Impact of the geriatric emergency medicine specialist intervention on final emergency department disposition.

BACKGROUND: The Geriatric Emergency Medicine Specialist (GEMS) pilot program is an innovative approach that utilizes geriatric-trained advanced practice providers to facilitate geriatric assessments and care planning for older adults in the emergency department (ED). The objective of this study was to explore the effect of GEMS on the use of observation status and final ED disposition. METHODS: This was a retrospective study under a target trial emulation framework. Geriatric patients (65+ years old) who presented to two ED sites within a large regional healthcare system between December 2020 and December 2022 were included. The primary outcome was final ED disposition (discharge, hospital inpatient admission, or hospital observation admission). Secondary outcomes included ED observation and ED length of stay. Non-GEMS patients were propensity score matched 5:1 to GEMS patients. Doubly robust regression was used to estimate the odds ratios and 95% confidence intervals of inpatient admission, discharge, hospital observation admission, ED observation admission, and estimate the mean ED length of stay. RESULTS: A total of 427 of 43,064 total patients (1.0%) received a GEMS intervention during the study period. Our analysis included 2,302 geriatric ED patients (410 GEMS, 1,892 non-GEMS) after propensity score matching. Hospital admission rates were 34.1% for GEMS compared to 56.4% for conventional treatment. GEMS patients had decreased odds of inpatient admission (OR: 0.41, 95 CI: 0.34-0.51, p < 0.001), increased odds of discharge (OR: 1.19 95 CI: 1.00-1.42, p = 0.047), hospital observation admission (OR: 2.97, 95 CI: 2.35-3.75, p < 0.001), ED observation admission (OR: 4.84 95 CI: 3.67-6.38, p < 0.001), and had a longer average ED length of stay (170 min, 95 CI: 84.6-256, p < 0.001) compared to non-GEMS patients. CONCLUSIONS: Patients seen by GEMS during their ED visit were associated with higher rates of hospital discharge and lower rates of hospital admissions.

Formative evaluation of an emergency department clinical decision support system for agitation symptoms: a study protocol.

INTRODUCTION: The burden of mental health-related visits to emergency departments (EDs) is growing, and agitation episodes are prevalent with such visits. Best practice guidance from experts recommends early assessment of at-risk populations and pre-emptive intervention using de-escalation techniques to prevent agitation. Time pressure, fluctuating work demands, and other systems-related factors pose challenges to efficient decision-making and adoption of best practice recommendations during an unfolding behavioural crisis. As such, we propose to design, develop and evaluate a computerised clinical decision support (CDS) system, Early Detection and Treatment to Reduce Events with Agitation Tool (ED-TREAT). We aim to identify patients at risk of agitation and guide ED clinicians through appropriate risk assessment and timely interventions to prevent agitation with a goal of minimising restraint use and improving patient experience and outcomes. METHODS AND ANALYSIS: This study describes the formative evaluation of the health record embedded CDS tool. Under aim 1, the study will collect qualitative data to design and develop ED-TREAT using a contextual design approach and an iterative user-centred design process. Participants will include potential CDS users, that is, ED physicians, nurses, technicians, as well as patients with lived experience of restraint use for behavioural crisis management during an ED visit. We will use purposive sampling to ensure the full spectrum of perspectives until we reach thematic saturation. Next, under aim 2, the study will conduct a pilot, randomised controlled trial of ED-TREAT at two adult ED sites in a regional health system in the Northeast USA to evaluate the feasibility, fidelity and bedside acceptability of ED-TREAT. We aim to recruit a total of at least 26 eligible subjects under the pilot trial. ETHICS AND DISSEMINATION: Ethical approval by the Yale University Human Investigation Committee was obtained in 2021 (HIC# 2000030893 and 2000030906). All participants will provide informed verbal consent prior to being enrolled in the study. Results will be disseminated through publications in open-access, peer-reviewed journals, via scientific presentations or through direct email notifications. TRIAL REGISTRATION NUMBER: NCT04959279; Pre-results.

Correction: How Does ChatGPT Perform on the United States Medical Licensing Examination (USMLE)? The Implications of Large Language Models for Medical Education and Knowledge Assessment.

[This corrects the article DOI: 10.2196/45312.].

Predicting relations between SOAP note sections: The value of incorporating a clinical information model.

Physician progress notes are frequently organized into Subjective, Objective, Assessment, and Plan (SOAP) sections. The Assessment section synthesizes information recorded in the Subjective and Objective sections, and the Plan section documents tests and treatments to narrow the differential diagnosis and manage symptoms. Classifying the relationship between the Assessment and Plan sections has been suggested to provide valuable insight into clinical reasoning. In this work, we use a novel human-in-the-loop pipeline to classify the relationships between the Assessment and Plan sections of SOAP notes as a part of the n2c2 2022 Track 3 Challenge. In particular, we use a clinical information model constructed from both the entailment logic expected from the aforementioned Challenge and the problem-oriented medical record. This information model is used to label named entities as primary and secondary problems/symptoms, events and complications in all four SOAP sections. We iteratively train separate Named Entity Recognition models and use them to annotate entities in all notes/sections. We fine-tune a downstream RoBERTa-large model to classify the Assessment-Plan relationship. We evaluate multiple language model architectures, preprocessing parameters, and methods of knowledge integration, achieving a maximum macro-F1 score of 82.31%. Our initial model achieves top-2 performance during the challenge (macro-F1: 81.52%, competitors' macro-F1 range: 74.54%-82.12%). We improved our model by incorporating post-challenge annotations (S&O sections), outperforming the top model from the Challenge. We also used Shapley additive explanations to investigate the extent of language model clinical logic, under the lens of our clinical information model. We find that the model often uses shallow heuristics and nonspecific attention when making predictions, suggesting language model knowledge integration requires further research.

How Does ChatGPT Perform on the United States Medical Licensing Examination (USMLE)? The Implications of Large Language Models for Medical Education and Knowledge Assessment.

BACKGROUND: Chat Generative Pre-trained Transformer (ChatGPT) is a 175-billion-parameter natural language processing model that can generate conversation-style responses to user input. OBJECTIVE: This study aimed to evaluate the performance of ChatGPT on questions within the scope of the United States Medical Licensing Examination (USMLE) Step 1 and Step 2 exams, as well as to analyze responses for user interpretability. METHODS: We used 2 sets of multiple-choice questions to evaluate ChatGPT's performance, each with questions pertaining to Step 1 and Step 2. The first set was derived from AMBOSS, a commonly used question bank for medical students, which also provides statistics on question difficulty and the performance on an exam relative to the user base. The second set was the National Board of Medical Examiners (NBME) free 120 questions. ChatGPT's performance was compared to 2 other large language models, GPT-3 and InstructGPT. The text output of each ChatGPT response was evaluated across 3 qualitative metrics: logical justification of the answer selected, presence of information internal to the question, and presence of information external to the question. RESULTS: Of the 4 data sets, AMBOSS-Step1, AMBOSS-Step2, NBME-Free-Step1, and NBME-Free-Step2, ChatGPT achieved accuracies of 44% (44/100), 42% (42/100), 64.4% (56/87), and 57.8% (59/102), respectively. ChatGPT outperformed InstructGPT by 8.15% on average across all data sets, and GPT-3 performed similarly to random chance. The model demonstrated a significant decrease in performance as question difficulty increased (P=.01) within the AMBOSS-Step1 data set. We found that logical justification for ChatGPT's answer selection was present in 100% of outputs of the NBME data sets. Internal information to the question was present in 96.8% (183/189) of all questions. The presence of information external to the question was 44.5% and 27% lower for incorrect answers relative to correct answers on the NBME-Free-Step1 (P<.001) and NBME-Free-Step2 (P=.001) data sets, respectively. CONCLUSIONS: ChatGPT marks a significant improvement in natural language processing models on the tasks of medical question answering. By performing at a greater than 60% threshold on the NBME-Free-Step-1 data set, we show that the model achieves the equivalent of a passing score for a third-year medical student. Additionally, we highlight ChatGPT's capacity to provide logic and informational context across the majority of answers. These facts taken together make a compelling case for the potential applications of ChatGPT as an interactive medical education tool to support learning.

Mapping emergency department asthma visits to identify poor-quality housing in New Haven, CT, USA: a retrospective cohort study.

BACKGROUND: Housing conditions are a key driver of asthma incidence and severity. Previous studies have shown increased emergency department visits for asthma among residents living in poor-quality housing. Interventions to improve housing conditions have been shown to reduce emergency department visits for asthma, but identification and remediation of poor housing conditions is often delayed or does not occur. This study evaluates whether emergency department visits for asthma can be used to identify poor-quality housing to support proactive and early intervention. METHODS: We conducted a retrospective cohort study of children and adults living in and around New Haven, CT, USA, who were seen for asthma in an urban, tertiary emergency department between March 1, 2013, and Aug 31, 2017. We geocoded and mapped patient addresses to city parcels, and calculated a composite estimate of the incidence of emergency department use for asthma for each parcel (Nv × Np/log2[P], where Nv is the estimated mean number of visits per patient, Np is the number of patients, and P is the estimated population). To determine whether parcel-level emergency department use for asthma was associated with public housing inspection scores, we used regression analyses, adjusting for neighbourhood-level and individual-level factors contributing to emergency department use for asthma. Public housing complex inspection scores were obtained from standardised home inspections, which are conducted every 1-3 years for publicly funded housing. We used a sliding-window approach to estimate how far in advance of a failed inspection the model could identify elevated use of emergency departments for asthma, using the city-wide 90th percentile as a cutoff for elevated incidence. FINDINGS: 11 429 asthma-related emergency department visits from 6366 unique patients were included in the analysis. Mean patient age was 32·4 years (SD 12·8); 3836 (60·3%) patients were female, 2530 (39·7%) were male, 3461 (57·2%) were Medicaid-insured, and 2651 (41·6%) were Black. Incidence of emergency department use for asthma was strongly correlated with lower housing inspection scores (Pearson's r=-0·55 [95% CI -0·70 to -0·35], p=3·5 × 10-6), and this correlation persisted after adjustment for patient-level and neighbourhood-level demographics using a linear regression model (r=-0·54 [-0·69 to -0·33], p=7·1 × 10-6) and non-linear regression model (r=-0·44 [-0·62 to -0·21], p=3·8 × 10-4). Elevated asthma incidence rates were typically detected around a year before a housing complex failed a housing inspection. INTERPRETATION: Emergency department visits for asthma are an early indicator of failed housing inspections. This approach represents a novel method for the early identification of poor housing conditions and could help to reduce asthma-related morbidity and mortality. FUNDING: Harvard-National Institute of Environmental Health Sciences (NIEHS) Center for Environmental Health.

Patterns of Care Partner Communication for Persons Living with Dementia in the Emergency Department.

Introduction: Nearly half of all persons living with dementia (PLwD) will visit the emergency department (ED) in any given year and ED visits by PLwD are associated with short-term adverse outcomes. Care partner engagement is critical in the care of PLwD, but little is known about their patterns of communication with ED clinicians. Methods: We performed a retrospective electronic health record (EHR) review of a random sampling of patients ≥ 65 years with a historical diagnosis code of dementia who visited an ED within a large regional health network between 1/2014 and 1/2022. ED notes within the EHRs were coded for documentation of care partner communication and presence of a care partner in the ED. Logistic regression was used to identify patient characteristics associated with the composite outcome of either care partner communication or care partner presence in the ED. Results: A total of 460 patients were included. The median age was 83.0 years, 59.3% were female, 11.3% were Black, and 7.6% Hispanic. A care partner was documented in the ED for 22.4% of the visits and care partner communication documented for 43.9% of visits. 54.8% of patients had no documentation of care partner communication nor evidence of a care partner at the bedside. In multivariate logistic regression, increasing age (OR, (95% CI): 1.06 (1.04-1.09)), altered mental status (OR: 2.26 (1.01-5.05)), and weakness (OR: 3.38 (1.49-7.65)) significantly increased the probability of having care partner communication documented or a care partner at the bedside. Conclusion: More than half of PLwD in our sample did not have clinician documentation of communication with a care partner or a care partner in the ED. Further studies are needed to use these insights to improve communication with care partners of PLwD in the ED.

Incorporating Domain Knowledge Into Language Models by Using Graph Convolutional Networks for Assessing Semantic Textual Similarity: Model Development and Performance Comparison.

BACKGROUND: Although electronic health record systems have facilitated clinical documentation in health care, they have also introduced new challenges, such as the proliferation of redundant information through the use of copy and paste commands or templates. One approach to trimming down bloated clinical documentation and improving clinical summarization is to identify highly similar text snippets with the goal of removing such text. OBJECTIVE: We developed a natural language processing system for the task of assessing clinical semantic textual similarity. The system assigns scores to pairs of clinical text snippets based on their clinical semantic similarity. METHODS: We leveraged recent advances in natural language processing and graph representation learning to create a model that combines linguistic and domain knowledge information from the MedSTS data set to assess clinical semantic textual similarity. We used bidirectional encoder representation from transformers (BERT)-based models as text encoders for the sentence pairs in the data set and graph convolutional networks (GCNs) as graph encoders for corresponding concept graphs that were constructed based on the sentences. We also explored techniques, including data augmentation, ensembling, and knowledge distillation, to improve the model's performance, as measured by the Pearson correlation coefficient (r). RESULTS: Fine-tuning the BERT_base and ClinicalBERT models on the MedSTS data set provided a strong baseline (Pearson correlation coefficients: 0.842 and 0.848, respectively) compared to those of the previous year's submissions. Our data augmentation techniques yielded moderate gains in performance, and adding a GCN-based graph encoder to incorporate the concept graphs also boosted performance, especially when the node features were initialized with pretrained knowledge graph embeddings of the concepts (r=0.868). As expected, ensembling improved performance, and performing multisource ensembling by using different language model variants, conducting knowledge distillation with the multisource ensemble model, and taking a final ensemble of the distilled models further improved the system's performance (Pearson correlation coefficients: 0.875, 0.878, and 0.882, respectively). CONCLUSIONS: This study presents a system for the MedSTS clinical semantic textual similarity benchmark task, which was created by combining BERT-based text encoders and GCN-based graph encoders in order to incorporate domain knowledge into the natural language processing pipeline. We also experimented with other techniques involving data augmentation, pretrained concept embeddings, ensembling, and knowledge distillation to further increase our system's performance. Although the task and its benchmark data set are in the early stages of development, this study, as well as the results of the competition, demonstrates the potential of modern language model-based systems to detect redundant information in clinical notes.

Clinical characteristics and outcomes for 7,995 patients with SARS-CoV-2 infection.

OBJECTIVE: Severe acute respiratory syndrome virus (SARS-CoV-2) has infected millions of people worldwide. Our goal was to identify risk factors associated with admission and disease severity in patients with SARS-CoV-2. DESIGN: This was an observational, retrospective study based on real-world data for 7,995 patients with SARS-CoV-2 from a clinical data repository. SETTING: Yale New Haven Health (YNHH) is a five-hospital academic health system serving a diverse patient population with community and teaching facilities in both urban and suburban areas. POPULATIONS: The study included adult patients who had SARS-CoV-2 testing at YNHH between March 1 and April 30, 2020. MAIN OUTCOME AND PERFORMANCE MEASURES: Primary outcomes were admission and in-hospital mortality for patients with SARS-CoV-2 infection as determined by RT-PCR testing. We also assessed features associated with the need for respiratory support. RESULTS: Of the 28605 patients tested for SARS-CoV-2, 7995 patients (27.9%) had an infection (median age 52.3 years) and 2154 (26.9%) of these had an associated admission (median age 66.2 years). Of admitted patients, 2152 (99.9%) had a discharge disposition at the end of the study period. Of these, 329 (15.3%) required invasive mechanical ventilation and 305 (14.2%) expired. Increased age and male sex were positively associated with admission and in-hospital mortality (median age 80.7 years), while comorbidities had a much weaker association with the risk of admission or mortality. Black race (OR 1.43, 95%CI 1.14-1.78) and Hispanic ethnicity (OR 1.81, 95%CI 1.50-2.18) were identified as risk factors for admission, but, among discharged patients, age-adjusted in-hospital mortality was not significantly different among racial and ethnic groups. CONCLUSIONS: This observational study identified, among people testing positive for SARS-CoV-2 infection, older age and male sex as the most strongly associated risks for admission and in-hospital mortality in patients with SARS-CoV-2 infection. While minority racial and ethnic groups had increased burden of disease and risk of admission, age-adjusted in-hospital mortality for discharged patients was not significantly different among racial and ethnic groups. Ongoing studies will be needed to continue to evaluate these risks, particularly in the setting of evolving treatment guidelines.

Clinical Characteristics and Outcomes for 7,995 Patients with SARS-CoV-2 Infection.

OBJECTIVE: Severe acute respiratory syndrome virus (SARS-CoV-2) has infected millions of people worldwide. Our goal was to identify risk factors associated with admission and disease severity in patients with SARS-CoV-2. DESIGN: This was an observational, retrospective study based on real-world data for 7,995 patients with SARS-CoV-2 from a clinical data repository. SETTING: Yale New Haven Health (YNHH) is a five-hospital academic health system serving a diverse patient population with community and teaching facilities in both urban and suburban areas. POPULATIONS: The study included adult patients who had SARS-CoV-2 testing at YNHH between March 1 and April 30, 2020. MAIN OUTCOME AND PERFORMANCE MEASURES: Primary outcomes were admission and in-hospital mortality for patients with SARS-CoV-2 infection as determined by RT-PCR testing. We also assessed features associated with the need for respiratory support. RESULTS: Of the 28605 patients tested for SARS-CoV-2, 7995 patients (27.9%) had an infection (median age 52.3 years) and 2154 (26.9%) of these had an associated admission (median age 66.2 years). Of admitted patients, 2152 (99.9%) had a discharge disposition at the end of the study period. Of these, 329 (15.3%) required invasive mechanical ventilation and 305 (14.2%) expired. Increased age and male sex were positively associated with admission and in-hospital mortality (median age 80.7 years), while comorbidities had a much weaker association with the risk of admission or mortality. Black race (OR 1.43, 95%CI 1.14-1.78) and Hispanic ethnicity (OR 1.81, 95%CI 1.50-2.18) were identified as risk factors for admission, but, among discharged patients, age-adjusted in-hospital mortality was not significantly different among racial and ethnic groups. CONCLUSIONS: This observational study identified, among people testing positive for SARSCoV-2 infection, older age and male sex as the most strongly associated risks for admission and in-hospital mortality in patients with SARS-CoV-2 infection. While minority racial and ethnic groups had increased burden of disease and risk of admission, age-adjusted in-hospital mortality for discharged patients was not significantly different among racial and ethnic groups. Ongoing studies will be needed to continue to evaluate these risks, particularly in the setting of evolving treatment guidelines.

Generating contextual embeddings for emergency department chief complaints.

OBJECTIVE: We learn contextual embeddings for emergency department (ED) chief complaints using Bidirectional Encoder Representations from Transformers (BERT), a state-of-the-art language model, to derive a compact and computationally useful representation for free-text chief complaints. MATERIALS AND METHODS: Retrospective data on 2.1 million adult and pediatric ED visits was obtained from a large healthcare system covering the period of March 2013 to July 2019. A total of 355 497 (16.4%) visits from 65 737 (8.9%) patients were removed for absence of either a structured or unstructured chief complaint. To ensure adequate training set size, chief complaint labels that comprised less than 0.01%, or 1 in 10 000, of all visits were excluded. The cutoff threshold was incremented on a log scale to create seven datasets of decreasing sparsity. The classification task was to predict the provider-assigned label from the free-text chief complaint using BERT, with Long Short-Term Memory (LSTM) and Embeddings from Language Models (ELMo) as baselines. Performance was measured as the Top-k accuracy from k = 1:5 on a hold-out test set comprising 5% of the samples. The embedding for each free-text chief complaint was extracted as the final 768-dimensional layer of the BERT model and visualized using t-distributed stochastic neighbor embedding (t-SNE). RESULTS: The models achieved increasing performance with datasets of decreasing sparsity, with BERT outperforming both LSTM and ELMo. The BERT model yielded Top-1 accuracies of 0.65 and 0.69, Top-3 accuracies of 0.87 and 0.90, and Top-5 accuracies of 0.92 and 0.94 on datasets comprised of 434 and 188 labels, respectively. Visualization using t-SNE mapped the learned embeddings in a clinically meaningful way, with related concepts embedded close to each other and broader types of chief complaints clustered together. DISCUSSION: Despite the inherent noise in the chief complaint label space, the model was able to learn a rich representation of chief complaints and generate reasonable predictions of their labels. The learned embeddings accurately predict provider-assigned chief complaint labels and map semantically similar chief complaints to nearby points in vector space. CONCLUSION: Such a model may be used to automatically map free-text chief complaints to structured fields and to assist the development of a standardized, data-driven ontology of chief complaints for healthcare institutions.

Patient factors associated with SARS-CoV-2 in an admitted emergency department population.

Background: The SARS-CoV-2 (COVID-19) virus has wide community spread. The aim of this study was to describe patient characteristics and to identify factors associated with COVID-19 among emergency department (ED) patients under investigation for COVID-19 who were admitted to the hospital. Methods: This was a retrospective observational study from 8 EDs within a 9-hospital health system. Patients with COVID-19 testing around the time of hospital admission were included. The primary outcome measure was COVID-19 test result. Patient characteristics were described and a multivariable logistic regression model was used to identify factors associated with a positive COVID-19 test. Results: During the study period from March 1, 2020 to April 8, 2020, 2182 admitted patients had a test resulted for COVID-19. Of these patients, 786 (36%) had a positive test result. For COVID-19-positive patients, 63 (8.1%) died during hospitalization. COVID-19-positive patients had lower pulse oximetry (0.91 [95% confidence interval, CI], [0.88-0.94]), higher temperatures (1.36 [1.26-1.47]), and lower leukocyte counts than negative patients (0.78 [0.75-0.82]). Chronic lung disease (odds ratio [OR] 0.68, [0.52-0.90]) and histories of alcohol (0.64 [0.42-0.99]) or substance abuse (0.39 [0.25-0.62]) were less likely to be associated with a positive COVID-19 result. Conclusion: We observed a high percentage of positive results among an admitted ED cohort under investigation for COVID-19. Patient factors may be useful in early differentiation of patients with COVID-19 from similarly presenting respiratory illnesses although no single factor will serve this purpose.

Synthesis of fracture radiographs with deep neural networks.

PURPOSE: We describe a machine learning system for converting diagrams of fractures into realistic X-ray images. We further present a method for iterative, human-guided refinement of the generated images and show that the resulting synthetic images can be used during training to increase the accuracy of deep classifiers on clinically meaningful subsets of fracture X-rays. METHODS: A neural network was trained to reconstruct images from programmatically created line drawings of those images. The images were then further refined with an optimization-based technique. Ten physicians were recruited into a study to assess the realism of synthetic radiographs created by the neural network. They were presented with mixed sets of real and synthetic images and asked to identify which images were synthetic. Two classifiers were trained to detect humeral shaft fractures: one only on true fracture images, and one on both true and synthetic images. RESULTS: Physicians were 49.63% accurate in identifying whether images were synthetic or real. This is close to what would be expected by pure chance (i.e. random guessing). A classifier trained only on real images detected fractures with 67.21% sensitivity when no fracture fixation hardware was present. A classifier trained on both real images and synthetic images was 75.54% sensitive. CONCLUSION: Our method generates X-rays realistic enough to be indistinguishable from real X-rays. We also show that synthetic images generated using this method can be used to increase the accuracy of deep classifiers on clinically meaningful subsets of fracture X-rays.

Hypertension and Acute Mountain Sickness in Himalayan Trekkers in Nepal: An Observational Cohort Study.

INTRODUCTION: A history of preexisting hypertension is common in people participating in mountain activities; however, the relationship between blood pressure (BP), preexisting hypertension, and acute mountain sickness (AMS) is not well studied. We sought to determine these relationships among trekkers in the Everest region of Nepal. METHODS: This was a prospective observational cohort study of a convenience sample of adult, nonpregnant volunteers trekking in the Everest Base Camp region in Nepal. We recorded Lake Louise Scores for AMS and measured BP at 2860 m, 3400 m, and 4300 m. The primary outcome was AMS. RESULTS: A total of 672 trekkers (including 60 with history of preexisting hypertension) were enrolled at 2860 m. We retained 529 at 3400 m and 363 at 4300 m. At 3400 m, 11% of participants had AMS, and 13% had AMS at 4300 m. We found no relationship between AMS and measured BP values (P>0.05), nor was there any relation of BP to AMS severity as measured by higher Lake Louise Scores (P>0.05). Preexisting hypertension (odds ratio [OR] 0.16; 95% CI 0.025-0.57), male sex (OR 0.59; 95% CI 0.37-0.96), and increased SpO2 (OR 0.93; 95% CI 0.87-0.98) were associated with reduced rates of AMS in multivariate analyses adjusting for known risk factors for AMS. CONCLUSIONS: AMS is common in trekkers in Nepal, even at 3400 m. There is no relationship between measured BP and AMS. However, a medical history of hypertension may be associated with a lower risk of AMS. More work is needed to confirm this novel finding.

Benchmark and Best Practices for Biomedical Knowledge Graph Embeddings.

Much of biomedical and healthcare data is encoded in discrete, symbolic form such as text and medical codes. There is a wealth of expert-curated biomedical domain knowledge stored in knowledge bases and ontologies, but the lack of reliable methods for learning knowledge representation has limited their usefulness in machine learning applications. While text-based representation learning has significantly improved in recent years through advances in natural language processing, attempts to learn biomedical concept embeddings so far have been lacking. A recent family of models called knowledge graph embeddings have shown promising results on general domain knowledge graphs, and we explore their capabilities in the biomedical domain. We train several state-of-the-art knowledge graph embedding models on the SNOMED-CT knowledge graph, provide a benchmark with comparison to existing methods and in-depth discussion on best practices, and make a case for the importance of leveraging the multi-relational nature of knowledge graphs for learning biomedical knowledge representation. The embeddings, code, and materials will be made available to the community.

Validation of a Machine Learning Model That Outperforms Clinical Risk Scoring Systems for Upper Gastrointestinal Bleeding.

BACKGROUND & AIMS: Scoring systems are suboptimal for determining risk in patients with upper gastrointestinal bleeding (UGIB); these might be improved by a machine learning model. We used machine learning to develop a model to calculate the risk of hospital-based intervention or death in patients with UGIB and compared its performance with other scoring systems. METHODS: We analyzed data collected from consecutive unselected patients with UGIB from medical centers in 4 countries (the United States, Scotland, England, and Denmark; n = 1958) from March 2014 through March 2015. We used the data to derive and internally validate a gradient-boosting machine learning model to identify patients who met a composite endpoint of hospital-based intervention (transfusion or hemostatic intervention) or death within 30 days. We compared the performance of the machine learning prediction model with validated pre-endoscopic clinical risk scoring systems (the Glasgow-Blatchford score [GBS], admission Rockall score, and AIMS65). We externally validated the machine learning model using data from 2 Asia-Pacific sites (Singapore and New Zealand; n = 399). Performance was measured by area under receiver operating characteristic curve (AUC) analysis. RESULTS: The machine learning model identified patients who met the composite endpoint with an AUC of 0.91 in the internal validation set; the clinical scoring systems identified patients who met the composite endpoint with AUC values of 0.88 for the GBS (P = .001), 0.73 for Rockall score (P < .001), and 0.78 for AIMS65 score (P < .001). In the external validation cohort, the machine learning model identified patients who met the composite endpoint with an AUC of 0.90, the GBS with an AUC of 0.87 (P = .004), the Rockall score with an AUC of 0.66 (P < .001), and the AIMS65 with an AUC of 0.64 (P < .001). At cutoff scores at which the machine learning model and GBS identified patients who met the composite endpoint with 100% sensitivity, the specificity values were 26% with the machine learning model versus 12% with GBS (P < .001). CONCLUSIONS: We developed a machine learning model that identifies patients with UGIB who met a composite endpoint of hospital-based intervention or death within 30 days with a greater AUC and higher levels of specificity, at 100% sensitivity, than validated clinical risk scoring systems. This model could increase identification of low-risk patients who can be safely discharged from the emergency department for outpatient management.