Electronic Health Record Vendors Research Articles

Patient Portals Fail to Collect Structured Information About Who Else is Involved in a Person's Care.

The US health care delivery system does not systematically engage or support family or friend care partners. Meanwhile, the uptake and familiarity of portals to personal health information are increasing among patients. Technology innovations, such as shared access to the portal, use separate identity credentials to differentiate between patients and care partners. Although not well-known, or commonly used, shared access allows patients to identify who they do and do not want to be involved in their care. However, the processes for patients to grant shared access to portals are often limited or so onerous that interested patients and care partners often circumvent the process entirely. As a result, the vast majority of care partners resort to accessing portals using a patient's identity credentials-a "do-it-yourself" solution in conflict with a health systems' legal responsibility to protect patient privacy and autonomy. The personal narratives in this viewpoint (shared by permission) elaborate on quantitative studies and provide first-person snapshots of challenges faced by patients and families as they attempt to gain or grant shared access during crucial moments in their lives. As digital modalities increase patient roles in health care interactions, so does the importance of making shared access work for all stakeholders involved-patients, clinicians, and care partners. Electronic health record vendors must recognize that both patients and care partners are important users of their products, and health care organizations must acknowledge and support the critical contributions of care partners as distinct from patients.

Improved Accuracy/Completeness of EHR Race/Ethnicity Data: A Requisite Step to Address Disparities in Care.

Accurate and complete racial/ethnic data in the electronic health record are a requisite step to addressing disparities in neurologic care, and at local, regional, and national levels. The current data pertaining to the patients' race and ethnicity contained in the electronic health record are inadequate. This article outlines recommendations at the individual practice and electronic health record vendor level to improve documentation of race and ethnicity.

Evaluation of SNOMED CT Grouper Accuracy and Coverage in Organizing the Electronic Health Record Problem List by Clinical System: Observational Study.

The problem list (PL) is a repository of diagnoses for patients' medical conditions and health-related issues. Unfortunately, over time, our PLs have become overloaded with duplications, conflicting entries, and no-longer-valid diagnoses. The lack of a standardized structure for review adds to the challenges of clinical use. Previously, our default electronic health record (EHR) organized the PL primarily via alphabetization, with other options available, for example, organization by clinical systems or priority settings. The system's PL was built with limited groupers, resulting in many diagnoses that were inconsistent with the expected clinical systems or not associated with any clinical systems at all. As a consequence of these limited EHR configuration options, our PL organization has poorly supported clinical use over time, particularly as the number of diagnoses on the PL has increased. We aimed to measure the accuracy of sorting PL diagnoses into PL system groupers based on Systematized Nomenclature of Medicine Clinical Terms (SNOMED CT) concept groupers implemented in our EHR. We transformed and developed 21 system- or condition-based groupers, using 1211 SNOMED CT hierarchal concepts refined with Boolean logic, to reorganize the PL in our EHR. To evaluate the clinical utility of our new groupers, we extracted all diagnoses on the PLs from a convenience sample of 50 patients with 3 or more encounters in the previous year. To provide a spectrum of clinical diagnoses, we included patients from all ages and divided them by sex in a deidentified format. Two physicians independently determined whether each diagnosis was correctly attributed to the expected clinical system grouper. Discrepancies were discussed, and if no consensus was reached, they were adjudicated by a third physician. Descriptive statistics and Cohen κ statistics for interrater reliability were calculated. Our 50-patient sample had a total of 869 diagnoses (range 4-59; median 12, IQR 9-24). The reviewers initially agreed on 821 system attributions. Of the remaining 48 items, 16 required adjudication with the tie-breaking third physician. The calculated κ statistic was 0.7. The PL groupers appropriately associated diagnoses to the expected clinical system with a sensitivity of 97.6%, a specificity of 58.7%, a positive predictive value of 96.8%, and an F1-score of 0.972. We found that PL organization by clinical specialty or condition using SNOMED CT concept groupers accurately reflects clinical systems. Our system groupers were subsequently adopted by our vendor EHR in their foundation system for PL organization.

Foundation Models, Generative AI, and Large Language Models

We are in a booming era of artificial intelligence, particularly with the increased availability of technologies that can help generate content, such as ChatGPT. Healthcare institutions are discussing or have started utilizing these innovative technologies within their workflow. Major electronic health record vendors have begun to leverage large language models to process and analyze vast amounts of clinical natural language text, performing a wide range of tasks in healthcare settings to help alleviate clinicians' burden. Although such technologies can be helpful in applications such as patient education, drafting responses to patient questions and emails, medical record summarization, and medical research facilitation, there are concerns about the tools' readiness for use within the healthcare domain and acceptance by the current workforce. The goal of this article is to provide nurses with an understanding of the currently available foundation models and artificial intelligence tools, enabling them to evaluate the need for such tools and assess how they can impact current clinical practice. This will help nurses efficiently assess, implement, and evaluate these tools to ensure these technologies are ethically and effectively integrated into healthcare systems, while also rigorously monitoring their performance and impact on patient care.

Guiding Principles for Adolescent Web-Based Portal Access Policies: Interviews With Informatics Administrators.

Web-based patient portals are tools that could support adolescents in managing their health and developing autonomy. However, informatics administrators must navigate competing interests when developing portal access policies for adolescents and their parents. We aimed to assess the perspectives of informatics administrators on guiding principles for the development of web-based health care portal access policies in adolescent health care. We interviewed informatics administrators from US hospitals with ≥50 dedicated pediatric beds. We performed a thematic analysis of guiding principles for developing and implementing adolescent portal access policies. We interviewed 65 informatics leaders who represented 63 pediatric hospitals, 58 health care systems, 29 states, and 14,379 pediatric hospital beds. Participants described 9 guiding principles related to three overarching themes: (1) balancing confidentiality and other care needs, (2) balancing simplicity and granularity, and (3) collaborating and advocating. Participants described the central importance of prioritizing the health and safety of the adolescent while also complying with state and federal laws. However, there were differing beliefs about how to prioritize health and safety and what role parents should play in supporting the adolescent's health care. Participants also identified areas where clinicians and institutions can advocate for adolescents, especially with electronic health record vendors and legislators. Informatics administrators provided guiding principles for adolescent portal access policies that aimed to balance the competing needs of adolescent confidentiality and the usefulness of the portal. Portal access policies must prioritize the adolescent's health and safety while complying with state and federal laws. However, institutions must determine how to best enact these principles. Institutions and clinicians should strive for consensus on principles to strengthen advocacy efforts with institutional leadership, electronic health record vendors, and lawmakers.

Explaining Variability in Electronic Health Record Effort in Primary Care Ambulatory Encounters.

Electronic health record (EHR) user interface event logs are fast providing another perspective on the value and efficiency EHR technology brings to health care. Analysis of these detailed usage data has demonstrated their potential to identify EHR and clinical process design factors related to user efficiency, satisfaction, and burnout. This study aimed to analyze the event log data across 26 different health systems to determine the variability of use of a single vendor's EHR based on four event log metrics, at the individual, practice group, and health system levels. We obtained de-identified event log data recorded from June 1, 2018, to May 31, 2019, from 26 health systems' primary care physicians. We estimated the variability in total Active EHR Time, Documentation Time, Chart Review Time, and Ordering Time across health systems, practice groups, and individual physicians. In total, 5,444 physicians (Family Medicine: 3,042 and Internal Medicine: 2,422) provided care in a total of 2,285 different practices nested in 26 health systems. Health systems explain 1.29, 3.55, 3.45, and 3.30% of the total variability in Active Time, Documentation Time, Chart Review Time, and Ordering Time, respectively. Practice-level variability was estimated to be 7.96, 13.52, 8.39, and 5.57%, respectively, and individual physicians explained the largest proportion of the variability for those same outcomes 17.09, 27.49, 17.51, and 19.75%, respectively. The most variable physician EHR usage patterns occurs at the individual physician level and decreases as you move up to the practice and health system levels. This suggests that interventions to improve individual users' EHR usage efficiency may have the most potential impact compared with those directed at health system or practice levels.

Clinical Knowledge Model for the Prevention of Healthcare-Associated Venous Thromboembolism.

Knowledge models inform organizational behavior through the logical association of documentation processes, definitions, data elements, and value sets. The development of a well-designed knowledge model allows for the reuse of electronic health record data to promote efficiency in practice, data interoperability, and the extensibility of data to new capabilities or functionality such as clinical decision support, quality improvement, and research. The purpose of this article is to describe the development and validation of a knowledge model for healthcare-associated venous thromboembolism prevention. The team used FloMap, an Internet-based survey resource, to compare metadata from six healthcare organizations to an initial draft model. The team used consensus decision-making over time to compare survey results. The resulting model included seven panels, 41 questions, and 231 values. A second validation step included completion of an Internet-based survey with 26 staff nurse respondents representing 15 healthcare organizations, two electronic health record vendors, and one academic institution. The final knowledge model contained nine Logical Observation Identifiers Names and Codes panels, 32 concepts, and 195 values representing an additional six panels (groupings), 15 concepts (questions), and the specification of 195 values (answers). The final model is useful for consistent documentation to demonstrate the contribution of nursing practice to the prevention of venous thromboembolism.

Standardizing default electronic health record tools to improve safety for hospitalized patients with Parkinson's disease.

Electronic Health Record (EHR) systems are often configured to address challenges and improve patient safety for persons with Parkinson's disease (PWP). For example, EHR systems can help identify Parkinson's disease (PD) patients across the hospital by flagging a patient's diagnosis in their chart, preventing errors in medication and dosing through the use of clinical decision support, and supplementing staff education through care plans that provide step-by-step road maps for disease-based care of a specific patient population. However, most EHR-based solutions are locally developed and, thus, difficult to scale widely or apply uniformly across hospital systems. In 2020, the Parkinson's Foundation, a national and international leader in PD research, education, and advocacy, and Epic, a leading EHR vendor with more than 35% market share in the United States, launched a partnership to reduce risks to hospitalized PWP using standardized EHR-based solutions. This article discusses that project which included leadership from physician informaticists, movement disorders specialists, hospital quality officers, the Parkinson's Foundation and members of the Parkinson's community. We describe the best practice solutions developed through this project. We highlight those that are currently available as standard defaults or options within the Epic EHR, discuss the successes and limitations of these solutions, and consider opportunities for scalability in environments beyond a single EHR vendor. The Parkinson's Foundation and Epic launched a partnership to develop best practice solutions in the Epic EHR system to improve safety for PWP in the hospital. The goal of the partnership was to create the EHR tools that will have the greatest impact on outcomes for hospitalized PWP.

Provider Perspectives on Adolescent Confidentiality and the Electronic Health Record Postimplementation of the 21st Century Cures Act Final Rule

PurposeThe 21st Century Cures Act Final Rule mandates increased transparency and accessibility of the Electronic Health Record (EHR). This mandate has exacerbated tensions in adolescent healthcare between transparency and confidentiality. This study evaluates clinicians' perspectives on how well the EHR maintains confidentiality for adolescents in the post-Cures Act era. MethodsA web-based survey was distributed via the national listserv of the Society for Adolescent Health and Medicine in March–April 2022. English-speaking members practicing in the Unites States who used an EHR to document clinical visits with minors were eligible. The survey included questions about EHR training, features, and approach to information sharing with adolescent patients and proxies. ResultsParticipants included 97 respondents from 32 states. Most participants were physicians (98%) and identified themselves as attending (89%). Several EHR vendors were represented, although the majority used Epic (76%). As has been seen in prior studies, there remains significant variability in approaches to patient portal access for adolescent patients and their adult proxies. Respondents report that training around adolescent-specific privacy issues remains infrequent and is perceived as inadequate. Adolescent providers, despite feeling confident in navigating the EHR generally, continue to report low rates of confidence in how well their EHR may protect adolescent privacy. DiscussionClinicians have persistent concerns about adolescent confidentiality postimplementation of the 21st Century Cures Act. Sharing sufficient information while protecting adolescent confidentiality might require standardization by EHR vendors to improve granularity of proxy information sharing. Healthcare institutions must also commit to training providers on management of adolescent confidentiality.

Analysis of Inclusive Gender, Sexuality and Sexual Orientation Data Elements in Academic Electronic Health Records.

Findings from an analysis of three vended academic electronic health records used in health science education are presented in this article. The quality assurance project examined the lexical and semantic fit and content coverage of gender, sexuality, and sexual orientation data elements within the academic electronic health records. A semantic comparative content analysis using a cognitive walkthrough was conducted as a means of comparing the ideal set of gender, sexuality, and sexual orientation data elements with those found in the three vended academic electronic health records. The results indicated a need for alignment to the research literature, expert consensus, and technical standards similar to what is expected for electronic health records used in clinical practice because of a lack of ideal state data elements. The findings align with ongoing issues with bias and disparities seen in the care of the lesbian, gay, bisexual, and transgender population and a lack of diverse, inclusive media and teaching technologies in health science education. The quality project and findings can inform academic electronic health record vendors on how they can create more inclusive systems and bring awareness to healthcare educators about the potential for implicit and explicit bias in their teaching technologies.

Enhancing Simulation: A Roadmap for Integrating the Electronic Health Record into the Nurse Practitioner Curriculum

The electronic health record (EHR) is an important health information technology (IT) used to support efficient clinical workflows, patient safety, and quality of care. The EHR is an individual's medical chart stored in digital format for complete real-time information to reduce medical errors and improve outcomes. EHR technology is complex, and users may struggle with efficiently navigating the system for data retrieval, documentation, and time management. Academic institutions are positioned to collaborate with EHR vendors to offer training playgrounds in the curriculum that simulate the clinical interface and support readiness to practice in relevant technology-rich clinical environments. A strategically selected educational EHR vendor matched to the platforms used by major health care organizations may influence clinical partnerships and employment opportunities. EHR competency may alleviate some of the onboarding burden for health care institutions and new graduates at a time of role transition. This article provides strategies for leveraging health IT in the academic curriculum to prepare a technology-proficient workforce capable of meeting the needs of diverse individuals, populations, and healthcare systems.

Clinical Decision Support to Reduce Opioid Prescriptions for Dental Extractions using SMART on FHIR: Implementation Report.

Clinical decision support (CDS) has the potential to improve clinical decision-making consistent with evidence-based care. CDS can be designed to save health care providers time and help them provide safe and personalized analgesic prescribing. The aim of this report is to describe the development of a CDS system designed to provide dentists with personalized pain management recommendations to reduce opioid prescribing following extractions. The use of CDS is also examined. This study was conducted in HealthPartners, which uses an electronic health record (EHR) system that integrates both medical and dental information upon which the CDS application was developed based on SMART (Substitutable Medical Applications and Reusable Technologies) on FHIR (Fast Healthcare Interoperability Resources). The various tools used to bring relevant medical conditions, medications, patient history, and other relevant data into the CDS interface are described. The CDS application runs a drug interaction algorithm developed by our organization and provides patient-specific recommendations. The CDS included access to the state Prescription Monitoring Program database. The pain management CDS was implemented as part of a study examining opioid prescribing among patients undergoing dental extraction procedures from February 17, 2020, to May 14, 2021. Provider-level use of CDS at extraction encounters ranged from 0% to 87.4% with 12.1% of providers opening the CDS for no encounters, 39.4% opening the CDS for 1%-20% of encounters, 36.4% opening it for 21%-50% of encounters, and 12.1% opening it for 51%-87% of encounters. The pain management CDS is an EHR-embedded, provider-facing tool to help dentists make personalized pain management recommendations following dental extractions. The SMART on FHIR-based pain management CDS adapted well to the point-of-care dental setting and led to the design of a scalable CDS tool that is EHR vendor agnostic. ClinicalTrials.gov NCT03584789; https://clinicaltrials.gov/study/NCT03584789.

Experts' Perspectives on Use of Fast Healthcare Interoperable Resources for Computerized Clinical Decision Support.

Barriers to improving the US healthcare system include a lack of interoperability across digital health information and delays in seeking preventative and recommended care. Interoperability can be seen as the lynch pin to reducing fragmentation and improving outcomes related to digital health systems. The prevailing standard for information exchange to enable interoperability is the Health Level Seven International Fast Healthcare Interoperable Resources standard. To better understand Fast Healthcare Interoperable Resources within the context of computerized clinical decision support expert interviews of health informaticists were conducted and used to create a modified force field analysis. Current barriers and future recommendations to scale adoption of Fast Healthcare Interoperable Resources were explored through qualitative analysis of expert interviews. Identified barriers included variation in electronic health record implementation, limited electronic health record vendor support, ontology variation, limited workforce knowledge, and testing limitations. Experts recommended research funders require Fast Healthcare Interoperable Resource usage, development of an "app store," incentives for clinical organizations and electronic health record vendors, and Fast Healthcare Interoperable Resource certification development.

Identifying and Addressing Barriers to Implementing Core Electronic Health Record Use Metrics for Ambulatory Care: Virtual Consensus Conference Proceedings.

Precise, reliable, valid metrics that are cost-effective and require reasonable implementation time and effort are needed to drive electronic health record (EHR) improvements and decrease EHR burden. Differences exist between research and vendor definitions of metrics. PROCESS: We convened three stakeholder groups (health system informatics leaders, EHR vendor representatives, and researchers) in a virtual workshop series to achieve consensus on barriers, solutions, and next steps to implementing the core EHR use metrics in ambulatory care. CONCLUSION: Actionable solutions identified to address core categories of EHR metric implementation challenges include: (1) maintaining broad stakeholder engagement, (2) reaching agreement on standardized measure definitions across vendors, (3) integrating clinician perspectives, and (4) addressing cognitive and EHR burden. Building upon the momentum of this workshop's outputs offers promise for overcoming barriers to implementing EHR use metrics.

Effectiveness of a Vendor Predictive Model for the Risk of Pediatric Asthma Exacerbation: A Difference-in-Differences Analysis.

Asthma is a common cause of morbidity and mortality in children. Predictive models may help providers tailor asthma therapies to an individual's exacerbation risk. The effectiveness of asthma risk scores on provider behavior and pediatric asthma outcomes remains unknown. Determine the impact of an electronic health record (EHR) vendor-released model on outcomes for children with asthma. The Epic Systems Risk of Pediatric Asthma Exacerbation model was implemented on February 24, 2021, for volunteer pediatric allergy and pulmonology providers as a noninterruptive risk score visible in the patient schedule view. Asthma hospitalizations, emergency department (ED) visits, or oral steroid courses within 90 days of the index visit were compared from February 24, 2019, to February 23, 2022, using a difference-in-differences design with a control group of visits to providers in the same departments. Volunteer providers were interviewed to identify barriers and facilitators to model use. In the intervention group, asthma hospitalizations within 90 days decreased from 1.4% (54/3,842) to 0.7% (14/2,165) after implementation with no significant change in the control group (0.9% [171/19,865] preimplementation to 1.0% [105/10,743] post). ED visits in the intervention group decreased from 5.8% (222/3,842) to 5.5% (118/2,164) but increased from 5.5% (1,099/19,865) to 6.8% (727/10,743) in the control group. The adjusted difference-in-differences estimators for hospitalization, ED visit, and oral steroid outcomes were -0.9% (95% confidence interval [CI]: -1.6 to -0.3), -2.4% (-3.9 to -0.8), and -1.9% (-4.3 to 0.5). In qualitative analysis, providers understood the purpose of the model and felt it was useful to flag high exacerbation risk. Trust in the model was calibrated against providers' own clinical judgement. This EHR vendor model implementation was associated with a significant decrease in asthma hospitalization and ED visits within 90 days of pediatric allergy and pulmonology clinic visits, but not oral steroid courses.

Variation in pediatric and adolescent electronic health data sharing practices under the 21st Century Cures Act.

To describe real-world practices and variation in implementation of the Information Blocking provisions amongst healthcare organizations caring for pediatric patients. An online survey regarding implementation practices was distributed to representatives from 10 participating US healthcare organizations located in 6 different states. The survey was followed by structured interviews conducted through video conference. Information was gathered about implementation practices at each organization, with a focus on patient and proxy portal access to, and segmentation capabilities of, certain data classes listed in the United States Core Data for Interoperability Version 1. All organizations had implemented the information blocking provisions at their institution. All organizations utilized different portal account types for proxies and users. All organizations reported the capability of sharing labs, medications, problem lists, imaging, and notes with the parent/guardian of the non-adolescent minor user with differences in how sensitive elements within the data classes were protected. Variability existed in how data was shared with the remaining user types. Significant variability exists in how organizations have implemented the information blocking rules. Variation in data sharing and data access between institutions can result in privacy breaches and create confusion about completeness of data for patients and families. Healthcare organizations have utilized varying strategies to comply with the information blocking provisions of the 21st Century Cures Act. Increased clarity from the Office of the National Coordinator for Health Information Technology on minor, adolescent, and caregiver privacy and improved segmentation capabilities from Electronic Health Record vendors is needed.

A Multiyear Survey Evaluating Clinician Electronic Health Record Satisfaction.

We assessed how clinician satisfaction with a vendor electronic health record (EHR) changed over time in the 4 years following the transition from a homegrown EHR system to identify areas for improvement. We conducted a multiyear survey of clinicians across a large health care system after transitioning to a vendor EHR. Eligible clinicians from the first institution to transition received a survey invitation by email in fall 2016 and then eligible clinicians systemwide received surveys in spring 2018 and spring 2019. The survey included items assessing ease/difficulty of completing tasks and items assessing perceptions of the EHR's value, usability, and impact. One item assessing overall satisfaction and one open-ended question were included. Frequencies and means were calculated, and comparison of means was performed between 2018 and 2019 on all clinicians. A multivariable generalized linear model was performed to predict the outcome of overall satisfaction. Response rates for the surveys ranged from 14 to 19%. The mean response from 3 years of surveys for one institution, Brigham and Women's Hospital, increased for overall satisfaction between 2016 (2.85), 2018 (3.01), and 2019 (3.21, p < 0.001). We found no significant differences in mean response for overall satisfaction between all responders of the 2018 survey (3.14) and those of the 2019 survey (3.19). Systemwide, tasks rated the most difficult included "Monitoring patient medication adherence," "Identifying when a referral has not been completed," and "Making a list of patients based on clinical information (e.g., problem, medication)." Clinicians disagreed the most with "The EHR helps me focus on patient care rather than the computer" and "The EHR allows me to complete tasks efficiently." Survey results indicate room for improvement in clinician satisfaction with the EHR. Usability of EHRs should continue to be an area of focus to ease clinician burden and improve clinician experience.

Informing Visual Display Design of Electronic Health Records: A Human Factors Cross-Industry Perspective

Background Despite their prevalence, poorly designed electronic health records (EHRs) are common, and research shows poor design consequences include clinician burnout, diagnostic error, and even patient harm. One of the major difficulties of EHR design is the visual display of information, which aims to present information in an easily digestible form for the user. High-risk industries like aviation, automotive, and nuclear have guidelines for visual displays based on human factors principles for optimized design. Purpose In this study, we reviewed the visual display guidelines from three high-risk industries—automotive, aviation, nuclear—for their applicability to EHR design and safety. Methods Human factors experts extracted guidelines related to visual displays from automotive, aviation, and nuclear human factors guideline documents. Human factors experts and a clinical expert excluded guidelines irrelevant to EHR. Human factors experts used a modified reflexive thematic analysis to group guidelines into meaningful topics. Disagreements were discussed until a consensus was reached. Results A total of 449 guidelines were extracted from the industry documents, and 283 (63.0%) were deemed relevant to EHRs. By industry, 12 of 44 (27.3%) automotive industry guidelines were relevant, 43 of 115 (37.4%) aviation industry guidelines were relevant, and 228 of 290 (78.6%) nuclear industry guidelines were relevant. Guidelines were grouped into six categories: alphanumeric; color, brightness, contrast, and luminance; comprehension; design characteristics; symbols, pictograms, and icons; and tables, figures, charts, and lists. Conclusion Our analysis identified visual display guidelines organized around six topics from the automotive, aviation, and nuclear industries to inform EHR design. Multiple stakeholders, including EHR vendors, healthcare facilities, and policymakers, can apply these guidelines to design new EHRs and optimize EHRs already in use.

Refining Vendor-Defined Measures to Accurately Quantify EHR Workload Outside Time Scheduled With Patients.

Accurately quantifying clinician time spent on electronic health record (EHR) activities outside the time scheduled with patients is critical for understanding occupational stress associated with ambulatory clinic environments. We make 3 recommendations regarding EHR workload measures that are intended to capture time working in the EHR outside time scheduled with patients, formally defined as work outside of work (WOW): (1) separate all time working in the EHR outside of time scheduled with patients from time working in the EHR during time scheduled with patients, (2) do not exclude any time before or after scheduled time with patients, and (3) encourage the EHR vendor and research communities to develop and standardize validated, vendor-agnostic methods for measuring active EHR use. Attributing all EHR work outside time scheduled with patients to WOW, regardless of when it occurs, will produce an objective and standardized measure better suited for use in efforts to reduce burnout, set policy, and facilitate research.

Deployment of Real-time Natural Language Processing and Deep Learning Clinical Decision Support in the Electronic Health Record: Pipeline Implementation for an Opioid Misuse Screener in Hospitalized Adults.

The clinical narrative in electronic health records (EHRs) carries valuable information for predictive analytics; however, its free-text form is difficult to mine and analyze for clinical decision support (CDS). Large-scale clinical natural language processing (NLP) pipelines have focused on data warehouse applications for retrospective research efforts. There remains a paucity of evidence for implementing NLP pipelines at the bedside for health care delivery. We aimed to detail a hospital-wide, operational pipeline to implement a real-time NLP-driven CDS tool and describe a protocol for an implementation framework with a user-centered design of the CDS tool. The pipeline integrated a previously trained open-source convolutional neural network model for screening opioid misuse that leveraged EHR notes mapped to standardized medical vocabularies in the Unified Medical Language System. A sample of 100 adult encounters were reviewed by a physician informaticist for silent testing of the deep learning algorithm before deployment. An end user interview survey was developed to examine the user acceptability of a best practice alert (BPA) to provide the screening results with recommendations. The planned implementation also included a human-centered design with user feedback on the BPA, an implementation framework with cost-effectiveness, and a noninferiority patient outcome analysis plan. The pipeline was a reproducible workflow with a shared pseudocode for a cloud service to ingest, process, and store clinical notes as Health Level 7 messages from a major EHR vendor in an elastic cloud computing environment. Feature engineering of the notes used an open-source NLP engine, and the features were fed into the deep learning algorithm, with the results returned as a BPA in the EHR. On-site silent testing of the deep learning algorithm demonstrated a sensitivity of 93% (95% CI 66%-99%) and specificity of 92% (95% CI 84%-96%), similar to published validation studies. Before deployment, approvals were received across hospital committees for inpatient operations. Five interviews were conducted; they informed the development of an educational flyer and further modified the BPA to exclude certain patients and allow the refusal of recommendations. The longest delay in pipeline development was because of cybersecurity approvals, especially because of the exchange of protected health information between the Microsoft (Microsoft Corp) and Epic (Epic Systems Corp) cloud vendors. In silent testing, the resultant pipeline provided a BPA to the bedside within minutes of a provider entering a note in the EHR. The components of the real-time NLP pipeline were detailed with open-source tools and pseudocode for other health systems to benchmark. The deployment of medical artificial intelligence systems in routine clinical care presents an important yet unfulfilled opportunity, and our protocol aimed to close the gap in the implementation of artificial intelligence-driven CDS. ClinicalTrials.gov NCT05745480; https://www.clinicaltrials.gov/ct2/show/NCT05745480.