Future healthcare systems will rely heavily on clinical decision support (CDSS) to improve the decision-making processes of clinicians. To explore design future CDSS, we developed a research-focused CDSS for management patients in intensive care unit that leverages Internet Things (IoT) devices capable collecting streaming physiologic data from ventilators and other medical devices. We then created machine learning (ML) models could analyze collected determine if ventilator was delivering...

Background Commonly used cardiovascular risk calculators do not provide estimation of stroke, a major postoperative complication with high morbidity and mortality. We developed validated an accurate prediction tool for cardiac complications (myocardial infarction or arrest), mortality after non‐cardiac surgery. Methods Results This retrospective cohort study included 1 165 750 surgical patients over 4‐year period (2007–2010) from the American College Surgeons National Surgical Quality...

We introduce a new paradigm for single-channel target source separation where the sources of interest can be distinguished using non-mutually exclusive concepts (e.g., loudness, gender, language, spatial location, etc). Our proposed heterogeneous framework seamlessly leverage datasets with large distribution shifts and learn cross-domain representations under variety used as conditioning. experiments show that training models conditions facilitates generalization to unseen out-of-domain data...

Recent studies in talking face generation have focused on building a model that can generalize from any source speech to target identity. A number of works already claimed this functionality and added their models will also language. However, we show, using languages different language families, these do not translate well when the training testing are sufficiently different. We reduce scope problem language-robust system seen identities, i.e., identity is same as In work, introduce...

Mobile electrocardiograms (ECGs) (mECGs) using smartphone applications are an emerging technology.In the coronavirus disease 2019 (COVID-19) era, minimizing patient contact has gained increasing importance.Additionally, increased QT/corrected QT (QTc) monitoring concurrently been required.The KardiaMobile 6L ECG device, cleared by United States Food and Drug Administration (FDA) for recording ECGs, along with KardiaStation tablet application is a platform (AliveCor, Mountain View, CA, USA)...

We propose EnCLAP, a novel framework for automated audio captioning. EnCLAP employs two acoustic representation models, EnCodec and CLAP, along with pretrained language model, BART. also introduce new training objective called masked codec modeling that improves awareness of the model. Experimental results on AudioCaps Clotho demonstrate our model surpasses performance baseline models. Source code will be available at https://github.com/jaeyeonkim99/EnCLAP. <sup...

IgG commonly co-exists with IgA in the glomerular mesangium of patients nephropathy (IgAN) unclear clinical relevance. Autoantibody (autoAb) biomarkers to detect and track progression IgAN are an unmet need. The objective study was identify IgA-specific autoAbs specific IgAN.High-density protein microarrays were evaluated serum (n = 22) controls 10). Clinical parameters, including annual GFR urine measurements, collected on all over 5 years. Bioinformatic data analysis performed select...

AKI after surgery is associated with high mortality and morbidity. The purpose of this study to develop validate a risk prediction tool for the occurrence postoperative requiring RRT (AKI-dialysis).This retrospective cohort had 2,299,502 surgical patients over 2015-2017 from American College Surgeons National Surgical Quality Improvement Program Database (ACS NSQIP). Eleven predictors were selected predictive model: age, history congestive heart failure, diabetes, ascites, emergency surgery,...

IgA plays a key role in nephropathy (IgAN) by forming immune complexes and depositing the glomeruli, leading to an inflammatory response. However, antigenic targets functional characterization of have been incompletely defined this disease.This study was performed sera from patients who were studied as part prospective, observational IgAN. These (n=22) all had biopsy-proven IgAN within 3 years initiation, complete clinical data, annual urinary inulin clearance for GFRs, at least 5 follow-up....

Coronavirus disease 2019 (COVID-19) carries high morbidity and mortality globally. Identification of patients at risk for clinical deterioration upon presentation would aid in triaging, prognostication, allocation resources experimental treatments.

Individual cytokines have been reported to be associated with the pathogenesis and prognosis of immunoglobulin A (IgA) nephropathy (IgAN). We attempted characterize a broad cytokine matrix in serum patients, investigate its association renal function. Thirty-two were simultaneous-ly measured from IgAN patients (n=22) healthy controls (n=12). Clinical variables including annual inulin based GFR 24-h proteinuria collected on over 5-year period. There was significant difference between...

We propose EnCLAP, a novel framework for automated audio captioning. EnCLAP employs two acoustic representation models, EnCodec and CLAP, along with pretrained language model, BART. also introduce new training objective called masked codec modeling that improves awareness of the model. Experimental results on AudioCaps Clotho demonstrate our model surpasses performance baseline models. Source code will be available at https://github.com/jaeyeonkim99/EnCLAP . An online demo is...

In this technical report, we describe our submission to DCASE2024 Challenge Task6 (Automated Audio Captioning) and Task8 (Language-based Retrieval). We develop approach building upon the EnCLAP audio captioning framework optimizing it for of challenge. Notably, outline changes in underlying components incorporation reranking process. Additionally, submit a supplementary retriever model, byproduct modified framework, Task8. Our proposed systems achieve FENSE score 0.542 on mAP@10 0.386 Task8,...

In this work, we aim to analyze and optimize the EnCLAP framework, a state-of-the-art model in automated audio captioning. We investigate impact of modifying acoustic encoder components, explore pretraining with different dataset scales, study effectiveness reranking scheme. Through extensive experimentation quantitative analysis generated captions, develop EnCLAP++, an enhanced version that significantly surpasses original.

ABSTRACT Background Coronavirus disease 2019 (COVID-19) carries high morbidity and mortality globally. Identification of patients at risk for clinical deterioration upon presentation would aid in triaging, prognostication, allocation resources experimental treatments. Research Question Can we develop validate a web-based prediction model identification who may severe COVID-19, defined as intensive care unit (ICU) admission, mechanical ventilation, and/or death? Methods This retrospective...

Troubleshooting a vehicle often requires some form of customer service; to help guide towards resolution. Consumer reporting faulty vehicular components can be made via telephone- based service call, or gathered through telemetry embedded intelligent transportation systems (ITS). During data transmission, free text is created in the dialogue between and representative. While generated, key details discussion extracted recorded. This paper describes methods used process recorded data. method...

Abstract Background Clostridioides difficile infection is associated with significant morbidity, mortality and increased costs. Assessment of the postoperative C. risk necessary to improve outcome surgical patients. Objective To develop validate a prediction tool for after surgery. Methods In this retrospective cohort study, 2,451,169 patients from American College Surgeons National Surgical Quality Improvement Program Database (ACS-NSQIP) over 2015-2017 were included. Nine predictors...

Patients with sepsis are at increased risk for cardiovascular complications, including myocardial infarction (MI), ischemic stroke (IS), and venous thromboembolism (VTE). Our objective is to assess whether genetic score (GRS) can differentiate these complications.A population-based prospective cohort of 483,177 subjects, derived from the UK Biobank, was followed diagnosis its complications (MI, IS, VTE) after study recruitment. GRS each complication calculated based on established...