Developing a deep learning natural language processing algorithm for automated reporting of adverse drug reactions.

The detection of adverse drug reactions (ADRs) is critical to our understanding of the safety and risk-benefit profile of medications. With an incidence that has not changed over the last 30 years, ADRs are a significant source of patient morbidity, responsible for 5%-10% of acute care hospital admissions worldwide. Spontaneous reporting of ADRs has long been the standard method of reporting, however this approach is known to have high rates of under-reporting, a problem that limits pharmacovigilance efforts. Automated ADR reporting presents an alternative pathway to increase reporting rates, although this may be limited by over-reporting of other drug-related adverse events. We developed a deep learning natural language processing algorithm to identify ADRs in discharge summaries at a single academic hospital centre. Our model was developed in two stages: first, a pre-trained model (DeBERTa) was further pre-trained on 1.1 million unlabelled clinical documents; secondly, this model was fine-tuned to detect ADR mentions in a corpus of 861 annotated discharge summaries. This model was compared to a version without the pre-training step, and a previously published RoBERTa model pretrained on MIMIC III, which has demonstrated strong performance on other pharmacovigilance tasks. To ensure that our algorithm could differentiate ADRs from other drug-related adverse events, the annotated corpus was enriched for both validated ADR reports and confounding drug-related adverse events using. The final model demonstrated good performance with a ROC-AUC of 0.955 (95% CI 0.933 - 0.978) for the task of identifying discharge summaries containing ADR mentions, significantly outperforming the two comparator models.

Virtual Care Initiatives for Older Adults in Australia: Scoping Review.

BACKGROUND: There has been a rapid shift toward the adoption of virtual health care services in Australia. It is unknown how widely virtual care has been implemented or evaluated for the care of older adults in Australia. OBJECTIVE: We aimed to review the literature evaluating virtual care initiatives for older adults across a wide range of health conditions and modalities and identify key challenges and opportunities for wider adoption at both patient and system levels in Australia. METHODS: A scoping review of the literature was conducted. We searched MEDLINE, Embase, PsycINFO, CINAHL, AgeLine, and gray literature (January 1, 2011, to March 8, 2021) to identify virtual care initiatives for older Australians (aged ≥65 years). The results were reported according to the World Health Organization's digital health evaluation framework. RESULTS: Among the 6296 documents in the search results, we identified 94 that reported 80 unique virtual care initiatives. Most (69/80, 89%) were at the pilot stage and targeted community-dwelling older adults (64/79, 81%) with chronic diseases (52/80, 65%). The modes of delivery included videoconference, telephone, apps, device or monitoring systems, and web-based technologies. Most initiatives showed either similar or better health and behavioral outcomes compared with in-person care. The key barriers for wider adoption were physical, cognitive, or sensory impairment in older adults and staffing issues, legislative issues, and a lack of motivation among providers. CONCLUSIONS: Virtual care is a viable model of care to address a wide range of health conditions among older adults in Australia. More embedded and integrative evaluations are needed to ensure that virtually enabled care can be used more widely by older Australians and health care providers.

Resources for Improving Pediatric Readiness and Quality of Care in Rural Communities and Emergency Departments.

OBJECTIVES: To share the process and products of an 8-year, federally funded grant from the Health Resources and Services Administration Emergency Medical Services for Children program to increase pediatric emergency readiness and quality of care provided in rural communities located within 2 underserved local emergency medical services agencies (LEMSAs) in Northern California. METHODS: In 2 multicounty LEMSAs with 24 receiving hospital emergency departments, we conducted focus groups and interviews with patients and parents, first responders, receiving hospital personnel, and other community stakeholders. From this, we (a regional, urban children's hospital) provided a variety of resources for improving the regionalization and quality of pediatric emergency care provided by prehospital providers and healthcare staff at receiving hospitals in these rural LEMSAs. RESULTS: From this project, we provided resources that included regularly scheduled pediatric-specific training and education programs, pediatric-specific quality improvement initiatives, expansion of telemedicine services, and cultural competency training. We also enhanced community engagement and investment in pediatric readiness. CONCLUSIONS: The resources we provided from our regional, urban children's hospital to 2 rural LEMSAs facilitated improvements in a regionalized system of care for critically ill and injured children. Our shared resources framework can be adapted by other regional children's hospitals to increase readiness and quality of pediatric emergency care in rural and underserved communities and LEMSAs.

A Scoping Review and Content Analysis of Common Depressive Symptoms of Young People.

School nurses are the most accessible health care providers for many young people including adolescents and young adults. Early identification of depression results in improved outcomes, but little information is available comprehensively describing depressive symptoms specific to this population. The aim of this study was to develop a taxonomy of depressive symptoms that were manifested and described by young people based on a scoping review and content analysis. Twenty-five journal articles that included narrative descriptions of depressive symptoms in young people were included. A total of 60 depressive symptoms were identified and categorized into five dimensions: behavioral (n = 8), cognitive (n = 14), emotional (n = 15), interpersonal (n = 13), and somatic (n = 10). This comprehensive depression symptom taxonomy can help school nurses to identify young people who may experience depression and will support future research to better screen for depression.

Telehealth and virtual health monitoring in cystic fibrosis.

PURPOSE OF REVIEW: At many institutions, the Covid-19 pandemic made it necessary to rapidly change the way services are provided to patients, including those with cystic fibrosis (CF). The purpose of this review is to explore the past, present and future of telehealth and virtual monitoring in CF and to highlight certain challenges/considerations in developing such services. RECENT FINDINGS: The Covid-19 pandemic has proven that telehealth and virtual monitoring are a feasible means for safely providing services to CF patients when traditional care is not possible. However, both telehealth and virtual monitoring can also provide further support in the future in a post-covid era through a hybrid-model incorporating traditional care, remote data collection and sophisticated platforms to manage and share data with CF teams. SUMMARY: We provide a detailed overview of telehealth and virtual monitoring including examples of how paediatric and adult CF services adapted to the need for rapid change. Such services have proven popular with people with CF meaning that co-design with stakeholders will likely improve systems further. In the future, telehealth and virtual monitoring will become more sophisticated by harnessing increasingly powerful technologies such as artificial intelligence, connected monitoring devices and wearables. In this review, we harmonise definitions and terminologies before highlighting considerations and limitations for the future of telehealth and virtual monitoring in CF.

Adaptation of an NLP system to a new healthcare environment to identify social determinants of health.

Social determinants of health (SDoH) are increasingly important factors for population health, healthcare outcomes, and care delivery. However, many of these factors are not reliably captured within structured electronic health record (EHR) data. In this work, we evaluated and adapted a previously published NLP tool to include additional social risk factors for deployment at Vanderbilt University Medical Center in an Acute Myocardial Infarction cohort. We developed a transformation of the SDoH outputs of the tool into the OMOP common data model (CDM) for re-use across many potential use cases, yielding performance measures across 8 SDoH classes of precision 0.83 recall 0.74 and F-measure of 0.78.

Examining the association of social risk with heart failure readmission in the Veterans Health Administration.

BACKGROUND: Previous research has found that social risk factors are associated with an increased risk of 30-day readmission. We aimed to assess the association of 5 social risk factors (living alone, lack of social support, marginal housing, substance abuse, and low income) with 30-day Heart Failure (HF) hospital readmissions within the Veterans Health Affairs (VA) and the impact of their inclusion on hospital readmission model performance. METHODS: We performed a retrospective cohort study using chart review and VA and Centers for Medicare and Medicaid Services (CMS) administrative data from a random sample of 1,500 elderly (≥ 65 years) Veterans hospitalized for HF in 2012. Using logistic regression, we examined whether any of the social risk factors were associated with 30-day readmission after adjusting for age alone and clinical variables used by CMS in its 30-day risk stratified readmission model. The impact of these five social risk factors on readmission model performance was assessed by comparing c-statistics, likelihood ratio tests, and the Hosmer-Lemeshow goodness-of-fit statistic. RESULTS: The prevalence varied among the 5 risk factors; low income (47 % vs. 47 %), lives alone (18 % vs. 19 %), substance abuse (14 % vs. 16 %), lacks social support (2 % vs. <1 %), and marginal housing (< 1 % vs. 3 %) among readmitted and non-readmitted patients, respectively. Controlling for clinical factors contained in CMS readmission models, a lack of social support was found to be associated with an increased risk of 30-day readmission (OR 4.8, 95 %CI 1.35-17.88), while marginal housing was noted to decrease readmission risk (OR 0.21, 95 %CI 0.03-0.87). Living alone (OR: 0.9, 95 %CI 0.64-1.26), substance abuse (OR 0.91, 95 %CI 0.67-1.22), and having low income (OR 1.01, 95 %CI 0.77-1.31) had no association with HF readmissions. Adding the five social risk factors to a CMS-based model (age and comorbid conditions; c-statistic 0.62) did not improve model performance (c-statistic: 0.62). CONCLUSIONS: While a lack of social support was associated with 30-day readmission in the VA, its prevalence was low. Moreover, the inclusion of some social risk factors did not improve readmission model performance. In an integrated healthcare system like the VA, social risk factors may have a limited effect on 30-day readmission outcomes.

Portable Automated Surveillance of Surgical Site Infections Using Natural Language Processing: Development and Validation.

OBJECTIVES: We present the development and validation of a portable NLP approach for automated surveillance of SSIs. SUMMARY OF BACKGROUND DATA: The surveillance of SSIs is labor-intensive limiting the generalizability and scalability of surgical quality surveillance programs. METHODS: We abstracted patient clinical text notes after surgical procedures from 2 independent healthcare systems using different electronic healthcare records. An SSI detected as part of the American College of Surgeons' National Surgical Quality Improvement Program was used as the reference standard. We developed a rules-based NLP system (Easy Clinical Information Extractor [CIE]-SSI) for operative event-level detection of SSIs using an training cohort (4574 operative events) from 1 healthcare system and then conducted internal validation on a blind cohort from the same healthcare system (1850 operative events) and external validation on a blind cohort from the second healthcare system (15,360 operative events). EasyCIE-SSI performance was measured using sensitivity, specificity, and area under the receiver-operating-curve (AUC). RESULTS: The prevalence of SSI was 4% and 5% in the internal and external validation corpora. In internal validation, EasyCIE-SSI had a sensitivity, specificity, AUC of 94%, 88%, 0.912 for the detection of SSI, respectively. In external validation, EasyCIE-SSI had sensitivity, specificity, AUC of 79%, 92%, 0.852 for the detection of SSI, respectively. The sensitivity of EasyCIE-SSI decreased in clean, skin/subcutaneous, and outpatient procedures in the external validation compared to internal validation. CONCLUSION: Automated surveillance of SSIs can be achieved using NLP of clinical notes with high sensitivity and specificity.

Emergency Medical Services Utilization by Children.

OBJECTIVE: The aim of this study was to compare demographic and clinical features of children (0-14 years old) who arrived at general emergency departments (EDs) by emergency medical services (EMS) to those who arrived by private vehicles and other means in a rural, 3-county region of northern California. METHODS: We reviewed 507 ED records of children who arrived at EDs by EMS and those who arrived by other means in 2013. We also analyzed prehospital procedures performed on all children transported to an area hospital by EMS. RESULTS: Children arriving by EMS were older (9.0 vs 6.0 years; P < 0.001), more ill (mean Severity Classification Score, 2.9 vs 2.4; P < 0.001), and had longer lengths of stay (3.6 vs 2.1 hours; P < 0.001) compared with children who were transported to the EDs by other means. Children transported by EMS received more subspecialty consultations (18.7% vs 6.9%; P < 0.05) and had more diagnostic testing, including laboratory testing (22.9% vs 10.6%; P < 0.001), radiography (39.7% vs 20.8%; P < 0.001), and computed tomography scans (16.8% vs 2.9%; P < 0.001). Children arriving by EMS were transferred more frequently (8.8% vs 1.6%; P < 0.001) and had higher mean Severity Classification Scores compared with children arriving by other transportation even after adjusting for age and sex (ß = 0.48; 95% confidence interval, 0.35-0.61; P < 0.001). Older children received more prehospital procedures compared with younger children, and these were of greater complexity and a wider spectrum. CONCLUSIONS: Children transported to rural EDs via EMS are more ill and use more medical resources compared with those who arrive to the ED by other means of transportation.

Using clinical Natural Language Processing for health outcomes research: Overview and actionable suggestions for future advances.

The importance of incorporating Natural Language Processing (NLP) methods in clinical informatics research has been increasingly recognized over the past years, and has led to transformative advances. Typically, clinical NLP systems are developed and evaluated on word, sentence, or document level annotations that model specific attributes and features, such as document content (e.g., patient status, or report type), document section types (e.g., current medications, past medical history, or discharge summary), named entities and concepts (e.g., diagnoses, symptoms, or treatments) or semantic attributes (e.g., negation, severity, or temporality). From a clinical perspective, on the other hand, research studies are typically modelled and evaluated on a patient- or population-level, such as predicting how a patient group might respond to specific treatments or patient monitoring over time. While some NLP tasks consider predictions at the individual or group user level, these tasks still constitute a minority. Owing to the discrepancy between scientific objectives of each field, and because of differences in methodological evaluation priorities, there is no clear alignment between these evaluation approaches. Here we provide a broad summary and outline of the challenging issues involved in defining appropriate intrinsic and extrinsic evaluation methods for NLP research that is to be used for clinical outcomes research, and vice versa. A particular focus is placed on mental health research, an area still relatively understudied by the clinical NLP research community, but where NLP methods are of notable relevance. Recent advances in clinical NLP method development have been significant, but we propose more emphasis needs to be placed on rigorous evaluation for the field to advance further. To enable this, we provide actionable suggestions, including a minimal protocol that could be used when reporting clinical NLP method development and its evaluation.

Using natural language processing to improve efficiency of manual chart abstraction in research: the case of breast cancer recurrence.

The increasing availability of electronic health records (EHRs) creates opportunities for automated extraction of information from clinical text. We hypothesized that natural language processing (NLP) could substantially reduce the burden of manual abstraction in studies examining outcomes, like cancer recurrence, that are documented in unstructured clinical text, such as progress notes, radiology reports, and pathology reports. We developed an NLP-based system using open-source software to process electronic clinical notes from 1995 to 2012 for women with early-stage incident breast cancers to identify whether and when recurrences were diagnosed. We developed and evaluated the system using clinical notes from 1,472 patients receiving EHR-documented care in an integrated health care system in the Pacific Northwest. A separate study provided the patient-level reference standard for recurrence status and date. The NLP-based system correctly identified 92% of recurrences and estimated diagnosis dates within 30 days for 88% of these. Specificity was 96%. The NLP-based system overlooked 5 of 65 recurrences, 4 because electronic documents were unavailable. The NLP-based system identified 5 other recurrences incorrectly classified as nonrecurrent in the reference standard. If used in similar cohorts, NLP could reduce by 90% the number of EHR charts abstracted to identify confirmed breast cancer recurrence cases at a rate comparable to traditional abstraction.

Evaluating the effects of machine pre-annotation and an interactive annotation interface on manual de-identification of clinical text.

The Health Insurance Portability and Accountability Act (HIPAA) Safe Harbor method requires removal of 18 types of protected health information (PHI) from clinical documents to be considered "de-identified" prior to use for research purposes. Human review of PHI elements from a large corpus of clinical documents can be tedious and error-prone. Indeed, multiple annotators may be required to consistently redact information that represents each PHI class. Automated de-identification has the potential to improve annotation quality and reduce annotation time. For instance, using machine-assisted annotation by combining de-identification system outputs used as pre-annotations and an interactive annotation interface to provide annotators with PHI annotations for "curation" rather than manual annotation from "scratch" on raw clinical documents. In order to assess whether machine-assisted annotation improves the reliability and accuracy of the reference standard quality and reduces annotation effort, we conducted an annotation experiment. In this annotation study, we assessed the generalizability of the VA Consortium for Healthcare Informatics Research (CHIR) annotation schema and guidelines applied to a corpus of publicly available clinical documents called MTSamples. Specifically, our goals were to (1) characterize a heterogeneous corpus of clinical documents manually annotated for risk-ranked PHI and other annotation types (clinical eponyms and person relations), (2) evaluate how well annotators apply the CHIR schema to the heterogeneous corpus, (3) compare whether machine-assisted annotation (experiment) improves annotation quality and reduces annotation time compared to manual annotation (control), and (4) assess the change in quality of reference standard coverage with each added annotator's annotations.

Clinical Experiences of Working with Patient-Generated Health Data in Primary Care.

This research aims to provide insight into the GP experience with patient-generated health data (PGHD) in a virtual care visit. Despite the prevalence of wearables, including smartwatches, the acceptability of generated data in primary care is understudied. The result of this study from mixed-method analysis showed the basic capabilities of PGHD to enhance clinical decision-making and positive impact on collaboration with the patient. The impact of PGHD on clinician satisfaction was not determined, highlighting the importance of rigorous methodology in future research.

Analyzing the Spread of Informatics with PubMed.

We analyzed PubMed citations since 1988 to explore the dissemination of medical/health informatics concepts between countries and across medical domains. We extracted countries from the PubMed author affiliation field to identify and analyze the top 10 informatics publishing countries. We found that the informatics publications are becoming more similar over time and that the rate of exchange across countries has increased with the introduction of e-publishing. Nonetheless, with the exception of machine learning, the impact of core informatics concepts on mainstream medicine and radiology publications remains small.

A Hybrid Physical-Digital Simulation Laboratory to Expedite Context-Aware Design and Usability Testing in Digital Health.

Fit within existing physical and digitalised workflows is a critical aspect of digital health software usability. Early, iterative exploration of contextual usability issues is complicated by barriers of access to healthcare settings. The Validitron SimLab is a new facility for digital health prototyping that augments immersive, realistic physical environments with a digital sandbox allowing new and existing software to be easily set up and tested in the physical space.

Adapting an Environmental Scan for 'Insights Reporting': Learnings from an Online Brain Cancer Peer Support Platform.

When developing a digital health solution, product owners, healthcare professionals, researchers, IT teams, and consumers require timely, accurate contextual information to inform solution development. Insights Reporting can rapidly draw together information from literature, end users and existing technology to inform the development process. This was the case when creating an online brain cancer peer support platform where solution development was conducted in parallel with contextual information synthesis. This paper discusses the novel adaptation of an environmental scan methodology using codesign and multiple layers of qualitative rigor, to create Insights Reporting. This seven-step process can be completed in two months and results in salient points of knowledge that can rapidly inform the design of a solution, creating a shared understanding of a digital health phenomenon. Project members noted that Insights Reporting surfaces previously inaccessible knowledge, catalyzes decision-making and allows all stakeholders to influence the report agenda, affirming principles of digital health equity.

Automatic Extraction of Skin and Soft Tissue Infection Status from Clinical Notes.

The reliable identification of skin and soft tissue infections (SSTIs) from electronic health records is important for a number of applications, including quality improvement, clinical guideline construction, and epidemiological analysis. However, in the United States, types of SSTIs (e.g. is the infection purulent or non-purulent?) are not captured reliably in structured clinical data. With this work, we trained and evaluated a rule-based clinical natural language processing system using 6,576 manually annotated clinical notes derived from the United States Veterans Health Administration (VA) with the goal of automatically extracting and classifying SSTI subtypes from clinical notes. The trained system achieved mention- and document-level performance metrics of the range 0.39 to 0.80 for mention level classification and 0.49 to 0.98 for document level classification.

Designing an Interprofessional Online Course to Foster Learning Health Systems.

The Learning Health Systems (LHS) framework demonstrates the potential for iterative interrogation of health data in real time and implementation of insights into practice. Yet, the lack of appropriately skilled workforce results in an inability to leverage existing data to design innovative solutions. We developed a tailored professional development program to foster a skilled workforce. The short course is wholly online, for interdisciplinary professionals working in the digital health arena. To transform healthcare systems, the workforce needs an understanding of LHS principles, data driven approaches, and the need for diversly skilled learning communities that can tackle these complex problems together.

Centering Digital Health Equity During Technology Innovation: Protocol for a Comprehensive Scoping Review of Evidence-Based Tools and Approaches.

BACKGROUND: In the rush to develop health technologies for the COVID-19 pandemic, the unintended consequence of digital health inequity or the inability of priority communities to access, use, and receive equal benefits from digital health technologies was not well examined. OBJECTIVE: This scoping review will examine tools and approaches that can be used during digital technology innovation to improve equitable inclusion of priority communities in the development of digital health technologies. The results from this study will provide actionable insights for professionals in health care, health informatics, digital health, and technology development to proactively center equity during innovation. METHODS: Based on the Arksey and O'Malley framework, this scoping review will consider priority communities' equitable involvement in digital technology innovation. Bibliographic databases in health, medicine, computing, and information sciences will be searched. Retrieved citations will be double screened against the inclusion and exclusion criteria using Covidence (Veritas Health Innovation). Data will be charted using a tailored extraction tool and mapped to a digital health innovation pathway defined by the Centre for eHealth Research roadmap for eHealth technologies. An accompanying narrative synthesis will describe the outcomes in relation to the review's objectives. RESULTS: This scoping review is currently in progress. The search of databases and other sources returned a total of 4868 records. After the initial screening of titles and abstracts, 426 studies are undergoing dual full-text review. We are aiming to complete the full-text review stage by May 30, 2024, data extraction in October 2024, and subsequent synthesis in December 2024. Funding was received on October 1, 2023, from the Centre for Health Equity Incubator Grant Scheme, University of Melbourne, Australia. CONCLUSIONS: This paper will identify and recommend a series of validated tools and approaches that can be used by health care stakeholders and IT developers to produce equitable digital health technology across the Centre for eHealth Research roadmap. Identified evidence gaps, possible implications, and further research will be discussed. INTERNATIONAL REGISTERED REPORT IDENTIFIER (IRRID): DERR1-10.2196/53855.

Using chief complaints for syndromic surveillance: a review of chief complaint based classifiers in North America.

A major goal of Natural Language Processing in the public health informatics domain is the automatic extraction and encoding of data stored in free text patient records. This extracted data can then be utilized by computerized systems to perform syndromic surveillance. In particular, the chief complaint--a short string that describes a patient's symptoms--has come to be a vital resource for syndromic surveillance in the North American context due to its near ubiquity. This paper reviews fifteen systems in North America--at the city, county, state and federal level--that use chief complaints for syndromic surveillance.