Electronic Health Record Analysis Research Articles

Leveraging Electronic Health Records (EHR) for Enhanced Patient Care

This study analyzes how electronic health records (EHR) can be utilized to enhance patient care. The objectives of the study are to; design predictive models to recognize patients are risk of developing some specific conditions; evaluate the effect of EHR systems on patient care outcomes; ascertain the impact of EHR systems on physicians’ work; and develop a framework for incorporating EHR analysis into hospital systems, thus enabling real-time decision-making by healthcare specialists. This study adopts a qualitative research using semi-structured interviews, which was directed to physicians who had at least a year experience at Primary Health Care Centres (PHCs) to gather information about the current content and scope of Electronic Health Record Systems. Most of the respondents interviewed said that the health information system’s adaptation and the availability of Electronic Health Records influenced encounters for patients positively in their PHCs. They talked about the many benefits of EHR in terms of effectiveness, which include the availability of patient information and accuracy of patient documentation; these are often hard to find in their paper documents. The interviewees also indicated that EHRs allow eprescribe of medication, which is a huge plus to the old system of prescribing and administering medicine to patients. The physicians stressed the need for private clinics and primary healthcare centres to share and link their systems to enhance the functionality of the EHR. The study recommends that; there is need for the PHC EHR system to integrate across other government and private institutions. There is also a need for increased capacity to support patients with physical activity and nutrition.

Graph learning with label attention and hyperbolic embedding for temporal event prediction in healthcare

The digitization of healthcare systems has led to the proliferation of electronic health records (EHRs), serving as comprehensive repositories of patient information. However, the vast volume and complexity of EHR data present challenges in extracting meaningful insights. This paper addresses the need for automated analysis of EHRs by proposing a novel graph learning model with label attention (GLLA) for temporal event prediction. GLLA utilizes graph neural networks to capture intricate relationships between medical codes and patients, incorporating hierarchical structures and shared risk factors. Furthermore, it introduces the Label Attention and Attention-based Transformer (LAAT) algorithm to analyze unstructured clinical notes as a multi-label classification problem. Evaluation on the widely-used MIMIC III dataset demonstrates the efficacy of GLLA in enhancing diagnostic prediction performance. The contributions of this research include a comprehensive analysis of existing models, the identification of limitations, and the development of innovative approaches to improve the accuracy and effectiveness of EHR analysis. Ultimately, GLLA aims to advance healthcare decision-making, disease management strategies, and patient outcomes.

A Quantitative Bias Analysis Approach to Informative Presence Bias in Electronic Health Records.

Accurate outcome and exposure ascertainment in electronic health record (EHR) data, referred to as EHR phenotyping, relies on the completeness and accuracy of EHR data for each individual. However, some individuals, such as those with a greater comorbidity burden, visit the health care system more frequently and thus have more complete data, compared with others. Ignoring such dependence of exposure and outcome misclassification on visit frequency can bias estimates of associations in EHR analysis. We developed a framework for describing the structure of outcome and exposure misclassification due to informative visit processes in EHR data and assessed the utility of a quantitative bias analysis approach to adjusting for bias induced by informative visit patterns. Using simulations, we found that this method produced unbiased estimates across all informative visit structures, if the phenotype sensitivity and specificity were correctly specified. We applied this method in an example where the association between diabetes and progression-free survival in metastatic breast cancer patients may be subject to informative presence bias. The quantitative bias analysis approach allowed us to evaluate robustness of results to informative presence bias and indicated that findings were unlikely to change across a range of plausible values for phenotype sensitivity and specificity. Researchers using EHR data should carefully consider the informative visit structure reflected in their data and use appropriate approaches such as the quantitative bias analysis approach described here to evaluate robustness of study findings.

Development of an Automatic Rule-Based Algorithm for the Detection of Ovarian Cancer Recurrence From Electronic Health Records.

As the onset of cancer recurrence is not explicitly recorded in the electronic health record (EHR), a high volume of manual chart review is required to detect the cancer recurrence. This study aims to develop an automatic rule-based algorithm for detecting ovarian cancer (OC) recurrence on the basis of minimally preprocessed EHR data. The automatic rule-based recurrence detection algorithm (Auto-Recur), using notes on image reading (positron emission tomography-computed tomography [PET-CT], CT, magnetic resonance imaging [MRI]), biomarker (CA125), and treatment information (surgery, chemotherapy, radiotherapy), was developed to detect the first OC recurrence. Auto-Recur contains three single algorithms (images, biomarkers, treatments) and hybrid algorithms (combinations of the single algorithms). The performance of Auto-Recur was assessed using sensitivity, specificity, and accuracy of the recurrence time detected. The recurrence-free survival probabilities were estimated and compared with the retrospective chart review results. The proposed Auto-Recur considerably reduced human resources and time; it saved approximately 1,340 days when scaled to 100,000 patients compared with the conventional retrospective chart review. The hybrid algorithm on the basis of a combination of image, biomarker, and treatment information was the most efficient (sensitivity: 93.4%, specificity: 97.4%) and precisely captured recurrence time (average time error: 8.5 days). The estimated 3-year recurrence-free survival probability (44%) was close to the estimates by the retrospective chart review (45%, log-rank P value = .894). Our rule-based algorithm effectively captured the first OC recurrence from large-scale EHR while closely approximating the recurrence-free survival estimates obtained by conventional retrospective chart reviews. The study findings facilitate large-scale EHR analysis, enhancing clinical research opportunities.

Applying Advanced Practice Nurses in Rural Japan: A Mixed-Methods Perspective.

Introduction Rural hospitals, particularly those in geographically isolated regions like Shimane Prefecture, Japan, face significant healthcare delivery challenges. These include limited resources, an aging population, and a scarcity of healthcare professionals. Advanced practice nurses (APNs) have emerged as pivotal in addressing these gaps, offering specialized patient assessment, diagnosis, and management skills. This study aimed to evaluate the demand for APNs in rural community hospitals, focusing on the specific educational needs and clinical competencies required to improve healthcare outcomes in these settings. Method Employing a mixed-methods approach, this research combined qualitative insights from stakeholder interviews with quantitative data analysis of electronic health records (EHRs) at Unnan City Hospital. This sequential exploratory design aimed to capture comprehensive educational needs and outcomes, integrating the depth of qualitative data with the breadth of quantitative evidence to tailor a curriculum for APNs in rural healthcare contexts. Results The study revealed a critical demand for APNs skilled in managing common medical issues in rural settings, such as infections, circulatory failures, and respiratory problems. Stakeholder interviews highlighted the necessity for a curriculum that enhances clinical competencies and emphasizes soft skills like communication and leadership. An analysis of EHRs identified 21 specific diseases across six categories, underlining the importance of targeted education on these prevalent conditions. Conclusion The findings underscore the urgent need for specialized education programs for APNs in rural hospitals to address these communities' unique healthcare challenges. Developing a curriculum that focuses on clinical and soft skills essential for rural healthcare delivery can significantly enhance the quality of care. This study advocates for implementing such tailored educational programs to empower APNs, thereby contributing to healthcare equity and improving patient outcomes in rural settings.

Clinical risk prediction using language models: benefits and considerations.

The use of electronic health records (EHRs) for clinical risk prediction is on the rise. However, in many practical settings, the limited availability of task-specific EHR data can restrict the application of standard machine learning pipelines. In this study, we investigate the potential of leveraging language models (LMs) as a means toincorporate supplementary domain knowledge for improving the performance of various EHR-based risk prediction tasks. We propose two novel LM-based methods, namely "LLaMA2-EHR" and "Sent-e-Med." Our focus is on utilizing the textual descriptions within structured EHRs to make risk predictions about future diagnoses. We conduct a comprehensive comparison with previous approaches across various data types and sizes. Experiments across 6 different methods and 3 separate risk prediction tasks reveal that employing LMs to represent structured EHRs, such as diagnostic histories, results in significant performance improvements when evaluated using standard metrics such as area under the receiver operating characteristic (ROC) curve and precision-recall (PR) curve. Additionally, they offer benefits such as few-shot learning, the ability to handle previously unseen medical concepts, and adaptability to various medical vocabularies. However, it is noteworthy that outcomes may exhibit sensitivity to a specific prompt. LMs encompass extensive embedded knowledge, making them valuable for the analysis of EHRs in the context of risk prediction. Nevertheless, it is important to exercise caution in their application, as ongoing safety concerns related to LMs persist and require continuous consideration.

Value of Electronic Health Records Measured Using Financial and Clinical Outcomes: Quantitative Study.

The Health Information Technology for Economic and Clinical Health Act of 2009 was legislated to reduce health care costs, improve quality, and increase patient safety. Providers and organizations were incentivized to exhibit meaningful use of certified electronic health record (EHR) systems in order to achieve this objective. EHR adoption is an expensive investment, given the resources and capital that are invested. Due to the cost of the investment, a return on the EHR adoption investment is expected. This study performed a value analysis of EHRs. The objective of this study was to investigate the relationship between EHR adoption levels and financial and clinical outcomes by combining both financial and clinical outcomes into one conceptual model. We examined the multivariate relationships between different levels of EHR adoption and financial and clinical outcomes, along with the time variant control variables, using moderation analysis with a longitudinal fixed effects model. Since it is unknown as to when hospitals begin experiencing improvements in financial outcomes, additional analysis was conducted using a 1- or 2-year lag for profit margin ratios. A total of 5768 hospital-year observations were analyzed over the course of 4 years. According to the results of the moderation analysis, as the readmission rate increases by 1 unit, the effect of a 1-unit increase in EHR adoption level on the operating margin decreases by 5.38%. Hospitals with higher readmission payment adjustment factors have lower penalties. This study fills the gap in the literature by evaluating individual relationships between EHR adoption levels and financial and clinical outcomes, in addition to evaluating the relationship between EHR adoption level and financial outcomes, with clinical outcomes as moderators. This study provided statistically significant evidence (P<.05), indicating that there is a relationship between EHR adoption level and operating margins when this relationship is moderated by readmission rates, meaning hospitals that have adopted EHRs could see a reduction in their readmission rates and an increase in operating margins. This finding could further be supported by evaluating more recent data to analyze whether hospitals increasing their level of EHR adoption would decrease readmission rates, resulting in an increase in operating margins. Hospitals would incur lower penalties as a result of improved readmission rates, which would contribute toward improved operating margins.

Lung cancer registry to promote person-centered care.&amp;nbsp;

Introduction: Lung cancer is a leading cause of death and major burden on healthcare systems. Patient-centered outcome measurements allow the optimization of care organization and quality improvement. The aim of this project was to outline a set of variables and outcomes to evaluate the quality of care provided to lung cancer patients. &#x0D; Methods: This project took place at Hospital Italiano de Buenos Aires, a community-based tertiary care hospital in Argentina; which provides thoracic surgery, pneumology, oncology, pathology, radiology and radiotherapy services. It also has a Lung Cancer Multidisciplinary Board. All patient information is stored in a Data Repository fed by the electronic health record (EHR).&#x0D; We established a team of experts in management, IT, thoracic surgery, pneumology, oncology, and internal medicine. We reviewed care processes and patient trajectory along the usual pathway, through a retrospective EHR analysis of a consecutive sample of lung cancer patients. &#x0D; We performed a literature review to outline possible variables and health outcomes to be collected, instruments, and frequency of measurement. We determined needs, usefulness, and feasibility of collecting variables and outcomes in group meetings. &#x0D; Results: We reviewed 15 cases of lung cancer, 53% male, median age 68 years old (range: 29-85). We detected heterogeneity in the time elapsed between critical points of management (eg first visit to diagnosis 21 to 642 days; biopsy report to chemotherapy 19 to 163). The proportion of use of the healthcare system was also variable (eg: 87% consulted pneumonolory or oncology, only 13% had pain management visits and 7% received respiratory kinesic therapy). Patient-reported outcomes were not systematically evaluated.&#x0D; The selected baseline variables were: age, sex, weight loss, smoking status, comorbidities, patient-reported health status (EORTC European Organization for Research and treatment of cancer, QLC-LC30 Core quality of life questionnaire, QLC-LC13 Lung cancer-specific quality of life questionnaire), performance status (ECOG Eastern Cooperative Oncology Group), pulmonary function (VEF1), histology, clinical and pathological stage (AJCC-TNM 8th Ed), molecular tests and treatment intent. The outcomes (collected annually): clinical staging performed before treatment, time from diagnosis to treatment, performance status, patient-reported health status, major surgical complications, major chemotherapy/radiotherapy complications, time from diagnosis to treatment, completed treatment, overall survival, cause of death, treatment-related mortality, place of death, time spent in hospital at the end of life (last 30 days).&#x0D; Discussion: Cancer registries comprise a set of systematically and prospectively collected data. Quality measurements performed in the context of cancer registries could assist in monitoring and evaluating quality of care. &#x0D; Conclusions: The project determined essential variables and outcomes to be measured, considering their relevance and feasibility.&#x0D; Lessons learnt: Teamwork is essential when developing a registry, as different stakeholders' opinions are reflected. Integrated care is a useful solution to overcome fragmentation.&#x0D; Limitations: Patients did not participate directly in this development. &#x0D; Suggestions for future research: These results will assist in outlining a care pathway for patients with lung cancer, which is currently under development. Evaluation before and after the implementation of care pathways may evidence quality improvement.

Longitudinal Study of Comorbidities and Clinical Outcomes in Persons with Gallstone Disease Using Electronic Health Records

BackgroundGallstone disease (GSD) is common and leads to significant morbidity, mortality, and health care utilization in the USA. We examined comorbidities and clinical outcomes among persons with GSD using electronic health records (EHR). MethodsIn this retrospective study of 1,381,004 adults, GSD was defined by ICD-9 code 574 or ICD-10 code K80 using Optum® longitudinal EHR from January 2007 to March 2021. We obtained diagnosis, procedure, prescription, and vital sign records and evaluated associations between demographics, comorbidities, and medications with cholecystectomy, digestive cancers, and mortality. ResultsAmong persons with GSD, 30% had a cholecystectomy and were more likely to be women, White, and younger, and less likely to have comorbidities, except for obesity, gastroesophageal reflux disease (GERD), abdominal pain, hyperlipidemia, and pancreatitis. Among persons with GSD, 2.2% had a non-colorectal digestive cancer diagnosis during follow-up and risk was 40% lower among persons with a cholecystectomy. Non-colorectal digestive cancer predictors included older age, male sex, non-White race-ethnicity, lower BMI, other cancers, diabetes, chronic liver disease, pancreatitis, GERD, and abdominal pain. Among persons with GSD, mortality was 15.1% compared with 9.7% for the whole EHR sample. Persons with a cholecystectomy had 40% lower mortality risk and mortality predictors included older age, male sex, Black race, lower BMI, and most comorbidities. ConclusionsIn this EHR analysis of persons with GSD, 30% had a cholecystectomy. Mortality was higher compared with the whole EHR sample. Persons with cholecystectomy were less likely to have non-colorectal digestive cancer or to die.

Inequalities in Communicating about Nutrition Information during Prenatal Care: The After-Visit Summary

ABSTRACT Background Though prenatal nutrition information is critical, it is not known whether information is shared equitably by patient race, financial security, or English proficiency. Purpose To evaluate whether delivery or receipt of ChooseYourFish.org nutrition information in the first prenatal visit differed by patient demographics. Methods Analysis of clinician-document electronic health record (EHR) or patient-reported surveys compared delivery and receipt of fish-related nutrition information in the first prenatal visit. Inferential statistics were used to compare delivery or receipt and race, ethnicity, payor, or interpreter use. Results EHR analysis (n = 2,329) revealed Medicaid patients who used an interpreter were half as likely to have the fish nutrition message in their after-visit summary compared to those with Medicaid who did not use an interpreter (OR = 0.54, 95% CL: 0.35–0.84). The same was not true for non-Medicaid patients. Survey analysis (n = 52) showed respondents identifying as Black or African American were 25% less likely to report receiving the after-visit summary compared to respondents who identified as white (p < .01). Discussion The results presented here illustrate how nutrition communication in the prenatal period can differ by patient race, financial security, and language. Translation to Health Education Practice: Culturally humble efforts to understand drivers of healthcare communication are needed to eliminate inequalities. A AJHE Self-Study quiz is online for this article via the SHAPE America Online Institute (SAOI) http://portal.shapeamerica.org/trn-Webinars

Using Natural Language Processing and Machine Learning to Identify Opioids in Electronic Health Record Data.

This study evaluates the utility of machine learning (ML) and natural language processing (NLP) in the processing and initial analysis of data within the electronic health record (EHR). We present and evaluate a method to classify medication names as either opioids or non-opioids using ML and NLP. A total of 4216 distinct medication entries were obtained from the EHR and were initially labeled by human reviewers as opioid or non-opioid medications. An approach incorporating bag-of-words NLP and supervised ML classification was implemented in MATLAB and used to automatically classify medications. The automated method was trained on 60% of the input data, evaluated on the remaining 40%, and compared to manual classification results. A total of 3991 medication strings were classified as non-opioid medications (94.7%), and 225 were classified as opioid medications by the human reviewers (5.3%). The algorithm achieved a 99.6% accuracy, 97.8% sensitivity, 94.6% positive predictive value, F1 value of 0.96, and a receiver operating characteristic (ROC) curve with 0.998 area under the curve (AUC). A secondary analysis indicated that approximately 15-20 opioids (and 80-100 non-opioids) were needed to achieve accuracy, sensitivity, and AUC values of above 90-95%. The automated approach achieved excellent performance in classifying opioids or non-opioids, even with a practical number of human reviewed training examples. This will allow a significant reduction in manual chart review and improve data structuring for retrospective analyses in pain studies. The approach may also be adapted to further analysis and predictive analytics of EHR and other "big data" studies.

A Deep-Learning Neural Network Approach for Secure Wireless Communication in the Surveillance of Electronic Health Records

The electronic health record (EHR) surveillance process relies on wireless security administered in application technology, such as the Internet of Things (IoT). Automated supervision with cutting-edge data analysis methods may be a viable strategy to enhance treatment in light of the increasing accessibility of medical narratives in the electronic health record. EHR analysis structured data structure code was used to obtain data on initial fatality risk, infection rate, and hazard ratio of death from EHRs for prediction of unexpected deaths. Patients utilizing EHRs in general must keep in mind the significance of security. With the rise of the IoT and sensor-based Healthcare 4.0, cyber-resilience has emerged as a need for the safekeeping of patient information across all connected devices. Security for access, amendment, and storage is cumulatively managed using the common paradigm. For improving the security of surveillance in the aforementioned services, this article introduces an endorsed joint security scheme (EJSS). This scheme recognizes the EHR utilization based on the aforementioned processes. For each process, different security measures are administered for sustainable security. Access control and storage modification require relative security administered using mutual key sharing between the accessing user and the EHR database. In this process, the learning identifies the variations in different processes for reducing adversarial interruption. The federated learning paradigm employed in this scheme identifies concurrent adversaries in the different processes initiated at the same time. Differentiating the adversaries under each process strengthens mutual authentication using individual attributes. Therefore, individual surveillance efficiency through log inspection and adversary detection is improved for heterogeneous and large-scale EHR databases.

Repurposing ketamine to treat cocaine use disorder: integration of artificial intelligence-based prediction, expert evaluation, clinical corroboration and mechanism of action analyses.

Cocaine use disorder (CUD) is a significant public health issue for which there is no Food and Drug Administration (FDA) approved medication. Drug repurposing looks for new cost-effective uses of approved drugs. This study presents an integrated strategy to identify repurposed FDA-approved drugs for CUD treatment. Our drug repurposing strategy combines artificial intelligence (AI)-based drug prediction, expert panel review, clinical corroboration and mechanisms of action analysis being implemented in the National Drug Abuse Treatment Clinical Trials Network (CTN). Based on AI-based prediction and expert knowledge, ketamine was ranked as the top candidate for clinical corroboration via electronic health record (EHR) evaluation of CUD patient cohorts prescribed ketamine for anesthesia or depression compared with matched controls who received non-ketamine anesthesia or antidepressants/midazolam. Genetic and pathway enrichment analyses were performed to understand ketamine's potential mechanisms of action in the context of CUD. The study utilized TriNetX to access EHRs from more than 90 million patients world-wide. Genetic- and functional-level analyses used DisGeNet, Search Tool for Interactions of Chemicals and Kyoto Encyclopedia of Genes and Genomes databases. A total of 7742 CUD patients who received anesthesia (3871 ketamine-exposed and 3871 anesthetic-controlled) and 7910 CUD patients with depression (3955 ketamine-exposed and 3955 antidepressant-controlled) were identified after propensity score-matching. EHR analysis outcome was a CUD remission diagnosis within 1 year of drug prescription. Patients with CUD prescribed ketamine for anesthesia displayed a significantly higher rate of CUD remission compared with matched individuals prescribed other anesthetics [hazard ratio (HR) = 1.98, 95% confidence interval (CI) = 1.42-2.78]. Similarly, CUD patients prescribed ketamine for depression evidenced a significantly higher CUD remission ratio compared with matched patients prescribed antidepressants or midazolam (HR = 4.39, 95% CI = 2.89-6.68). The mechanism of action analysis revealed that ketamine directly targets multiple CUD-associated genes (BDNF, CNR1, DRD2, GABRA2, GABRB3, GAD1, OPRK1, OPRM1, SLC6A3, SLC6A4) and pathways implicated in neuroactive ligand-receptor interaction, cAMP signaling and cocaine abuse/dependence. Ketamine appears to be a potential repurposed drug for treatment of cocaine use disorder.

Documenting maternal and childcare information of mothers presented to substance use treatment services: A qualitative study of reports in a clinical case register

BackgroundMothers compose a significant proportion of women in substance use treatment services. These women have needs that, if not addressed, can negatively impact their capacity to parent. This study explores the feasibility of using free-text notes from electronic health records (EHRs) to identify factors that impair mothers' ability to care for their children. MethodThis study is a qualitative analysis of EHRs of 50 women attending substance use services in Southeast London who were parents of dependent children (defined as mothers of children aged <18 years independent of living together). We developed a sampling stratification process to ensure an adequate volume of data were available and analyzed per case. The study identified and tested search terms. We extracted data from clinical notes and letters of communication with other services/agencies (free-text notes) using the identified search terms and conducted deductive thematic analysis. ResultsThe mean number of documents per case was 92.17 (SD = 18.51). The study identified five themes with subthemes: childcare arrangements, family context, safeguarding issues, factors that might impact the treatment plan and care of the child, and communication between the health care and child welfare systems. ConclusionThe study demonstrates a novel approach for exploring parenting-related characteristics of mothers in substance use treatment. Despite a range of maternal and childcare-related information available on EHRs, the type of treatment and support being offered to patients in response to the reported information is less well documented. Findings highlight the need for further investments in implementing effective family centered strategies within substance use services.

Examining Real-World Adherence to Nusinersen for the Treatment of Spinal Muscular Atrophy Using Two Large US Data Sources.

Spinal muscular atrophy (SMA) is a rare neuromuscular disease characterized by progressive muscular atrophy and weakness. Nusinersen was the first treatment approved for SMA. Per the US label, the nusinersen administration schedule consists of three loading doses at 14-day intervals, a fourth loading dose 30days later, and maintenance doses every 4months thereafter. Using two large US databases, we evaluated real-world adherence to nusinersen with its unique dosing schedule among generalizable populations of patients with SMA. Patients with SMA treated with nusinersen, likely to have complete information on date of treatment initiation, were identified in the Optum® de-identified electronic health records (EHR) database (7/2017-9/2019), and in the Merative™ MarketScan® Research Databases from commercial (1/2017-6/2020) and Medicaid claims (1/2017-12/2019). Baseline demographics, number of nusinersen administrations on time, and distribution of inter-dose intervals were summarized. Totals of 67 and 291 patients were identified in the EHR and claims databases, respectively. Most nusinersen doses were received on time (93.9% EHR, 80.5% claims). Adherence was higher during the maintenance phase (90.6%) than the loading phase (71.1%) in the claims analysis, in contrast with the EHR analysis (95.5% and 92.6%, respectively), suggesting that not all loading doses of nusinersen may be accurately captured in claims. Inter-dose intervals captured in both databases aligned with the expected dosing schedule. Most nusinersen doses were received on time, consistent with the recommended schedule. Our findings also highlight the importance of careful methodological approaches when using real-world administrative databases for evaluation of nusinersen treatment patterns.

Patient journey of individuals tested for HCV in Spain: LiverTAI, a retrospective analysis of EHRs through natural language processing

ObjectivesLimited screening and delays in diagnosis and linkage-to-care are barriers for hepatitis C virus (HCV) elimination. The LiverTAI study focused on patients tested for HCV using AI technologies to describe their demographic and clinical characteristics and pre-testing patient journeys, reflecting clinical practice in hospitals. Patients and methodsLiverTAI is a retrospective, secondary analysis of electronic health records (EHRs) from 6 tertiary Spanish hospitals, extracting unstructured clinical data using natural language processing (NLP) EHRead® technology. Adult subjects with an HCV testing procedure from January 2014 to December 2018 were grouped according to HCV seropositivity and viremia. ResultsFrom 2,440,358 patients, 16,261 patients were tested for HCV (13,602 [83.6%] HCV seronegative; 2659 [16.4%] seropositive). Active HCV viremia appeared in 37.7% (n=1003) of patients, 18.6% (n=494) had negative viremia, and 43.7% (n=1162) unknown viremia. Patient journeys showed core departments (Gastroenterology, Internal Medicine, and Infectious Disease) and others including Emergency perform ample HCV testing in Spanish hospitals, whereas Medical Oncology lags. Patients were PCR-tested and genotyped significantly faster in core departments (p<.001). ConclusionsOur results highlight hospital departments responsible for HCV testing. However, further testing was sub-optimal during the study period. Therefore, we underscore the need for HCV screening and reflex testing to accelerate diagnosis and linkage-to-care.

Multiview Incomplete Knowledge Graph Integration with application to cross-institutional EHR data harmonization

Objective:The growing availability of electronic health records (EHR) data opens opportunities for integrative analysis of multi-institutional EHR to produce generalizable knowledge. A key barrier to such integrative analyses is the lack of semantic interoperability across different institutions due to coding differences. We propose a Multiview Incomplete Knowledge Graph Integration (MIKGI) algorithm to integrate information from multiple sources with partially overlapping EHR concept codes to enable translations between healthcare systems. Methods:The MIKGI algorithm combines knowledge graph information from (i) embeddings trained from the co-occurrence patterns of medical codes within each EHR system and (ii) semantic embeddings of the textual strings of all medical codes obtained from the Self-Aligning Pretrained BERT (SAPBERT) algorithm. Due to the heterogeneity in the coding across healthcare systems, each EHR source provides partial coverage of the available codes. MIKGI synthesizes the incomplete knowledge graphs derived from these multi-source embeddings by minimizing a spherical loss function that combines the pairwise directional similarities of embeddings computed from all available sources. MIKGI outputs harmonized semantic embedding vectors for all EHR codes, which improves the quality of the embeddings and enables direct assessment of both similarity and relatedness between any pair of codes from multiple healthcare systems. Results:With EHR co-occurrence data from Veteran Affairs (VA) healthcare and Mass General Brigham (MGB), MIKGI algorithm produces high quality embeddings for a variety of downstream tasks including detecting known similar or related entity pairs and mapping VA local codes to the relevant EHR codes used at MGB. Based on the cosine similarity of the MIKGI trained embeddings, the AUC was 0.918 for detecting similar entity pairs and 0.809 for detecting related pairs. For cross-institutional medical code mapping, the top 1 and top 5 accuracy were 91.0% and 97.5% when mapping medication codes at VA to RxNorm medication codes at MGB; 59.1% and 75.8% when mapping VA local laboratory codes to LOINC hierarchy. When trained with 500 labels, the lab code mapping attained top 1 and 5 accuracy at 77.7% and 87.9%. MIKGI also attained best performance in selecting VA local lab codes for desired laboratory tests and COVID-19 related features for COVID EHR studies. Compared to existing methods, MIKGI attained the most robust performance with accuracy the highest or near the highest across all tasks. Conclusions:The proposed MIKGI algorithm can effectively integrate incomplete summary data from biomedical text and EHR data to generate harmonized embeddings for EHR codes for knowledge graph modeling and cross-institutional translation of EHR codes.

Sleep apnea phenotyping and relationship to disease in a large clinical biobank.

ObjectiveSleep apnea is associated with a broad range of pathophysiology. While electronic health record (EHR) information has the potential for revealing relationships between sleep apnea and associated risk factors and outcomes, practical challenges hinder its use. Our objectives were to develop a sleep apnea phenotyping algorithm that improves the precision of EHR case/control information using natural language processing (NLP); identify novel associations between sleep apnea and comorbidities in a large clinical biobank; and investigate the relationship between polysomnography statistics and comorbid disease using NLP phenotyping.Materials and MethodsWe performed clinical chart reviews on 300 participants putatively diagnosed with sleep apnea and applied International Classification of Sleep Disorders criteria to classify true cases and noncases. We evaluated 2 NLP and diagnosis code-only methods for their abilities to maximize phenotyping precision. The lead algorithm was used to identify incident and cross-sectional associations between sleep apnea and common comorbidities using 4876 NLP-defined sleep apnea cases and 3× matched controls.ResultsThe optimal NLP phenotyping strategy had improved model precision (≥0.943) compared to the use of one diagnosis code (≤0.733). Of the tested diseases, 170 disorders had significant incidence odds ratios (ORs) between cases and controls, 8 of which were confirmed using polysomnography (n = 4544), and 281 disorders had significant prevalence OR between sleep apnea cases versus controls, 41 of which were confirmed using polysomnography data.Discussion and ConclusionAn NLP-informed algorithm can improve the accuracy of case-control sleep apnea ascertainment and thus improve the performance of phenome-wide, genetic, and other EHR analyses of a highly prevalent disorder.

Interpretability of time-series deep learning models: A study in cardiovascular patients admitted to Intensive care unit

Interpretability is fundamental in healthcare problems and the lack of it in deep learning models is currently the major barrier in the usage of such powerful algorithms in the field. The study describes the implementation of an attention layer for Long Short-Term Memory (LSTM) neural network that provides a useful picture on the influence of the several input variables included in the model.A cohort of 10,616 patients with cardiovascular diseases is selected from the MIMIC III dataset, an openly available database of electronic health records (EHRs) including all patients admitted to an ICU at Boston’s Medical Centre. For each patient, we consider a 10-length sequence of 1-hour windows in which 48 clinical parameters are extracted to predict the occurrence of death in the next 7 days. Inspired from the recent developments in the field of attention mechanisms for sequential data, we implement a recurrent neural network with LSTM cells incorporating an attention mechanism to identify features driving model’s decisions over time.The performance of the LSTM model, measured in terms of AUC, is 0.790 (SD = 0.015). Regard our primary objective, i.e. model interpretability, we investigate the role of attention weights. We find good correspondence with driving predictors of a transparent model (r = 0.611, 95% CI [0.395, 0.763]). Moreover, most influential features identified at the cohort-level emerge as known risk factors in the clinical context.Despite the limitations of study dataset, this work brings further evidence of the potential of attention mechanisms in making deep learning model more interpretable and suggests the application of this strategy for the sequential analysis of EHRs.

High performance implementation of the hierarchical likelihood for generalized linear mixed models: an application to estimate the potassium reference range in massive electronic health records datasets

BackgroundConverting electronic health record (EHR) entries to useful clinical inferences requires one to address the poor scalability of existing implementations of Generalized Linear Mixed Models (GLMM) for repeated measures. The major computational bottleneck concerns the numerical evaluation of multivariable integrals, which even for the simplest EHR analyses may involve millions of dimensions (one for each patient). The hierarchical likelihood (h-lik) approach to GLMMs is a methodologically rigorous framework for the estimation of GLMMs that is based on the Laplace Approximation (LA), which replaces integration with numerical optimization, and thus scales very well with dimensionality.MethodsWe present a high-performance, direct implementation of the h-lik for GLMMs in the R package TMB. Using this approach, we examined the relation of repeated serum potassium measurements and survival in the Cerner Real World Data (CRWD) EHR database. Analyzing this data requires the evaluation of an integral in over 3 million dimensions, putting this problem beyond the reach of conventional approaches. We also assessed the scalability and accuracy of LA in smaller samples of 1 and 10% size of the full dataset that were analyzed via the a) original, interconnected Generalized Linear Models (iGLM), approach to h-lik, b) Adaptive Gaussian Hermite (AGH) and c) the gold standard for multivariate integration Markov Chain Monte Carlo (MCMC).ResultsRandom effects estimates generated by the LA were within 10% of the values obtained by the iGLMs, AGH and MCMC techniques. The H-lik approach was 4–30 times faster than AGH and nearly 800 times faster than MCMC. The major clinical inferences in this problem are the establishment of the non-linear relationship between the potassium level and the risk of mortality, as well as estimates of the individual and health care facility sources of variations for mortality risk in CRWD.ConclusionsWe found that the direct implementation of the h-lik offers a computationally efficient, numerically accurate approach for the analysis of extremely large, real world repeated measures data via the h-lik approach to GLMMs. The clinical inference from our analysis may guide choices of treatment thresholds for treating potassium disorders in the clinic.