Haemosuccus pancreaticus and seven episodes of recurrent unlocalised upper gastrointestinal bleeding.

Upper gastrointestinal (GI) bleeding is a common medical condition that results in extensive morbidity and mortality, as well as substantial healthcare costs. While there is variation among society and consensus guidelines, the approaches to assessment and evaluation are generally consistent. Our case describes a man in his 40s who presented with seven episodes of recurrent upper GI bleeding over 2 years secondary to haemosuccus pancreaticus. While rare, this case study highlights key principles to the initial diagnostic approach that, in appropriate clinical contexts, should be applied to patients with unlocalised upper GI bleeding. We further perform a complete systematic review of similar cases available in PubMed (36 patients in 24 case reports) to further refine these diagnostic principles.

Empirical data drift detection experiments on real-world medical imaging data.

While it is common to monitor deployed clinical artificial intelligence (AI) models for performance degradation, it is less common for the input data to be monitored for data drift - systemic changes to input distributions. However, when real-time evaluation may not be practical (eg., labeling costs) or when gold-labels are automatically generated, we argue that tracking data drift becomes a vital addition for AI deployments. In this work, we perform empirical experiments on real-world medical imaging to evaluate three data drift detection methods' ability to detect data drift caused (a) naturally (emergence of COVID-19 in X-rays) and (b) synthetically. We find that monitoring performance alone is not a good proxy for detecting data drift and that drift-detection heavily depends on sample size and patient features. Our work discusses the need and utility of data drift detection in various scenarios and highlights gaps in knowledge for the practical application of existing methods.

Reliability and Variability of Ki-67 Digital Image Analysis Methods for Clinical Diagnostics in Breast Cancer.

Ki-67 is a nuclear protein associated with proliferation, and a strong potential biomarker in breast cancer, but is not routinely measured in current clinical management owing to a lack of standardization. Digital image analysis (DIA) is a promising technology that could allow high-throughput analysis and standardization. There is a dearth of data on the clinical reliability as well as intra- and interalgorithmic variability of different DIA methods. In this study, we scored and compared a set of breast cancer cases in which manually counted Ki-67 has already been demonstrated to have prognostic value (n = 278) to 5 DIA methods, namely Aperio ePathology (Lieca Biosystems), Definiens Tissue Studio (Definiens AG), Qupath, an unsupervised immunohistochemical color histogram algorithm, and a deep-learning pipeline piNET. The piNET system achieved high agreement (interclass correlation coefficient: 0.850) and correlation (R = 0.85) with the reference score. The Qupath algorithm exhibited a high degree of reproducibility among all rater instances (interclass correlation coefficient: 0.889). Although piNET performed well against absolute manual counts, none of the tested DIA methods classified common Ki-67 cutoffs with high agreement or reached the clinically relevant Cohen's κ of at least 0.8. The highest agreement achieved was a Cohen's κ statistic of 0.73 for cutoffs 20% and 25% by the piNET system. The main contributors to interalgorithmic variation and poor cutoff characterization included heterogeneous tumor biology, varying algorithm implementation, and setting assignments. It appears that image segmentation is the primary explanation for semiautomated intra-algorithmic variation, which involves significant manual intervention to correct. Automated pipelines, such as piNET, may be crucial in developing robust and reproducible unbiased DIA approaches to accurately quantify Ki-67 for clinical diagnosis in the future.

Impact of phthalate metabolites on vitamin D levels and subclinical inflammation: national health and nutrition examination survey, 2013-2018.

This study explores the association between phthalates and total vitamin D levels and the link between phthalates exposure and subclinical inflammation using monocyte percentage to high-density lipoprotein cholesterol ratio (MHR), utilizing three National Health and Nutrition Examination Survey (NHANES) survey cycles 2013-2018. This study is cross-sectional, utilizing one-time urine samples from randomly selected NHANES participants to assess phthalate metabolites. An inverse association between vitamin D and all Di(2-ethylhexyl) phthalate (DEHP) metabolites was found. The molar sum of DEHP metabolites was inversely associated with vitamin D (ß -2.329; 95% CI -3.937,-0.720). An inverse association was observed between monocarboxynonyl phthalate and vitamin D (ß -0.0278; 95% CI -0.0527,-0.00298). A similar relationship was found between monocarboxyoctyl phthalate and vitamin D (ß -0.0160; 95% CI -0.0242,-0.00775). There was no association between phthalate metabolites and MHR. Stratified analysis showed that the association between phthalate metabolites and MHR may vary according to vitamin D status.

American Black Authorship Has Decreased Across All Clinical Specialties Despite an Increasing Number of Black Physicians Between 1990 and 2020 in the USA.

PURPOSE: Many diversity, equity, and inclusion initiatives assume that attainment of a racially diverse healthcare workforce will translate to increased diversity elsewhere in the healthcare system (e.g., leadership roles or academic authorship). We sought to investigate these trends over time by examining the evolution of physician demographics in the USA, in concert with demographic changes in US authorship in US medical journals from 1990 to 2020 across 25 specialties. METHODS: We reviewed all articles indexed in PubMed, with a primary author affiliation located in the USA and limited to journals based in the USA, relative to the proportion of medical professionals in the CMS National Provider Registry. We employed a previously peer-reviewed/validated algorithm called "averaging-of-proportions" that probabilistically predicts racial identity from surname using the US Census to assess the relationship between diversity among medical professionals and diversity in medical journal authorship. RESULTS: Data reveals a sharp disconnect between the demographic breakdown of physicians and authors. Despite an increase in the number of Black physicians (from 8.5% in 2005 to 9.1% in 2020), there has been a decrease in Black early-career authorship from 7.2% in 1990 to 5.8% in 2020. The percentage of Black early-career authors across all specialties in 2020 is lower than the average per specialty in 1990. Similar trends were noted for Black senior authorship, decreasing from 7.6% in 1990 to 6.2% in 2020, as well as a plateau in Hispanic authorship over the same time interval despite an increasing number of Hispanic physicians. CONCLUSION: Modest advances in physician diversity have not translated to increased diversity in academic authorship. Increasing diversity requires initiatives focused beyond recruitment of underrepresented minorities to medical schools or residencies.

SAGES/AHPBA guidelines for the use of microwave and radiofrequency liver ablation for the surgical treatment of hepatocellular carcinoma or colorectal liver metastases less than 5 cm.

BACKGROUND: Primary hepatocellular carcinoma (HCC) and colorectal liver metastases (CRLM) represent the liver's two most common malignant neoplasms. Liver-directed therapies such as ablation have become part of multidisciplinary therapies despite a paucity of data. Therefore, an expert panel was convened to develop evidence-based recommendations regarding the use of microwave ablation (MWA) and radiofrequency ablation (RFA) for HCC or CRLM less than 5 cm in diameter in patients ineligible for other therapies. METHODS: A systematic review was conducted for six key questions (KQ) regarding MWA or RFA for solitary liver tumors in patients deemed poor candidates for first-line therapy. Subject experts used the GRADE methodology to formulate evidence-based recommendations and future research recommendations. RESULTS: The panel addressed six KQs pertaining to MWA vs. RFA outcomes and laparoscopic vs. percutaneous MWA. The available evidence was poor quality and individual studies included both HCC and CRLM. Therefore, the six KQs were condensed into two, recognizing that these were two disparate tumor groups and this grouping was somewhat arbitrary. With this significant limitation, the panel suggested that in appropriately selected patients, either MWA or RFA can be safe and feasible. However, this recommendation must be implemented cautiously when simultaneously considering patients with two disparate tumor biologies. The limited data suggested that laparoscopic MWA of anatomically more difficult tumors has a compensatory higher morbidity profile compared to percutaneous MWA, while achieving similar overall 1-year survival. Thus, either approach can be appropriate depending on patient-specific factors (very low certainty of evidence). CONCLUSION: Given the weak evidence, these guidelines provide modest guidance regarding liver ablative therapies for HCC and CRLM. Liver ablation is just one component of a multimodal approach and its use is currently limited to a highly selected population. The quality of the existing data is very low and therefore limits the strength of the guidelines.

Ethnicity and gender trends of UK authors in The British Medical Journal and the Lancet over the past two decades: a comprehensive longitudinal analysis.

Background: While gender equity among academic authors has been extensively investigated, there is a significant gap in our understanding of racial/ethnic authorship trends, despite the recognition of barriers to authorship along both ethnic and gender lines. Leveraging the meta-data for all articles published in The British Medical Journal (The BMJ) and the Lancet and between 2002 and 2022 (inclusive), we explore demographic trends among UK academic medicine authors in two of the world's leading British medical journals. Methods: We systematically searched PubMed's MEDLINE for all articles published in The BMJ and Lancet between January 1st 2002 and December 31st 2022. Filtering for articles with a UK affiliation, we predicted gender using a publicly-validated name-to-gender dictionary, while data was analysed to explore and investigate ethnicity using the Consumer Data Research Centre's (CDRC) Ethnicity Estimator. Data was analysed to explore and investigate: (a) the proportion of female/male author publications, (b) the proportion of the various UK author ethnicity groups, and (c) the overlap/intersection between gender and ethnic identities among first and last authors. This comprehensive longitudinal analysis was conducted on 82,143 articles (51,209 from The BMJ and 30,934 from the Lancet) which represents >97% of all published articles between 2002 and 2022. As we sought to understand how academic authorship reflects the diversity of the UK population, we limited our analysis to first and last authors who had a UK affiliation and excluded "news" and "comments" pieces (16,736 articles for The BMJ and 4678 articles from the Lancet). The main outcome measures were the trends in first and last authorship demographics of academic medicine, focusing on the proportion of female/male authors, ethnicity and their intersectionality. Findings: Our findings show that, while women have made substantial headway towards equity among first and last authorship in The BMJ (peaking at 42% and 43%), they remain under-represented in the Lancet (35% and 27%). In both The BMJ and Lancet, Black authors have remained severely under-represented as both first and last authors (below 1% for most of the two decades), while Asian authors have increased proportionally to match their fraction in the general population (ranging from 2 to 10%). Interpretation: Analysis over the past two decades has shown that the gender author gap is decreasing quickly in The BMJ and Lancet. However, despite the two journals' growing focus on structural inequalities in medical academia, little progress has been made in rectifying the large gap between White British authors and other ethnic groups, especially Black authors. Without more awareness, diversity initiatives which have resulted in positive gains for White women do not seem to translate well for authors of colour. Funding: None.

Clinical Phenotypes of PCOS: a Cross-Sectional Study.

This cross-sectional study examines the Doi-Alshoumer PCOS clinical phenotype classification in relation to measured clinical and biochemical characteristics of women with polycystic ovary syndrome (PCOS). Two cohorts of women (Kuwait and Rotterdam) diagnosed with PCOS (FAI > 4.5%) were examined. These phenotypes were created using neuroendocrine dysfunction (IRMA LH/FSH ratio > 1 or LH > 6 IU/L) and menstrual cycle status (oligo/amenorrhea) to create three phenotypes: (A) neuroendocrine dysfunction and oligo/amenorrhea, (B) without neuroendocrine dysfunction but with oligo/amenorrhea, and (C) without neuroendocrine dysfunction and with regular cycles. These phenotypes were compared in terms of hormonal, biochemical, and anthropometric measures. The three suggested phenotypes (A, B, and C) were shown to be sufficiently distinct in terms of hormonal, biochemical, and anthropometric measures. Patients who were classified as phenotype A had neuroendocrine dysfunction, excess LH (and LH/FSH ratio), irregular cycles, excess A4, infertility, excess T, highest FAI and E2, and excess 17αOHPG when compared to the other phenotypes. Patients classified as phenotype B had irregular cycles, no neuroendocrine dysfunction, obesity, acanthosis nigricans, and insulin resistance. Lastly, patients classified as phenotype C had regular cycles, acne, hirsutism, excess P4, and the highest P4 to E2 molar ratio. The differences across phenotypes suggested distinct phenotypic expression of this syndrome, and the biochemical and clinical correlates of each phenotype are likely to be useful in the management of women with PCOS. These phenotypic criteria are distinct from criteria used for diagnosis.

Tracing the path of 37,050 studies into practice across 18 specialties of the 2.4 million published between 2011 and 2020.

The absence of evidence to assess treatment efficacy partially underpins the unsustainable expenditure of the US healthcare system, a challenge exacerbated by a limited understanding of the factors influencing the translation of clinical research into practice. Leveraging a dataset of >10,000 UpToDate articles, sampled every 3 months between 2011 and 2020, we trace the path of research (37,050 newly added articles from 887 journals) from initial publication to the point-of-care, compared to the 2.4 million uncited studies published during the same time window across 18 medical specialties. Our analysis reveals substantial variation in how specialties prioritize/adopt research, with regards to a fraction of literature cited (0.4-2.4%) and quality-of-evidence incorporated. In 9 of 18 specialties, less than 1 in 10 clinical trials are ever cited. Furthermore, case reports represent one of the most cited article types in 12 of 18 specialties, comprising nearly a third of newly added references for some specialties (e.g. dermatology). Anesthesiology, cardiology, critical care, geriatrics, internal medicine, and oncology tended to favor higher-quality evidence. By modeling citations as a function of National Institutes of Health (NIH) department-specific funding, we estimate the cost of bringing one new clinical citation to the point-of-care as ranging from thousands to tens of thousands of dollars depending on specialty. The success of a subset of specialties in incorporating a larger proportion of published research, as well as high(er) quality of evidence, demonstrates the existence of translational strategies that should be applied more broadly. In addition to providing a baseline for monitoring the efficiency of research investments, we also describe new 'impact' indices to assess the efficacy of reforms to the clinical scientific enterprise.

Surgical approach to microwave and radiofrequency liver ablation for hepatocellular carcinoma and colorectal liver metastases less than 5 cm: a systematic review and meta-analysis.

BACKGROUND: Primary hepatocellular carcinoma (HCC) and colorectal liver metastases (CRLM) represent the two most common malignant neoplasms of the liver. The objective of this study was to assess outcomes of surgical approaches to liver ablation comparing laparoscopic versus percutaneous microwave ablation (MWA), and MWA versus radiofrequency ablation (RFA) in patients with HCC or CRLM lesions smaller than 5 cm. METHODS: A systematic review was conducted across seven databases, including PubMed, Embase, and Cochrane, to identify all comparative studies between 1937 and 2021. Two independent reviewers screened for eligibility, extracted data for selected studies, and assessed study bias using the modified Newcastle Ottawa Scale. Random effects meta-analyses were subsequently performed on all available comparative data. RESULTS: From 1066 records screened, 11 studies were deemed relevant to the study and warranted inclusion. Eight of the 11 studies were at high or uncertain risk for bias. Our meta-analyses of two studies revealed that laparoscopic MW ablation had significantly higher complication rates compared to a percutaneous approach (risk ratio = 4.66; 95% confidence interval = [1.23, 17.22]), but otherwise similar incomplete ablation rates, local recurrence, and oncologic outcomes. The remaining nine studies demonstrated similar efficacy of MWA and RFA, as measured by incomplete ablation, complication rates, local/regional recurrence, and oncologic outcomes, for both HCC and CRLM lesions less than 5 cm (p > 0.05 for all outcomes). There was no statistical subgroup interaction in the analysis of tumors < 3 cm. CONCLUSION: The available comparative evidence regarding both laparoscopic versus percutaneous MWA and MWA versus RFA is limited, evident by the few studies that suffer from high/uncertain risk of bias. Additional high-quality randomized trials or statistically matched cohort studies with sufficient granularity of patient variables, institutional experience, and physician specialty/training will be useful in informing clinical decision making for the ablative treatment of HCC or CRLM.

The Under-representation and Stagnation of Female, Black, and Hispanic Authorship in the Journal of the American Medical Association and the New England Journal of Medicine.

Publication in leading medical journals is critical to knowledge dissemination and academic advancement alike. Leveraging a novel dataset comprised of nearly all articles published in JAMA and NEJM from 1990 to 2020, along with established reference works for name identification, we explore changing authorship demographics in two of the world's leading medical journals. Our main outcomes are the annual proportion of male and female authors and the proportion of racial/ethnic identities in junior and senior authorship positions for articles published in JAMA and NEJM since 1990. We found that women remain under-represented in research authorship in both JAMA (at its peak, 38.1% of articles had a female first author in 2011) and NEJM (peaking at 28.2% in 2002). The rate of increase is so slow that it will take more than a century for both journals to reach gender parity. Black and Hispanic researchers have likewise remained under-represented as first and last authors in both journals, even using the best-case scenario. Their appearance as authors has remained stagnant for three decades, despite attention to structural inequalities in medical academia. Thus, analysis of authorship demographics in JAMA and NEJM over the past three decades reveals the existence of inequalities in high-impact medical journal authorship. Gender and racial/ethnic disparities in authorship may both reflect and further contribute to disparities in academic advancement.

Insights from full-text analyses of the Journal of the American Medical Association and the New England Journal of Medicine.

Analysis of the content of medical journals enables us to frame the shifting scientific, material, ethical, and epistemic underpinnings of medicine over time, including today. Leveraging a dataset comprised of nearly half-a-million articles published in the Journal of the American Medical Association (JAMA) and the New England Journal of Medicine (NEJM) over the past 200 years, we (a) highlight the evolution of medical language, and its manifestations in shifts of usage and meaning, (b) examine traces of the medical profession's changing self-identity over time, reflected in its shifting ethical and epistemic underpinnings, (c) analyze medicine's material underpinnings and how we describe where medicine is practiced, (d) demonstrate how the occurrence of specific disease terms within the journals reflects the changing burden of disease itself over time and the interests and perspectives of authors and editors, and (e) showcase how this dataset can allow us to explore the evolution of modern medical ideas and further our understanding of how modern disease concepts came to be, and of the retained legacies of prior embedded values.

A general framework for predicting the transcriptomic consequences of non-coding variation and small molecules.

Genome wide association studies (GWASs) for complex traits have implicated thousands of genetic loci. Most GWAS-nominated variants lie in noncoding regions, complicating the systematic translation of these findings into functional understanding. Here, we leverage convolutional neural networks to assist in this challenge. Our computational framework, peaBrain, models the transcriptional machinery of a tissue as a two-stage process: first, predicting the mean tissue specific abundance of all genes and second, incorporating the transcriptomic consequences of genotype variation to predict individual abundance on a subject-by-subject basis. We demonstrate that peaBrain accounts for the majority (>50%) of variance observed in mean transcript abundance across most tissues and outperforms regularized linear models in predicting the consequences of individual genotype variation. We highlight the validity of the peaBrain model by calculating non-coding impact scores that correlate with nucleotide evolutionary constraint that are also predictive of disease-associated variation and allele-specific transcription factor binding. We further show how these tissue-specific peaBrain scores can be leveraged to pinpoint functional tissues underlying complex traits, outperforming methods that depend on colocalization of eQTL and GWAS signals. We subsequently: (a) derive continuous dense embeddings of genes for downstream applications; (b) highlight the utility of the model in predicting transcriptomic impact of small molecules and shRNA (on par with in vitro experimental replication of external test sets); (c) explore how peaBrain can be used to model difficult-to-study processes (such as neural induction); and (d) identify putatively functional eQTLs that are missed by high-throughput experimental approaches.

Genome-Wide Association Study of Peripheral Artery Disease.

BACKGROUND: Peripheral artery disease (PAD) affects >200 million people worldwide and is associated with high mortality and morbidity. We sought to identify genomic variants associated with PAD overall and in the contexts of diabetes and smoking status. METHODS: We identified genetic variants associated with PAD and then meta-analyzed with published summary statistics from the Million Veterans Program and UK Biobank to replicate their findings. Next, we ran stratified genome-wide association analysis in ever smokers, never smokers, individuals with diabetes, and individuals with no history of diabetes and corresponding interaction analyses, to identify variants that modify the risk of PAD by diabetic or smoking status. RESULTS: We identified 5 genome-wide significant (Passociation ≤5×10-8) associations with PAD in 449 548 (Ncases=12 086) individuals of European ancestry near LPA (lipoprotein [a]), CDKN2BAS1 (CDKN2B antisense RNA 1), SH2B3 (SH2B adaptor protein 3) - PTPN11 (protein tyrosine phosphatase non-receptor type 11), HDAC9 (histone deacetylase 9), and CHRNA3 (cholinergic receptor nicotinic alpha 3 subunit) loci (which overlapped previously reported associations). Meta-analysis with variants previously associated with PAD showed that 18 of 19 published variants remained genome-wide significant. In individuals with diabetes, rs116405693 at the CCSER1 (coiled-coil serine rich protein 1) locus was associated with PAD (odds ratio [95% CI], 1.51 [1.32-1.74], Pdiabetes=2.5×10-9, Pinteractionwithdiabetes=5.3×10-7). Furthermore, in smokers, rs12910984 at the CHRNA3 locus was associated with PAD (odds ratio [95% CI], 1.15 [1.11-1.19], Psmokers=9.3×10-10, Pinteractionwithsmoking=3.9×10-5). CONCLUSIONS: Our analyses confirm the published genetic associations with PAD and identify novel variants that may influence susceptibility to PAD in the context of diabetes or smoking status.

Weight Loss Following Roux-en-Y Gastric Bypass Causally Implicated with Serum Levels of IL-22: A Mendelian Randomization and Phenome-Wide Association Study.

OBJECTIVE: Laparoscopic Roux-en-Y gastric bypass (RYGB) modulates the low-grade inflammatory state associated with severe obesity. This study sought to investigate whether weight loss is causally implicated with changes in serum levels of inflammatory molecules. METHODS: Using the largest genome-wide association study (n = 1,020 individuals), this study curated five genetic variants associated with weight loss following RYGB. Phenome-wide association studies (PheWAS) were performed to identify other phenotypes associated with these variants. Subsequently, two-sample Mendelian randomization was used to study the causal effects of weight loss on the serum levels of 382 inflammatory proteins (measured in 3,033 individuals). This is the first systematic quasi-experimental investigation of weight loss following RYGB and serum markers of inflammation. RESULTS: The PheWAS analysis revealed that four of the five variants are associated with phenotypes relating to metabolism and inflammation, including insulin response and levels of C-reactive protein. Two-sample Mendelian randomization of the 382 serum inflammatory markers revealed that weight loss following RYGB increases serum levels of interleukin 22 (IL-22) (beta = 0.021, P < 10-3 ; 95% CI: 0.010-0.031). Sensitivity analyses further supported the results and the causal direction. CONCLUSIONS: Weight loss following RYGB may cause an increase in IL-22 serum levels, suggesting that weight loss directly contributes to immune modulation following bypass. These results demonstrate the utility of genetic studies to disentangling molecular cause and effect following bariatric surgery.

Asynchrony Between Individual and Government Actions Accounts for Disproportionate Impact of COVID-19 on Vulnerable Communities.

INTRODUCTION: Previously estimated effects of social distancing do not account for changes in individual behavior before the implementation of stay-at-home policies or model this behavior in relation to the burden of disease. This study aims to assess the asynchrony between individual behavior and government stay-at-home orders, quantify the true impact of social distancing using mobility data, and explore the sociodemographic variables linked to variation in social distancing practices. METHODS: This study was a retrospective investigation that leveraged mobility data to quantify the time to behavioral change in relation to the initial presence of COVID-19 and the implementation of government stay-at-home orders. The impact of social distancing that accounts for both individual behavior and testing data was calculated using generalized mixed models. The role of sociodemographics in accounting for variation in social distancing behavior was modeled using a 10-fold cross-validated elastic net (linear machine learning model). Analysis was conducted in April‒July 2020. RESULTS: Across all the 1,124 counties included in this analysis, individuals began to socially distance at a median of 5 days (IQR=3-8) after 10 cumulative cases of COVID-19 were confirmed in their state, with state governments taking a median of 15 days (IQR=12-19) to enact stay-at-home orders. Overall, people began social distancing at a median of 12 days (IQR=8-17) before their state enacted stay-at-home orders. Of the 16 studies included in the review, 13 exclusively used government dates as a proxy for social distancing behavior, and none accounted for both testing and mobility. Using government stay-at-home dates as a proxy for social distancing (10.2% decrease in the number of daily cases) accounted for only 55% of the true impact of the intervention when compared with estimates using mobility (18.6% reduction). Using 10-fold cross-validation, 23 of 43 sociodemographic variables were significantly and independently predictive of variation in individual social distancing, with delays corresponding to an increase in a county's proportion of people without a high school diploma and proportion of racial and ethnic minorities. CONCLUSIONS: This retrospective analysis of mobility patterns found that social distancing behavior occurred well before the onset of government stay-at-home dates. This asynchrony leads to the underestimation of the impact of social distancing. Sociodemographic characteristics associated with delays in social distancing can help explain the disproportionate case burden and mortality among vulnerable communities.

A Multi-omic Integrative Scheme Characterizes Tissues of Action at Loci Associated with Type 2 Diabetes.

Resolving the molecular processes that mediate genetic risk remains a challenge because most disease-associated variants are non-coding and functional characterization of these signals requires knowledge of the specific tissues and cell-types in which they operate. To address this challenge, we developed a framework for integrating tissue-specific gene expression and epigenomic maps to obtain "tissue-of-action" (TOA) scores for each association signal by systematically partitioning posterior probabilities from Bayesian fine-mapping. We applied this scheme to credible set variants for 380 association signals from a recent GWAS meta-analysis of type 2 diabetes (T2D) in Europeans. The resulting tissue profiles underscored a predominant role for pancreatic islets and, to a lesser extent, adipose and liver, particularly among signals with greater fine-mapping resolution. We incorporated resulting TOA scores into a rule-based classifier and validated the tissue assignments through comparison with data from cis-eQTL enrichment, functional fine-mapping, RNA co-expression, and patterns of physiological association. In addition to implicating signals with a single TOA, we found evidence for signals with shared effects in multiple tissues as well as distinct tissue profiles between independent signals within heterogeneous loci. Lastly, we demonstrated that TOA scores can be directly coupled with eQTL colocalization to further resolve effector transcripts at T2D signals. This framework guides mechanistic inference by directing functional validation studies to the most relevant tissues and can gain power as fine-mapping resolution and cell-specific annotations become richer. This method is generalizable to all complex traits with relevant annotation data and is made available as an R package.

Estimating access to health care in Yemen, a complex humanitarian emergency setting: a descriptive applied geospatial analysis.

BACKGROUND: In conflict settings, data to guide humanitarian and development responses are often scarce. Although geospatial analyses have been used to estimate health-care access in many countries, such techniques have not been widely applied to inform real-time operations in protracted health emergencies. Doing so could provide a more robust approach for identifying and prioritising populations in need, targeting assistance, and assessing impact. We aimed to use geospatial analyses to overcome such data gaps in Yemen, the site of one of the world's worst ongoing humanitarian crises. METHODS: We derived geospatial coordinates, functionality, and service availability data for Yemen health facilities from the Health Resources and Services Availability Monitoring System assessment done by WHO and the Yemen Ministry of Public Health and Population. We modelled population spatial distribution using high-resolution satellite imagery, UN population estimates, and census data. A road network grid was built from OpenStreetMap and satellite data and modified using UN Yemen Logistics Cluster data and other datasets to account for lines of conflict and road accessibility. Using this information, we created a geospatial network model to deduce the travel time of Yemeni people to their nearest health-care facilities. FINDINGS: In 2018, we estimated that nearly 8·8 million (30·6%) of the total estimated Yemeni population of 28·7 million people lived more than 30-min travel time from the nearest fully or partially functional public primary health-care facility, and more than 12·1 million (42·4%) Yemeni people lived more than 1 h from the nearest fully or partially functional public hospital, assuming access to motorised transport. We found that access varied widely by district and type of health service, with almost 40% of the population living more than 2 h from comprehensive emergency obstetric and surgical care. We identified and ranked districts according to the number of people living beyond acceptable travel times to facilities and services. We found substantial variability in access and that many front-line districts were among those with the poorest access. INTERPRETATION: These findings provide the most comprehensive estimates of geographical access to health care in Yemen since the outbreak of the current conflict, and they provide proof of concept for how geospatial techniques can be used to address data gaps and rigorously inform health programming. Such information is of crucial importance for humanitarian and development organisations seeking to improve effectiveness and accountability. FUNDING: Global Financing Facility for Women, Children, and Adolescents Trust Fund; Development and Data Science grant; and the Yemen Emergency Health and Nutrition Project, a partnership between the World Bank, UNICEF, and WHO.

Exploring the Privacy-Preserving Properties of Word Embeddings: Algorithmic Validation Study.

BACKGROUND: Word embeddings are dense numeric vectors used to represent language in neural networks. Until recently, there had been no publicly released embeddings trained on clinical data. Our work is the first to study the privacy implications of releasing these models. OBJECTIVE: This paper aims to demonstrate that traditional word embeddings created on clinical corpora that have been deidentified by removing personal health information (PHI) can nonetheless be exploited to reveal sensitive patient information. METHODS: We used embeddings created from 400,000 doctor-written consultation notes and experimented with 3 common word embedding methods to explore the privacy-preserving properties of each. RESULTS: We found that if publicly released embeddings are trained from a corpus anonymized by PHI removal, it is possible to reconstruct up to 68.5% (n=411/600) of the full names that remain in the deidentified corpus and associated sensitive information to specific patients in the corpus from which the embeddings were created. We also found that the distance between the word vector representation of a patient's name and a diagnostic billing code is informative and differs significantly from the distance between the name and a code not billed for that patient. CONCLUSIONS: Special care must be taken when sharing word embeddings created from clinical texts, as current approaches may compromise patient privacy. If PHI removal is used for anonymization before traditional word embeddings are trained, it is possible to attribute sensitive information to patients who have not been fully deidentified by the (necessarily imperfect) removal algorithms. A promising alternative (ie, anonymization by PHI replacement) may avoid these flaws. Our results are timely and critical, as an increasing number of researchers are pushing for publicly available health data.