A History of the National Academy of Medicine: 50 Years of Transformational Leadership (2022)

This volume was originally intended to be published in 2020 to mark the occasion of the 50th anniversary of the founding of the Institute of Medicine (IOM). The COVID-19 pandemic slowed its completion, and, as the chapters that follow detail, led to lasting changes across the organization. The final volume describes events within and outside the organization through the end of 2021. The National Academy of Medicine's (NAM's) response to major events in 2022, such as Russia's invasion of Ukraine, racially motivated and other mass shootings in the United States, and the U.S. Supreme Court's decision to overturn Roe v. Wade, are covered in the Epilogue. © National Academy of Sciences. All rights reserved.

Uncoupling Inequality: Reflections on the Ethics of Benchmarks for Digital Media

Our collaboration seeks to demonstrate shared interrogation by exploring the ethics of machine learning benchmarks from a socio-technical management perspective with insight from public health and ethnic studies. Benchmarks, such as ImageNet, are annotated open data sets for training algorithms. The COVID-19 pandemic reinforced the practical need for ethical information infrastructures to analyze digital and social media, especially related to medicine and race. Social media analysis that obscures Black teen mental health and ignores anti-Asian hate fails as information infrastructure. Despite inadequately handling non-dominant voices, machine learning benchmarks are the basis for analysis in operational systems. Turning to the management literature, we interrogate cross-cutting problems of benchmarks through the lens of coupling, or mutual interdependence between people, technologies, and environments. Uncoupling inequality from machine learning benchmarks may require conceptualizing the social dependencies that build structural barriers to inclusion. © 2022 IEEE Computer Society. All rights reserved.

SPATIAL GRAPH ATTENTION AND CURIOSITY-DRIVEN POLICY FOR ANTIVIRAL DRUG DISCOVERY

We developed Distilled Graph Attention Policy Network (DGAPN), a reinforcement learning model to generate novel graph-structured chemical representations that optimize user-defined objectives by efficiently navigating a physically constrained domain. The framework is examined on the task of generating molecules that are designed to bind, noncovalently, to functional sites of SARS-CoV-2 proteins. We present a spatial Graph Attention (sGAT) mechanism that leverages self-attention over both node and edge attributes as well as encoding the spatial structure - this capability is of considerable interest in synthetic biology and drug discovery. An attentional policy network is introduced to learn the decision rules for a dynamic, fragment-based chemical environment, and state-of-the-art policy gradient techniques are employed to train the network with stability. Exploration is driven by the stochasticity of the action space design and the innovation reward bonuses learned and proposed by random network distillation. In experiments, our framework achieved outstanding results compared to state-of-the-art algorithms, while reducing the complexity of paths to chemical synthesis. © 2022 ICLR 2022 - 10th International Conference on Learning Representationss. All rights reserved.

Impact of a multifaceted, outpatient antimicrobial stewardship intervention bundle on unnecessary antimicrobial prescribing in upper respiratory tract infections (URI)

Background. URIs are the most common indication for outpatient antibiotic prescribing. Given high rates of unnecessary prescribing, these indications have been identified as a high-priority target for outpatient antimicrobial stewardship programs (ASP). Our primary objective was to evaluate the impact of a system-wide, multifaceted, outpatient ASP intervention bundle on unnecessary antibiotic prescribing for URI. Methods. This quasi-experimental study was conducted from 2019 to 2021. ICD-10 codes for URIs were grouped into 3 tiers (i.e., tier I = antibiotics always indicated, tier II = sometimes, tier III = never). Encounters from 5 care specialties (i.e., family medicine, community internal medicine, express care, pediatrics, and emergency department) with a tier III URI primary ICD-10 code but without a secondary tier I or tier II code were included. COVID-19 ICD-10 codes were excluded. Interventions included construction of a prescribing data model, dissemination of clinician prescribing data and education, promotion of symptom management strategies, a patient-facing commitment poster, and a pre-populated URI order panel. Tools were designed at a system level and implemented by regional champions beginning in the 3rd quarter of 2020. The primary outcome was the rate of antibiotic prescribing, and the secondary outcome and counterbalance measure was the rate of repeat URI-related healthcare contact within 14 days. Outcomes were analyzed with chi-square with an alpha level of 0.05. Results. A total of 147403 encounters were included. The overall antibiotic prescribing rate decreased from 24.1% to 12.3% between 2019 and 2021 (p< 0.01). Significant reductions in tier III antibiotic prescribing were demonstrated for each region, care specialty, and syndrome evaluated (Table 1). A reduction in repeat healthcare contact was seen across the total cohort (9.5% in 2019 vs. 8.3% in 2021, p< 0.01);decreases in repeat contact rates were observed in those not initially receiving an antibiotic (10.3% vs. 8.6%, p< 0.01), but not in those who initially received an antibiotic (6.8% vs. 6.8%, p = 0.94). Tier III URI encounter level antimicrobial prescribing rates by region, care specialty, and syndrome Conclusion. A multifaceted, outpatient ASP intervention bundle decreased rates of unnecessary antimicrobial prescribing without increasing rates of 14-day repeat URI-related healthcare contact.

Gambling in Canada During the Pandemic: 6 Months After the National COVID Lockdown

Objective: To investigate the impact of the coronavirus disease (COVID) pandemic on gambling and problem gambling in Canada. Method(s): A stratified national cohort of regular gamblers (n=2790) completed a comprehensive online questionnaire 6 months before the onset of the pandemic (baseline;August-November 2019), during the nation-wide lockdown (May-June 2020), and 6 months after the lockdown (December 2020). Result(s): Significant decreases in gambling frequency, time spent in gambling sessions, money spent, and the number of game types played occurred during lockdown followed by significant increases in all of these same measures postlockdown. However, the level of postlockdown gambling behavior was still significantly lower than prepandemic baseline levels. A significant shift to online gambling was also observed during the lockdown, which persisted postlockdown. Problem gambling scores also declined during the lockdown, with no significant change in these scores postlockdown. Consistent with previous literature, it was found that a higher level of gambling engagement, online gambling participation, and known biopsychosocial factors (family history of problem gambling, gambling fallacies, substance use, male sex, mood disorder) were predictors of problem gambling at postlockdown. COVID-specific factors predictive of problem gambling postlockdown were the presence of negative COVID-related health impacts and increased levels of social interaction and leisure time. Conclusion(s): The COVID-related lockdowns have resulted in lower levels of gambling and problem gambling symptomatology as well as increased online gambling in Canada that have persisted to some extent 6 months postlockdown. Future studies are necessary to ascertain whether these represent permanent or temporary changes. Copyright © 2022 Lippincott Williams and Wilkins. All rights reserved.

The impact of forced abstinence on gambling: A natural ABA design study

The impacts of the COVID-19 pandemic, and responses used to mitigate the spread such as selective closure of non-essential businesses, have been far-reaching. Some of these impacts include changes in health, economic, social and recreation. Included among other non-essential business, in-person gambling venues were closed across Canada. Yet, online gambling opportunities remained available, making this period both a historical first in Canada, and a natural experiment. The current study examined quantifiable ramifications of the sudden forced abstinence from in-person gambling during the nation-wide lockdown in Canada, and what changes occurred six-months later upon reopening. For this cohort study, pre-pandemic base line data was provided six-month before the lockdown by online panel participants (n = 2,790), who were then re-surveyed during the national lockdown and again six-months postlockdown. Nearly one-third of gamblers reported a complete cessation of gambling during the lockdown period. For those who continued gambling, quantitative data indicated signifi-cant decreases on all gambling engagement measures: frequency, time spent in gambling sessions, money spent, and the number of game types played. This was followed by significant increases on each engagement measure six-months post-lockdown. Although these increases did not return to pre-pandemic engagement levels. Problem gambling within the whole sample generally declined during lockdown, however, significant increases in highrisk gambling were evidenced six-months later. In fact, engaging in online gambling and COVID-specific changes in health, employment, and social isolation across the closure and re-opening periods were independent predictors for classification as a problem gambler sixmonths after the lockdown.

MetaMap versus BERT models with explainable active learning: Ontology-based experiments with prior knowledge for COVID-19

Emergence of the Coronavirus 2019 Disease has highlighted further the need for timely support for clinicians as they manage severely ill patients. We combine Semantic Web technologies with Deep Learning for Natural Language Processing with the aim of converting human-readable best evidence/ practice for COVID-19 into that which is computer-interpretable. We present the results of experiments with 1212 clinical ideas (medical terms and expressions) from two UK national healthcare services specialty guides for COVID-19 and three versions of two BMJ Best Practice documents for COVID-19. The paper seeks to recognise and categorise clinical ideas, performing a Named Entity Recognition (NER) task, with an ontology providing extra terms as context and describing the intended meaning of categories understandable by clinicians. The paper investigates: 1) the performance of classical NER using MetaMap versus NER with fine-tuned BERT models;2) the integration of both NER approaches using a lightweight ontology developed in close collaboration with senior doctors;and 3) the easy interpretation by junior doctors of the main classes from the ontology once populated with NER results. We report the NER performance and the observed agreement for human audits. Copyright © 2022 for this paper by its authors.

The Evolution of Empiric Antimicrobial Prescribing Patterns and Patient-Associated Antimicrobial Risk Factors over Time in COVID-19

Introduction: Despite low rates of bacterial co-infection in patients admitted with COVID-19, antibiotics are frequently prescribed in acute care settings. Antimicrobial stewardship program (ASP) efforts have evolved during the progression of the COVID-19 pandemic. We sought to evaluate the overall antimicrobial prescribing rate in patients with COVID-19, as well as assess changes to these patterns over time. Research Question or Hypothesis: What factors are associated with increased empiric antibiotic prescribing in COVID-19, and what is the impact of ASPs on prescribing rates? Study Design: Retrospective chart review of patients admit to a tertiary care center with symptomatic COVID-19 between March 1st, 2020 and November 30th, 2020. Methods: Symptomatic adults admitted with a positive SARS-CoV-2 polymerase chain reaction test were included for review and stratified by disease severity. Patient and provider demographics, antimicrobial utilization, and culture data were collected. Poisson regression was used to assess changes in antimicrobial prescribing over time. Logistic regression was used to assess factors associated with the empiric use of antimicrobial agents among patients without an existing positive bacterial respiratory culture. Results: 654 patients were included for review;189 with mild, 242 with moderate, and 223 with severe COVID-19. Antibiotics were prescribed in 37.9% of the cohort, with an increased incidence by disease severity (16.9% mild, 29.8% moderate, and 64.6% severe, p < 0.001). 85.1% of antibiotics administered were prescribed within 48 hours of hospital admission. Over the course of the study, antimicrobial prescribing rates decreased by 8.7% per month despite a concurrent increase in COVID-19 admissions. Multivariate analysis found that ICU admission, obtainment of procalcitonin, intubation, heart failure, hemodialysis, and nursing home residence were associated with empiric antimicrobial prescribing. Conclusion: ASPs should take an active role on intervening in unnecessary antimicrobial use targeting populations most at risk for unnecessary exposure. The application of ASP techniques appear to impact antimicrobial use trends even during the COVID-19 pandemic.

Named Entity Recognition and Relation Extraction for COVID-19: Explainable Active Learning with Word2vec Embeddings and Transformer-Based BERT Models

Deep learning for natural language processing acquires dense vector representations for n-grams from large-scale unstructured corpora. Converting static embeddings of n-grams into a dataset of interlinked concepts with explicit contextual semantic dependencies provides the foundation to acquire reusable knowledge. However, the validation of this knowledge requires cross-checking with ground-truths that may be unavailable in an actionable or computable form. This paper presents a novel approach from the new field of explainable active learning that combines methods for learning static embeddings (word2vec models) with methods for learning dynamic contextual embeddings (transformer-based BERT models). We created a dataset for named entity recognition (NER) and relation extraction (REX) for the Coronavirus Disease 2019 (COVID-19). The COVID-19 dataset has 2,212 associations captured by 11 word2vec models with additional examples of use from the biomedical literature. We propose interpreting the NER and REX tasks for COVID-19 as Question Answering (QA) incorporating general medical knowledge within the question, e.g. “does ‘cough’ (n-gram) belong to ‘clinical presentation/symptoms’ for COVID-19?”. We evaluated biomedical-specific pre-trained language models (BioBERT, SciBERT, ClinicalBERT, BlueBERT, and PubMedBERT) versus general-domain pre-trained language models (BERT, and RoBERTa) for transfer learning with COVID-19 dataset, i.e. task-specific fine-tuning considering NER as a sequence-level task. Using 2,060 QA for training (associations from 10 word2vec models) and 152 QA for validation (associations from 1 word2vec model), BERT obtained an F-measure of 87.38%, with precision = 93.75% and recall = 81.82%. SciBERT achieved the highest F-measure of 94.34%, with precision = 98.04% and recall = 90.91%. © 2021, Springer Nature Switzerland AG.

"Not a perfect situation, but..." A single-practice survey of patient experience of phone consultations during COVID-19 Alert Level 4 in New Zealand

AIM: To explore patients' experiences of virtual consultations during the COVID-19 Alert Level 4 lockdown in New Zealand. METHOD: A single-practice retrospective phone survey exploring patients' satisfaction with the phone consultation process during Alert Level 4 lockdown. RESULTS: Of 259 eligible patients, 108 (42%) participated in the survey. Overall satisfaction with phone consultations was high, with a median score 9 out of 10 (95% CI 9-9). Participants were highly likely to recommend phone consultations to others, with a median score of 9 (95% CI 7-9). This was consistent across age groups, ethnicities and socioeconomic groupings. Men were less satisfied with phone consultations than women, with a 2 point (95% CI -3--1) lower median score than women, but they were not less likely to recommend phone consultations. Most participants found phone consultations to be convenient and time-saving and considered not seeing the doctor to be acceptable in the context of the lockdown. Few participants experienced technical difficulties over the phone. Issues of communication and appropriateness of consultations to the medium of the phone were raised. CONCLUSION: This single-centre study demonstrates the acceptability of phone consults for most patients presenting to general practice during a pandemic. These findings need further exploration in broader general practice settings and non-pandemic contexts.

IMPECCABLE: Integrated Modeling PipelinE for COVID Cure by Assessing Better LEads

The drug discovery process currently employed in the pharmaceutical industry typically requires about 10 years and $2-3 billion to deliver one new drug. This is both too expensive and too slow, especially in emergencies like the COVID-19 pandemic. In silico methodologies need to be improved both to select better lead compounds, so as to improve the efficiency of later stages in the drug discovery protocol, and to identify those lead compounds more quickly. No known methodological approach can deliver this combination of higher quality and speed. Here, we describe an Integrated Modeling PipEline for COVID Cure by Assessing Better LEads (IMPECCABLE) that employs multiple methodological innovations to overcome this fundamental limitation. We also describe the computational framework that we have developed to support these innovations at scale, and characterize the performance of this framework in terms of throughput, peak performance, and scientific results. We show that individual workflow components deliver 100 × to 1000 × improvement over traditional methods, and that the integration of methods, supported by scalable infrastructure, speeds up drug discovery by orders of magnitudes. IMPECCABLE has screened ∼1011 ligands and has been used to discover a promising drug candidate. These capabilities have been used by the US DOE National Virtual Biotechnology Laboratory and the EU Centre of Excellence in Computational Biomedicine. © 2021 ACM.

Scalable HPC & AI infrastructure for COVID-19 therapeutics

COVID-19 has claimed more than 2.7 × 106 lives and resulted in over 124 × 106 infections. There is an urgent need to identify drugs that can inhibit SARS-CoV-2. We discuss innovations in computational infrastructure and methods that are accelerating and advancing drug design. Specifically, we describe several methods that integrate artificial intelligence and simulation-based approaches, and the design of computational infrastructure to support these methods at scale. We discuss their implementation, characterize their performance, and highlight science advances that these capabilities have enabled. © 2021 ACM.

Changes to management of hypertension in pregnancy, and attitudes to self-management: An online survey of obstetricians, before and following the first wave of the COVID-19 pandemic.

OBJECTIVE: This study aimed to understand the views and practice of obstetricians regarding self-monitoring for hypertensive disorders of pregnancy (blood pressure (BP) and proteinuria), the potential for self-management (including actions taken on self-monitored parameters) and to understand the impact of the COVID-19 pandemic on such views. DESIGN: Cross-sectional online survey pre- and post- the first wave of the COVID-19 pandemic. SETTING AND SAMPLE: UK obstetricians recruited via an online portal. METHODS: A survey undertaken in two rounds: December 2019-January 2020 (pre-pandemic), and September-November 2020 (during pandemic) RESULTS: 251 responses were received across rounds one (150) and two (101). Most obstetricians considered that self-monitoring of BP and home urinalysis had a role in guiding clinical decisions and this increased significantly following the first wave of the COVID-19 pandemic (88%, (132/150) 95%CI: 83-93% first round vs 96% (95%CI: 92-94%), (97/101), second round; p = 0.039). Following the pandemic, nearly half were agreeable to women self-managing their hypertension by using their own readings to make a pre-agreed medication change themselves (47%, 47/101 (95%CI: 37-57%)). CONCLUSIONS: A substantial majority of UK obstetricians considered that self-monitoring had a role in the management of pregnancy hypertension and this increased following the pandemic. Around half are now supportive of women having a wider role in self-management of hypertensive treatment. Maximising the potential of such changes in pregnancy hypertension management requires further work to understand how to fully integrate women's own measurements into clinical care.

Recommendations for accurate genotyping of SARS-CoV-2 using amplicon-based sequencing of clinical samples

OBJECTIVES: Genotyping of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has been instrumental in monitoring viral evolution and transmission during the pandemic. The quality of the sequence data obtained from these genotyping efforts depends on several factors, including the quantity/integrity of the input material, the technology, and laboratory-specific implementation. The current lack of guidelines for SARS-CoV-2 genotyping leads to inclusion of error-containing genome sequences in genomic epidemiology studies. We aimed to establish clear and broadly applicable recommendations for reliable virus genotyping. METHODS: We established and used a sequencing data analysis workflow that reliably identifies and removes technical artefacts;such artefacts can result in miscalls when using alternative pipelines to process clinical samples and synthetic viral genomes with an amplicon-based genotyping approach. We evaluated the impact of experimental factors, including viral load and sequencing depth, on correct sequence determination. RESULTS: We found that at least 1000 viral genomes are necessary to confidently detect variants in the SARS-CoV-2 genome at frequencies of >=10%. The broad applicability of our recommendations was validated in over 200 clinical samples from six independent laboratories. The genotypes we determined for clinical isolates with sufficient quality cluster by sampling location and period. Our analysis also supports the rise in frequencies of 20A.EU1 and 20A.EU2, two recently reported European strains whose dissemination was facilitated by travel during the summer of 2020. CONCLUSIONS: We present much-needed recommendations for the reliable determination of SARS-CoV-2 genome sequences and demonstrate their broad applicability in a large cohort of clinical samples.