefficient from linear regression to compare changes in resistance over time for the four patient care groups. The trends in resistance rates to clinically relevant antibiotics were influenced by age and care setting. For example, ertapenem-resistant Enterobacter cloacae isolates from children overall increased significantly compared with adults (0.7% to 9.8%, 2.1% to 2.8%, P = 0.00013) and isolates from children in paediatric facilities increased significantly compared with facilities treating adults and children (0.1% to 27.1%, 0.9% to 3.8%, P = 0.0002). Large-scale analysis of EHR data from 166 facilities shows that AR patterns for some bug-drug combinations vary by care setting and patient age. We describe novel data visualizations to interpret large-scale EHR data on the prevalence and trends of AR that should influence antimicrobial prescribing and antimicrobial stewardship programme interventions.