RESUMEN

ABSTRCATO_ST_ABSPurposeC_ST_ABSThis study investigates whether graph-based fusion of imaging data with non-imaging EHR data can improve the prediction of disease trajectory for COVID-19 patients, beyond the prediction performance of only imaging or non-imaging EHR data. Materials and MethodsWe present a novel graph-based framework for fine-grained clinical outcome prediction (discharge, ICU admission, or death) that fuses imaging and non-imaging information using a similarity-based graph structure. Node features are represented by image embedding and edges are encoded with clinical or demographic similarity. ResultsOur experiments on data collected from Emory Healthcare network indicate that our fusion modeling scheme performs consistently better than predictive models using only imaging or non-imaging features, with f1-scores of 0.73, 0.77, and 0.66 for discharge from hospital, mortality, and ICU admission, respectively. External validation was performed on data collected from Mayo Clinic. Our scheme highlights known biases in the model prediction such as bias against patients with alcohol abuse history and bias based on insurance status. ConclusionThe study signifies the importance of fusion of multiple data modalities for accurate prediction of clinical trajectory. Proposed graph structure can model relationships between patients based on non-imaging EHR data and graph convolutional networks can fuse this relationship information with imaging data to effectively predict future disease trajectory more effectively than models employing only imaging or non-imaging data. Forecasting clinical events can enable intelligent resource allocation in hospitals. Our graph-based fusion modeling frameworks can be easily extended to other prediction tasks to efficiently combine imaging data with non-imaging clinical data.

RESUMEN

I.AO_SCPLOWBSTRACTC_SCPLOWThe Coronavirus Disease 2019 (COVID-19) has demonstrated that accurate forecasts of infection and mortality rates are essential for informing healthcare resource allocation, designing countermeasures, implementing public health policies, and increasing public awareness. However, there exist a multitude of modeling methodologies, and their relative performances in accurately forecasting pandemic dynamics are not currently comprehensively understood. In this paper, we introduce the non-mechanistic MIT-LCP forecasting model, and assess and compare its performance to various mechanistic and non-mechanistic models that have been proposed for forecasting COVID-19 dynamics. We performed a comprehensive experimental evaluation which covered the time period of November 2020 to April 2021, in order to determine the relative performances of MIT-LCP and seven other forecasting models from the United States Centers for Disease Control and Prevention (CDC) Forecast Hub. Our results show that there exist forecasting scenarios well-suited to both mechanistic and non-mechanistic models, with mechanistic models being particularly performant for forecasts that are further in the future when recent data may not be as informative, and non-mechanistic models being more effective with shorter prediction horizons when recent representative data is available. Improving our understanding of which forecasting approaches are more reliable, and in which forecasting scenarios, can assist effective pandemic preparation and management.

RESUMEN

Releasing preprints is a popular way to hasten the speed of research but may carry hidden risks for public discourse. The COVID-19 pandemic caused by the novel SARS-CoV-2 infection highlighted the risk of rushing the publication of unvalidated findings, leading to damaging scientific miscommunication in the most extreme scenarios. Several high-profile preprints, later found to be deeply flawed, have indeed exacerbated widespread skepticism about the risks of the COVID-19 disease - at great cost to public health. Here, preprint article quality during the pandemic is examined by distinguishing papers related to COVID-19 from other research studies. Importantly, our analysis also investigated possible factors contributing to manuscript quality by assessing the relationship between preprint quality and gender balance in authorship within each research discipline. Using a comprehensive data set of preprint articles from medRxiv and bioRxiv from January to May 2020, we construct both a new index of manuscript quality including length, readability, and spelling correctness and a measure of gender mix among a manuscripts authors. We find that papers related to COVID-19 are less well-written than unrelated papers, but that this gap is significantly mitigated by teams with better gender balance, even when controlling for variation by research discipline. Beyond contributing to a systematic evaluation of scientific publishing and dissemination, our results have broader implications for gender and representation as the pandemic has led female researchers to bear more responsibility for childcare under lockdown, inducing additional stress and causing disproportionate harm to women in science.

RESUMEN

PURPOSE: Cancer patients are at increased risk of treatment- or disease-related admission to the intensive care unit. Over the past decades, both critical care and cancer care have improved substantially. Due to increased cancer-specific survival, we hypothesized that the number of cancer patients admitted to the intensive care unit (ICU) and survival have increased. MATERIALS AND METHODS: MIMIC-III was used to study trends and outcomes of cancer patients admitted to the ICU between 2002 and 2011. Multiple logistic regression analysis was performed to adjust for confounders of mortality. RESULTS: Among 41,468 patients analyzed, 1,083 were hemato-oncologic, 4,330 were oncologic and 66 patients had both a hematological and solid malignancy. Admission numbers more than doubled and the proportion of cancer patients in the ICU increased steadily from 2002 to 2011. In both the univariate and multivariate analyses, solid cancers and hematologic cancers were strongly associated with 28-day mortality. This association was even stronger for 1-year mortality, with odds ratios of 4.02 (95% confidence interval [CI], 3.69 to 4.38) and 2.25 (95% CI, 1.93 to 2.62), respectively. Over the 10-year study period, both 28-day and 1-year mortality decreased, with hematologic patients showing the strongest annual adjusted decrease in the odds of death. There was considerable heterogeneity among solid cancer types. CONCLUSION: Between 2002 and 2011, the number of cancer patients admitted to the ICU more than doubled, while clinical severity scores remained overall unchanged, suggesting improved treatment. Although cancer patients had higher mortality rates, both 28-day and 1-year mortality of hematologic patients decreased faster than that of non-cancer patients, while mortality rates of cancer patients strongly depended on cancer type.

Asunto(s)

RESUMEN

OBJECTIVE: Medical practice generates and stores immense amounts of clinical process data, while integrating and utilization of these data requires interdisciplinary cooperation together with novel models and methods to further promote applications of medical big data and research of artificial intelligence. A "Datathon" model is a novel event of data analysis and is typically organized as intense, short-duration, competitions in which participants with various knowledge and skills cooperate to address clinical questions based on "real world" data. This article introduces the origin of Datathon, organization of the events and relevant practice. The Datathon approach provides innovative solutions to promote cross-disciplinary collaboration and new methods for conducting research of big data in healthcare. It also offers insight into teaming up multi-expertise experts to investigate relevant clinical questions and further accelerate the application of medical big data.

Asunto(s)