Development of a federated learning approach to predict acute kidney injury in adult hospitalized patients with COVID-19 in New York City.
Ontology highlight
ABSTRACT:
SUBMITTER: Jaladanki SK
PROVIDER: S-EPMC8328073 | biostudies-literature |
REPOSITORIES: biostudies-literature
Similar Datasets
Project description: Not available
| S-EPMC8325375 | biostudies-literature
Project description:Background and aimsCoronavirus disease 2019 (COVID-19) is a global threat, affecting more than 100 million people and causing over 2 million deaths. Liver laboratory test abnormalities are an extrapulmonary manifestation of COVID-19, yet characterization of hepatic injury is incomplete. Our objective was to further characterize and identify causes of liver injury in patients with COVID-19.MethodsWe conducted a retrospective cohort study of 551 patients hospitalized with COVID-19 at NewYork-Presbyterian Hospital/Columbia University Irving Medical Center between March 1, 2020 and May 31, 2020. We analyzed patient demographics, liver laboratory test results, vital signs, other relevant test results, and clinical outcomes (mortality and intensive care unit admission).ResultsAbnormal liver laboratory tests were common on hospital admission for COVID-19 and the incidence increased during hospitalization. Of those with elevated serum alanine aminotransferase and/or alkaline phosphatase activities on admission, 58.2% had a cholestatic injury pattern, 35.2% mixed, and 6.6% hepatocellular. Comorbid liver disease was not associated with outcome; however, abnormal direct bilirubin or albumin on admission were associated with intensive care unit stay and mortality. On average, patients who died had greater magnitudes of abnormalities in all liver laboratory tests than those who survived. Ischemic hepatitis was a mechanism of severe hepatocellular injury in some patients.ConclusionsLiver laboratory test abnormalities are common in hospitalized patients with COVID-19, and some are associated with increased odds of intensive care unit stay or death. Severe hepatocellular injury is likely attributable to secondary effects such as systemic inflammatory response syndrome, sepsis, and ischemic hepatitis.
| S-EPMC8369008 | biostudies-literature
Project description:BackgroundMachine learning models require large datasets that may be siloed across different health care institutions. Machine learning studies that focus on COVID-19 have been limited to single-hospital data, which limits model generalizability.ObjectiveWe aimed to use federated learning, a machine learning technique that avoids locally aggregating raw clinical data across multiple institutions, to predict mortality in hospitalized patients with COVID-19 within 7 days.MethodsPatient data were collected from the electronic health records of 5 hospitals within the Mount Sinai Health System. Logistic regression with L1 regularization/least absolute shrinkage and selection operator (LASSO) and multilayer perceptron (MLP) models were trained by using local data at each site. We developed a pooled model with combined data from all 5 sites, and a federated model that only shared parameters with a central aggregator.ResultsThe LASSOfederated model outperformed the LASSOlocal model at 3 hospitals, and the MLPfederated model performed better than the MLPlocal model at all 5 hospitals, as determined by the area under the receiver operating characteristic curve. The LASSOpooled model outperformed the LASSOfederated model at all hospitals, and the MLPfederated model outperformed the MLPpooled model at 2 hospitals.ConclusionsThe federated learning of COVID-19 electronic health record data shows promise in developing robust predictive models without compromising patient privacy.
| S-EPMC7842859 | biostudies-literature
Project description: Not available
| S-EPMC7377796 | biostudies-literature
Project description:IntroductionReports from the United States suggest that acute kidney injury (AKI) frequently complicates coronavirus disease 2019 (COVID-19), but understanding of AKI risks and outcomes is incomplete. In addition, whether kidney outcomes have evolved during the course of the pandemic is unknown.MethodsWe used electronic medical records to identify patients with COVID-19 with and without AKI admitted to 3 New York Hospitals between March 2 and August 25, 2020. Outcomes included AKI overall and according to admission week, AKI stage, the requirement for new renal replacement therapy (RRT), mortality, and recovery of kidney function. Logistic regression was used to assess associations of patient characteristics and outcomes.ResultsOf 4732 admissions, 1386 (29.3%) patients had AKI. Among those with AKI, 717 (51.7%) had stage 1 disease, 132 (9.5%) had stage 2 disease, 537 (38.7%) had stage 3 disease, and 237 (17.1%) required RRT initiation. In March, 536 of 1648 (32.5%) patients developed AKI compared with 15 of 87 (17.2%) in August (P < 0.001 for monthly trend), whereas RRT initiation was required in 6.9% and 0% of admissions in March and August, respectively. Mortality was higher with than without AKI (51.6% vs. 8.6%) and was 71.9% in individuals requiring RRT. However, most patients with AKI who survived hospitalization (77%) recovered to within 0.3 mg/dl of baseline creatinine. Among those surviving to discharge, 62% discontinued RRT.ConclusionsAKI impacts a high proportion of admitted patients with COVID-19 and is associated with high mortality, particularly when RRT is required. AKI incidence appears to be decreasing over time and kidney function frequently recovers in those who survive.
| S-EPMC7857986 | biostudies-literature
Project description:Deep venous thrombosis (DVT) is associated with significant morbidity, mortality, and increased healthcare costs. Standard scoring systems for DVT risk stratification often provide insufficient stratification of hospitalized patients and are unable to accurately predict which inpatients are most likely to present with DVT. There is a continued need for tools which can predict DVT in hospitalized patients. We performed a retrospective study on a database collected from a large academic hospital, comprised of 99,237 total general ward or ICU patients, 2,378 of whom experienced a DVT during their hospital stay. Gradient boosted machine learning algorithms were developed to predict a patient's risk of developing DVT at 12- and 24-hour windows prior to onset. The primary outcome of interest was diagnosis of in-hospital DVT. The machine learning predictors obtained AUROCs of 0.83 and 0.85 for DVT risk prediction on hospitalized patients at 12- and 24-hour windows, respectively. At both 12 and 24 hours before DVT onset, the most important features for prediction of DVT were cancer history, VTE history, and internal normalized ratio (INR). Improved risk stratification may prevent unnecessary invasive testing in patients for whom DVT cannot be ruled out using existing methods. Improved risk stratification may also allow for more targeted use of prophylactic anticoagulants, as well as earlier diagnosis and treatment, preventing the development of pulmonary emboli and other sequelae of DVT.
| S-EPMC7907939 | biostudies-literature
Project description: Not available
| S-EPMC11367008 | biostudies-literature
Project description:The significance of minimal increases in serum creatinine below the levels indicative of the acute kidney injury (AKI) stage is not well established. We aimed to investigate the influence of pre-stage AKI (pre-AKI) on clinical outcomes. We enrolled a total of 21,261 patients who were admitted to the Seoul National University Bundang Hospital from January 1, 2013 to December 31, 2013. Pre-AKI was defined as a 25-50% increase in peak serum creatinine levels from baseline levels during the hospital stay. In total, 5.4% of the patients had pre-AKI during admission. The patients with pre-AKI were predominantly female (55.0%) and had a lower body weight and lower baseline levels of serum creatinine (0.63 ± 0.18 mg/dl) than the patients with AKI and the patients without AKI (P < 0.001). The patients with pre-AKI had a higher prevalence of diabetes mellitus (25.1%) and malignancy (32.6%). The adjusted hazard ratio of in-hospital mortality for pre-AKI was 2.112 [95% confidence interval (CI), 1.143 to 3.903]. In addition, patients with pre-AKI had an increased length of stay (7.7 ± 9.7 days in patients without AKI, 11.4 ± 11.4 days in patients with pre-AKI, P < 0.001) and increased medical costs (4,061 ± 4,318 USD in patients without AKI, 4,966 ± 5,099 USD in patients with pre-AKI, P < 0.001) during admission. The adjusted hazard ratio of all-cause mortality for pre-AKI during the follow-up period of 2.0 ± 0.6 years was 1.473 (95% CI, 1.228 to 1.684). Although the adjusted hazard ratio of pre-AKI for overall mortality was not significant among the patients admitted to the surgery department or who underwent surgery, pre-AKI was significantly associated with mortality among the non-surgical patients (adjusted HR 1.542 [95% CI, 1.330 to 1.787]) and the patients admitted to the medical department (adjusted HR 1.384 [95% CI, 1.153 to 1.662]). Pre-AKI is associated with increased mortality, longer hospital stay, and increased medical costs during admission. More attention should be paid to the clinical significance of pre-AKI.
| S-EPMC5132306 | biostudies-literature
Project description:BackgroundMachine learning and artificial intelligence have shown promising results in many areas and are driven by the increasing amount of available data. However, these data are often distributed across different institutions and cannot be easily shared owing to strict privacy regulations. Federated learning (FL) allows the training of distributed machine learning models without sharing sensitive data. In addition, the implementation is time-consuming and requires advanced programming skills and complex technical infrastructures.ObjectiveVarious tools and frameworks have been developed to simplify the development of FL algorithms and provide the necessary technical infrastructure. Although there are many high-quality frameworks, most focus only on a single application case or method. To our knowledge, there are no generic frameworks, meaning that the existing solutions are restricted to a particular type of algorithm or application field. Furthermore, most of these frameworks provide an application programming interface that needs programming knowledge. There is no collection of ready-to-use FL algorithms that are extendable and allow users (eg, researchers) without programming knowledge to apply FL. A central FL platform for both FL algorithm developers and users does not exist. This study aimed to address this gap and make FL available to everyone by developing FeatureCloud, an all-in-one platform for FL in biomedicine and beyond.MethodsThe FeatureCloud platform consists of 3 main components: a global frontend, a global backend, and a local controller. Our platform uses a Docker to separate the local acting components of the platform from the sensitive data systems. We evaluated our platform using 4 different algorithms on 5 data sets for both accuracy and runtime.ResultsFeatureCloud removes the complexity of distributed systems for developers and end users by providing a comprehensive platform for executing multi-institutional FL analyses and implementing FL algorithms. Through its integrated artificial intelligence store, federated algorithms can easily be published and reused by the community. To secure sensitive raw data, FeatureCloud supports privacy-enhancing technologies to secure the shared local models and assures high standards in data privacy to comply with the strict General Data Protection Regulation. Our evaluation shows that applications developed in FeatureCloud can produce highly similar results compared with centralized approaches and scale well for an increasing number of participating sites.ConclusionsFeatureCloud provides a ready-to-use platform that integrates the development and execution of FL algorithms while reducing the complexity to a minimum and removing the hurdles of federated infrastructure. Thus, we believe that it has the potential to greatly increase the accessibility of privacy-preserving and distributed data analyses in biomedicine and beyond.
| S-EPMC10372562 | biostudies-literature
Project description:BackgroundHospitalization-associated acute kidney injury (AKI), affecting one-in-five inpatients, is associated with increased mortality and major adverse cardiac/kidney endpoints. Early AKI risk stratification may enable closer monitoring and prevention. Given the complexity and resource utilization of existing machine learning models, we aimed to develop a simpler prediction model.MethodsModels were trained and validated to predict risk of AKI using electronic health record (EHR) data available at 24 h of inpatient admission. Input variables included demographics, laboratory values, medications, and comorbidities. Missing values were imputed using multiple imputation by chained equations.Results26,410 of 209,300 (12.6%) inpatients developed AKI during admission between 13 July 2012 and 11 July 2018. The area under the receiver operating characteristic curve (AUROC) was 0.86 for Random Forest and 0.85 for LASSO. Based on Youden's Index, a probability cutoff of >0.15 provided sensitivity and specificity of 0.80 and 0.79, respectively. AKI risk could be successfully predicted in 91% patients who required dialysis. The model predicted AKI an average of 2.3 days before it developed.ConclusionsThe proposed simpler machine learning model utilizing data available at 24 h of admission is promising for early AKI risk stratification. It requires external validation and evaluation of effects of risk prediction on clinician behavior and patient outcomes.
| S-EPMC9573390 | biostudies-literature