Evaluating Natural Language Processing Packages for Predicting Hospital-Acquired Pressure Injuries From Clinical Notes.

Incidence of hospital-acquired pressure injury, a key indicator of nursing quality, is directly proportional to adverse outcomes, increased hospital stays, and economic burdens on patients, caregivers, and society. Thus, predicting hospital-acquired pressure injury is important. Prediction models use structured data more often than unstructured notes, although the latter often contain useful patient information. We hypothesize that unstructured notes, such as nursing notes, can predict hospital-acquired pressure injury. We evaluate the impact of using various natural language processing packages to identify salient patient information from unstructured text. We use named entity recognition to identify keywords, which comprise the feature space of our classifier for hospital-acquired pressure injury prediction. We compare scispaCy and Stanza, two different named entity recognition models, using unstructured notes in Medical Information Mart for Intensive Care III, a publicly available ICU data set. To assess the impact of vocabulary size reduction, we compare the use of all clinical notes with only nursing notes. Our results suggest that named entity recognition extraction using nursing notes can yield accurate models. Moreover, the extracted keywords play a significant role in the prediction of hospital-acquired pressure injury.

An AdaBoost-based algorithm to detect hospital-acquired pressure injury in the presence of conflicting annotations.

Hospital-acquired pressure injury is one of the most harmful events in clinical settings. Patients who do not receive early prevention and treatment can experience a significant financial burden and physical trauma. Several hospital-acquired pressure injury prediction algorithms have been developed to tackle this problem, but these models assume a consensus, gold-standard label (i.e., presence of pressure injury or not) is present for all training data. Existing definitions for identifying hospital-acquired pressure injuries are inconsistent due to the lack of high-quality documentation surrounding pressure injuries. To address this issue, we propose in this paper an ensemble-based algorithm that leverages truth inference methods to resolve label inconsistencies between various case definitions and the level of disagreements in annotations. Application of our method to MIMIC-III, a publicly available intensive care unit dataset, gives empirical results that illustrate the promise of learning a prediction model using truth inference-based labels and observed conflict among annotators.

Evaluation of available risk scores to predict multiple cardiovascular complications for patients with type 2 diabetes mellitus using electronic health records.

Aims: Various cardiovascular risk prediction models have been developed for patients with type 2 diabetes mellitus. Yet few models have been validated externally. We perform a comprehensive validation of existing risk models on a heterogeneous population of patients with type 2 diabetes using secondary analysis of electronic health record data. Methods: Electronic health records of 47,988 patients with type 2 diabetes between 2013 and 2017 were used to validate 16 cardiovascular risk models, including 5 that had not been compared previously, to estimate the 1-year risk of various cardiovascular outcomes. Discrimination and calibration were assessed by the c-statistic and the Hosmer-Lemeshow goodness-of-fit statistic, respectively. Each model was also evaluated based on the missing measurement rate. Sub-analysis was performed to determine the impact of race on discrimination performance. Results: There was limited discrimination (c-statistics ranged from 0.51 to 0.67) across the cardiovascular risk models. Discrimination generally improved when the model was tailored towards the individual outcome. After recalibration of the models, the Hosmer-Lemeshow statistic yielded p-values above 0.05. However, several of the models with the best discrimination relied on measurements that were often imputed (up to 39% missing). Conclusion: No single prediction model achieved the best performance on a full range of cardiovascular endpoints. Moreover, several of the highest-scoring models relied on variables with high missingness frequencies such as HbA1c and cholesterol that necessitated data imputation and may not be as useful in practice. An open-source version of our developed Python package, cvdm, is available for comparisons using other data sources.

A Comprehensive and Improved Definition for Hospital-Acquired Pressure Injury Classification Based on Electronic Health Records: Comparative Study.

BACKGROUND: Patients develop pressure injuries (PIs) in the hospital owing to low mobility, exposure to localized pressure, circulatory conditions, and other predisposing factors. Over 2.5 million Americans develop PIs annually. The Center for Medicare and Medicaid considers hospital-acquired PIs (HAPIs) as the most frequent preventable event, and they are the second most common claim in lawsuits. With the growing use of electronic health records (EHRs) in hospitals, an opportunity exists to build machine learning models to identify and predict HAPI rather than relying on occasional manual assessments by human experts. However, accurate computational models rely on high-quality HAPI data labels. Unfortunately, the different data sources within EHRs can provide conflicting information on HAPI occurrence in the same patient. Furthermore, the existing definitions of HAPI disagree with each other, even within the same patient population. The inconsistent criteria make it impossible to benchmark machine learning methods to predict HAPI. OBJECTIVE: The objective of this project was threefold. We aimed to identify discrepancies in HAPI sources within EHRs, to develop a comprehensive definition for HAPI classification using data from all EHR sources, and to illustrate the importance of an improved HAPI definition. METHODS: We assessed the congruence among HAPI occurrences documented in clinical notes, diagnosis codes, procedure codes, and chart events from the Medical Information Mart for Intensive Care III database. We analyzed the criteria used for the 3 existing HAPI definitions and their adherence to the regulatory guidelines. We proposed the Emory HAPI (EHAPI), which is an improved and more comprehensive HAPI definition. We then evaluated the importance of the labels in training a HAPI classification model using tree-based and sequential neural network classifiers. RESULTS: We illustrate the complexity of defining HAPI, with <13% of hospital stays having at least 3 PI indications documented across 4 data sources. Although chart events were the most common indicator, it was the only PI documentation for >49% of the stays. We demonstrate a lack of congruence across existing HAPI definitions and EHAPI, with only 219 stays having a consensus positive label. Our analysis highlights the importance of our improved HAPI definition, with classifiers trained using our labels outperforming others on a small manually labeled set from nurse annotators and a consensus set in which all definitions agreed on the label. CONCLUSIONS: Standardized HAPI definitions are important for accurately assessing HAPI nursing quality metric and determining HAPI incidence for preventive measures. We demonstrate the complexity of defining an occurrence of HAPI, given the conflicting and incomplete EHR data. Our EHAPI definition has favorable properties, making it a suitable candidate for HAPI classification tasks.

Subgingival Microbiome in Pregnancy and a Potential Relationship to Early Term Birth.

Background: Periodontal disease in pregnancy is considered a risk factor for adverse birth outcomes. Periodontal disease has a microbial etiology, however, the current state of knowledge about the subgingival microbiome in pregnancy is not well understood. Objective: To characterize the structure and diversity of the subgingival microbiome in early and late pregnancy and explore relationships between the subgingival microbiome and preterm birth among pregnant Black women. Methods: This longitudinal descriptive study used 16S rRNA sequencing to profile the subgingival microbiome of 59 Black women and describe microbial ecology using alpha and beta diversity metrics. We also compared microbiome features across early (8-14 weeks) and late (24-30 weeks) gestation overall and according to gestational age at birth outcomes (spontaneous preterm, spontaneous early term, full term). Results: In this sample of Black pregnant women, the top twenty bacterial taxa represented in the subgingival microbiome included a spectrum representative of various stages of biofilm progression leading to periodontal disease, including known periopathogens Porphyromonas gingivalis and Tannerella forsythia. Other organisms associated with periodontal disease reflected in the subgingival microbiome included several Prevotella spp., and Campylobacter spp. Measures of alpha or beta diversity did not distinguish the subgingival microbiome of women according to early/late gestation or full term/spontaneous preterm birth; however, alpha diversity differences in late pregnancy between women who spontaneously delivered early term and women who delivered full term were identified. Several taxa were also identified as being differentially abundant according to early/late gestation, and full term/spontaneous early term births. Conclusions: Although the composition of the subgingival microbiome is shifted toward complexes associated with periodontal disease, the diversity of the microbiome remains stable throughout pregnancy. Several taxa were identified as being associated with spontaneous early term birth. Two, in particular, are promising targets of further investigation. Depletion of the oral commensal Lautropia mirabilis in early pregnancy and elevated levels of Prevotella melaninogenica in late pregnancy were both associated with spontaneous early term birth.

GASP: Graph-based Approximate Sequential Pattern Mining for Electronic Health Records.

Sequential pattern mining can be used to extract meaningful sequences from electronic health records. However, conventional sequential pattern mining algorithms that discover all frequent sequential patterns can incur a high computational and be susceptible to noise in the observations. Approximate sequential pattern mining techniques have been introduced to address these shortcomings yet, existing approximate methods fail to reflect the true frequent sequential patterns or only target single-item event sequences. Multi-item event sequences are prominent in healthcare as a patient can have multiple interventions for a single visit. To alleviate these issues, we propose GASP, a graph-based approximate sequential pattern mining, that discovers frequent patterns for multi-item event sequences. Our approach compresses the sequential information into a concise graph structure which has computational benefits. The empirical results on two healthcare datasets suggest that GASP outperforms existing approximate models by improving recoverability and extracts better predictive patterns.

Privacy-preserving Sequential Pattern Mining in distributed EHRs for Predicting Cardiovascular Disease.

From electronic health records (EHRs), the relationship between patients' conditions, treatments, and outcomes can be discovered and used in various healthcare research tasks such as risk prediction. In practice, EHRs can be stored in one or more data warehouses, and mining from distributed data sources becomes challenging. Another challenge arises from privacy laws because patient data cannot be used without some patient privacy guarantees. Thus, in this paper, we propose a privacy-preserving framework using sequential pattern mining in distributed data sources. Our framework extracts patterns from each source and shares patterns with other sources to discover discriminative and representative patterns that can be used for risk prediction while preserving privacy. We demonstrate our framework using a case study of predicting Cardiovascular Disease in patients with type 2 diabetes and show the effectiveness of our framework with several sources and by applying differential privacy mechanisms.

How stress, discrimination, acculturation and the gut microbiome affect depression, anxiety and sleep among Chinese and Korean immigrants in the USA: a cross-sectional pilot study protocol.

INTRODUCTION: Although a considerable proportion of Asians in the USA experience depression, anxiety and poor sleep, these health issues have been underestimated due to the model minority myth about Asians, the stigma associated with mental illness, lower rates of treatment seeking and a shortage of culturally tailored mental health services. Indeed, despite emerging evidence of links between psychosocial risk factors, the gut microbiome and depression, anxiety and sleep quality, very few studies have examined how these factors are related in Chinese and Korean immigrants in the USA. The purpose of this pilot study was to address this issue by (a) testing the usability and feasibility of the study's multilingual survey measures and biospecimen collection procedure among Chinese and Korean immigrants in the USA and (b) examining how stress, discrimination, acculturation and the gut microbiome are associated with depression, anxiety and sleep quality in this population. METHOD AND ANALYSIS: This is a cross-sectional pilot study among first and second generations of adult Chinese and Korean immigrants in the greater Atlanta area (Georgia, USA). We collected (a) gut microbiome samples and (b) data on psychosocial risk factors, depression, anxiety and sleep disturbance using validated, online surveys in English, Chinese and Korean. We aim to recruit 60 participants (30 Chinese, 30 Korean). We will profile participants' gut microbiome using 16S rRNA V3-V4 sequencing data, which will be analysed by QIIME 2. Associations of the gut microbiome and psychosocial factors with depression, anxiety and sleep disturbance will be analysed using descriptive and inferential statistics, including linear regression. ETHICS AND DISSEMINATION: This study has been approved by the Institutional Review Board at Emory University (IRB ID: STUDY00000935). Results will be made available to Chinese and Korean community members, the funder and other researchers and the broader scientific community.

Predictors of progression through the cascade of care to a cure for hepatitis C patients using decision trees and random forests.

BACKGROUND: This study uses machine learning techniques to identify sociodemographic and clinical predictors of progression through the hepatitis C (HCV) cascade of care for patients in the 1945-1965 birth cohort in the Southern United States. METHODS: We compared sociodemographic and clinical variables between groups of patients for three care outcomes: linkage to care, initiation of antiviral treatment, and virologic cure. A decision tree model and random forest model were built for each outcome. RESULTS: Patients were primarily male, African American/Black or Caucasian/White, non-Hispanic or Latino, and insured. The average age at first HCV screening was 60 years old, and common medical diagnoses included chronic kidney disease, fibrosis and/or cirrhosis, transplanted liver, diabetes mellitus, and liver cell carcinoma. Variables used in predicting linkage to care included age at first HCV screening, insurance at first HCV screening, race, fibrosis and/or cirrhosis, other liver disease, ascites, and transplanted liver. Variables used in predicting initiation of antiviral treatment included insurance at first HCV screening, gender, other liver cancer, steatosis, and liver cell carcinoma. Variables used in predicting virologic cure included insurance at first HCV screening, transplanted liver, and ethnicity. CONCLUSION: These patients have a high hepatic health burden, likely reflecting complications of untreated HCV and highlighting the urgency to cure HCV in this birth cohort. We found differences in HCV care outcomes based on sociodemographic and clinical variables. More work is needed to understand the mechanisms of these differences in care outcomes and to improve HCV care.

Toolkit for Best Practice Use of Electronic Health Record Data in Quality Improvement.

This project piloted an educational intervention focused on use and management of EHR data by Doctor of Nursing Practice students in quality improvement initiatives. Recommendations from academic and clinical nursing promote the integration of EHR data findings into practice. Nursing's general lack of understanding about how to use and manage data is a barrier to using EHR data to guide quality improvement initiatives. Doctor of Nursing Practice students at a hospital-affiliated university participated in a pre-test, training, and post-test through an online learning management system. Training content and assessments focused on data and planning for its use in quality improvement initiatives. Sixteen students experienced a median of 17.6% increase in scores after completing the post-test. There was a statistically significant increase in scores between the pre-test and post-test (P = .0006). These results suggest educational content included in the Doctor of Nursing Practice Quality Improvement Toolkit increases knowledge about use and management of EHR data. Future considerations include use for educating a variety of students and healthcare staff.

Pressure Ulcer Injury in Unstructured Clinical Notes: Detection and Interpretation.

Hospital-acquired pressure ulcer injury (PUI) is a primary nursing quality metric, reflecting the caliber of nursing care within a hospital. Prior studies have used the Braden scale and structured data from the electronic health records to detect/predict PUI while the informative unstructured clinical notes have not been used. We propose automated PUI detection using a novel negation-detection algorithm applied to unstructured clinical notes. Our detection framework is on-demand, requiring minimal cost. In application to the MIMIC-III dataset, the text features produced using our algorithm resulted in improved PUI detection when evaluated using logistic regression, random forests, and neural networks compared to text features without negation detection. Exploratory analysis reveals substantial overlap between key classifier features and leading clinical attributes of PUI, adding interpretability to our solution. Our method could also considerably reduce nurses' evaluations by automatic detection of most cases, leaving only the most uncertain cases for nursing assessment.

NuRsing Research in the 21st Century: R You Ready?

Nurse scientists are adept at translating findings from basic science into useful clinical- and community-based interventions to improve health. Over time, the focus of some nursing research has grown to include the assessment and evaluation of genomic and other output from high-throughput, or "omic," technologies as indicators related to health and disease. To date, the growth in the application of omics technologies in nursing research has included calls to increase attention to omics in nursing school curricula and educational training opportunities, such as the Summer Genetics Institute offered by the National Institute of Nursing Research. However, there has been scant attention paid in the nursing literature to the complexity of data analysis or issues of reproducibility related to omics studies. The goals of this article are to (1) familiarize nurse scientists with tools that encourage reproducibility in omics studies, with a focus on the free and open-source data processing and analysis pipeline, and (2) provide a baseline understanding of how these tools can be used to improve collaboration and cohesion among interdisciplinary research team members. Knowledge of these tools and skill in applying them will be important for communication across disciplines and imperative for the advancement of omics research in nursing.

The Airplane Cabin Microbiome.

Serving over three billion passengers annually, air travel serves as a conduit for infectious disease spread, including emerging infections and pandemics. Over two dozen cases of in-flight transmissions have been documented. To understand these risks, a characterization of the airplane cabin microbiome is necessary. Our study team collected 229 environmental samples on ten transcontinental US flights with subsequent 16S rRNA sequencing. We found that bacterial communities were largely derived from human skin and oral commensals, as well as environmental generalist bacteria. We identified clear signatures for air versus touch surface microbiome, but not for individual types of touch surfaces. We also found large flight-to-flight beta diversity variations with no distinguishing signatures of individual flights, rather a high between-flight diversity for all touch surfaces and particularly for air samples. There was no systematic pattern of microbial community change from pre- to post-flight. Our findings are similar to those of other recent studies of the microbiome of built environments. In summary, the airplane cabin microbiome has immense airplane to airplane variability. The vast majority of airplane-associated microbes are human commensals or non-pathogenic, and the results provide a baseline for non-crisis-level airplane microbiome conditions.

Potential Role of the Gut Microbiome in ALS: A Systematic Review.

Amyotrophic lateral sclerosis (ALS) etiology and pathophysiology are not well understood. Recent data suggest that dysbiosis of gut microbiota may contribute to ALS etiology and progression. This review aims to explore evidence of associations between gut microbiota and ALS etiology and pathophysiology. Databases were searched for publications relevant to the gut microbiome in ALS. Three publications provided primary evidence of changes in microbiome profiles in ALS. An ALS mouse model revealed damaged tight junction structure and increased permeability in the intestine versus controls along with a shifted microbiome profile, including decreased levels of butyrate-producing bacteria. In a subsequent publication, again using an ALS mouse model, researchers showed that dietary supplementation with butyrate relieved symptoms and lengthened both time to onset of weight loss and survival time. In a small study of ALS patients and healthy controls, investigators also found decreased levels of butyrate-producing bacteria. Essential for maintaining gut barrier integrity, butyrate is the preferred energy source of intestinal epithelial cells. Ten other articles were reviews and commentaries providing indirect support for a role of gut microbiota in ALS pathophysiology. Thus, these studies provide a modicum of evidence implicating gut microbiota in ALS disease, although more research is needed to confirm the connection and determine pathophysiologic mechanisms. Nurses caring for these patients need to understand the gut microbiome and its potential role in ALS in order to effectively counsel patients and their families about emerging therapies (e.g., prebiotics, probiotics, and fecal microbial transplant) and their off-label uses.

The Risk of Cross Infection in the Emergency Department: A Simulation Study.

OBJECTIVESThe risk of cross infection in a busy emergency department (ED) is a serious public health concern, especially in times of pandemic threats. We simulated cross infections due to respiratory diseases spread by large droplets using empirical data on contacts (ie, close-proximity interactions of ≤1m) in an ED to quantify risks due to contact and to examine factors with differential risks associated with them.DESIGNProspective study.PARTICIPANTSHealth workers (HCWs) and patients.SETTINGA busy ED.METHODSData on contacts between participants were collected over 6 months by observing two 12-hour shifts per week using a radiofrequency identification proximity detection system. We simulated cross infection due to a novel agent across these contacts to determine risks associated with HCW role, chief complaint category, arrival mode, and ED disposition status.RESULTSCross-infection risk between HCWs was substantially greater than between patients or between patients and HCWs. Providers had the least risk, followed by nurses, and nonpatient care staff had the most risk. There were no differences by patient chief complaint category. We detected differential risk patterns by arrival mode and by HCW role. Although no differential risk was associated with ED disposition status, 0.1 infections were expected per shift among patients admitted to hospital.CONCLUSIONThese simulations demonstrate that, on average, 11 patients who were infected in the ED will be admitted to the hospital over the course of an 8-week local influenza outbreak. These patients are a source of further cross-infection risk once in the hospital.Infect Control Hosp Epidemiol 2018;39:688-693.

Behaviors, movements, and transmission of droplet-mediated respiratory diseases during transcontinental airline flights.

With over 3 billion airline passengers annually, the inflight transmission of infectious diseases is an important global health concern. Over a dozen cases of inflight transmission of serious infections have been documented, and air travel can serve as a conduit for the rapid spread of newly emerging infections and pandemics. Despite sensational media stories and anecdotes, the risks of transmission of respiratory viruses in an airplane cabin are unknown. Movements of passengers and crew may facilitate disease transmission. On 10 transcontinental US flights, we chronicled behaviors and movements of individuals in the economy cabin on single-aisle aircraft. We simulated transmission during flight based on these data. Our results indicate there is low probability of direct transmission to passengers not seated in close proximity to an infectious passenger. This data-driven, dynamic network transmission model of droplet-mediated respiratory disease is unique. To measure the true pathogen burden, our team collected 229 environmental samples during the flights. Although eight flights were during Influenza season, all qPCR assays for 18 common respiratory viruses were negative.

Contact networks in the emergency department: Effects of time, environment, patient characteristics, and staff role.

Emergency departments play a critical role in the public health system, particularly in times of pandemic. Infectious patients presenting to emergency departments bring a risk of cross-infection to other patients and staff through close proximity interactions or contacts. To understand factors associated with cross-infection risk, we measured close proximity interactions of emergency department staff and patients by radiofrequency identification in a working emergency department. The number of contacts (degree) is not related to patient demographic characteristics. However, the amount of time in close proximity (weighted degree) of patients with ED personnel did differ, with black patients having approximately 15 min more contact with staff than non-white patients. Patients arriving by EMS had fewer contacts with other patients than patients arriving by other means. There are differences in the number of contacts based on staff role and arrival mode. When crowding is low, providers have the most contact time with patients, while administrative staff have the least. However, when crowding is high, this differential is reversed. The effect of arrival mode is modified by the extent of crowding. When crowding is low, patients arriving by EMS had longer contact with administrative staff, compared to patients arriving by other means. However, when crowding is high, patients arriving by EMS had less contact with administrative staff compared to patients arriving by other means. Our findings should help designers of emergency care focus on higher risk situations for transmission of dangerous pathogens in an emergency department. For instance, the effects of arrival and crowding should be considered as targets for engineering or architectural interventions that could artificially increase social distances.

On the 2-Row Rule for Infectious Disease Transmission on Aircraft.

BACKGROUND: With over two billion airline passengers annually, in-flight transmission of infectious diseases is an important global health concern. Many instances of in-flight transmission have been documented, but the relative influence of the many factors (see below) affecting in-flight transmission has not been quantified. Long-standing guidance by public health agencies is that the primary transmission risk associated with air travel for most respiratory infectious diseases is associated with sitting within two rows of an infectious passenger. The effect of proximity may be one of these factors. OBJECTIVE: The aim of this study was to determine the risk of infection within and beyond the 2-row rule given by public health guidance. METHODS: We searched the literature for reports of in-flight transmission of infection which included seat maps indicating where the infectious and infected passengers were seated. FINDINGS: There is a ∼ 6% risk to passengers seated within the 2-rows of infected individual(s) and there is ∼ 2% risk to passengers seated beyond 2-rows from the infectious individual. DISCUSSION: Contact tracing limited to passengers within 2-rows of the infectious individual(s) could fail to detect other cases of infections. This has important consequences for assessing the spread of infectious diseases. CONCLUSIONS: Infection at a distance from the index case indicates other factors, such as airflow, movement of passenger/crew members, fomites and contacts between passengers in the departure gate before boarding, or after deplaning, are involved.

Measuring social contacts in the emergency department.

BACKGROUND: Infectious individuals in an emergency department (ED) bring substantial risks of cross infection. Data about the complex social and spatial structure of interpersonal contacts in the ED will aid construction of biologically plausible transmission risk models that can guide cross infection control. METHODS AND FINDINGS: We sought to determine the number and duration of contacts among patients and staff in a large, busy ED. This prospective study was conducted between 1 July 2009 and 30 June 2010. Two 12-hour shifts per week were randomly selected for study. The study was conducted in the ED of an urban hospital. There were 81 shifts in the planned random sample of 104 (78%) with usable contact data, during which there were 9183 patient encounters. Of these, 6062 (66%) were approached to participate, of which 4732 (78%) agreed. Over the course of the year, 88 staff members participated (84%). A radiofrequency identification (RFID) system was installed and the ED divided into 89 distinct zones structured so copresence of two individuals in any zone implied a very high probability of contact <1 meter apart in space. During study observation periods, patients and staff were given RFID tags to wear. Contact events were recorded. These were further broken down with respect to the nature of the contacts, i.e., patient with patient, patient with staff, and staff with staff. 293,171 contact events were recorded, with a median of 22 contact events and 9 contacts with distinct individuals per participant per shift. Staff-staff interactions were more numerous and longer than patient-patient or patient-staff interactions. CONCLUSIONS: We used RFID to quantify contacts between patients and staff in a busy ED. These results are useful for studies of the spread of infections. By understanding contact patterns most important in potential transmission, more effective prevention strategies may be implemented.