Dataset Information

Machine learning to predict venous thrombosis in acutely ill medical patients.

ABSTRACT: Background:The identification of acutely ill patients at high risk for venous thromboembolism (VTE) may be determined clinically or by use of integer-based scoring systems. These scores demonstrated modest performance in external data sets. Objectives:To evaluate the performance of machine learning models compared to the IMPROVE score. Methods:The APEX trial randomized 7513 acutely medically ill patients to extended duration betrixaban vs. enoxaparin. Including 68 variables, a super learner model (ML) was built to predict VTE by combining estimates from 5 families of candidate models. A "reduced" model (rML) was also developed using 16 variables that were thought, a priori, to be associated with VTE. The IMPROVE score was calculated for each patient. Model performance was assessed by discrimination and calibration to predict a composite VTE end point. The frequency of predicted risks of VTE were plotted and divided into tertiles. VTE risks were compared across tertiles. Results:The ML and rML algorithms outperformed the IMPROVE score in predicting VTE (c-statistic: 0.69, 0.68 and 0.59, respectively). The Hosmer-Lemeshow goodness-of-fit P-value was 0.06 for ML, 0.44 for rML, and <0.001 for the IMPROVE score. The observed event rate in the lowest tertile was 2.5%, 4.8% in tertile 2, and 11.4% in the highest tertile. Patients in the highest tertile of VTE risk had a 5-fold increase in odds of VTE compared to the lowest tertile. Conclusion:The super learner algorithms improved discrimination and calibration compared to the IMPROVE score for predicting VTE in acute medically ill patients.

SUBMITTER: Nafee T

PROVIDER: S-EPMC7040551 | biostudies-literature | 2020 Feb

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Machine learning to predict venous thrombosis in acutely ill medical patients.

Nafee Tarek T Gibson C Michael CM Travis Ryan R Yee Megan K MK Kerneis Mathieu M Chi Gerald G AlKhalfan Fahad F Hernandez Adrian F AF Hull Russell D RD Cohen Ander T AT Harrington Robert A RA Goldhaber Samuel Z SZ

Research and practice in thrombosis and haemostasis 20200121 2

<h4>Background</h4>The identification of acutely ill patients at high risk for venous thromboembolism (VTE) may be determined clinically or by use of integer-based scoring systems. These scores demonstrated modest performance in external data sets.<h4>Objectives</h4>To evaluate the performance of machine learning models compared to the IMPROVE score.<h4>Methods</h4>The APEX trial randomized 7513 acutely medically ill patients to extended duration betrixaban vs. enoxaparin. Including 68 variables ...[more]

PMID: 32110753

Similar Datasets

Project description:BackgroundAcutely ill medical patients experience deep venous thrombosis (DVT) during the hospitalization, however the time course of DVT is still unclear.ObjectivesTo evaluate risk factors in acutely ill hospitalized medical patients for proximal asymptomatic DVT (ADVT) and symptomatic DVT (SDVT) at admission and discharge.Patients/methodsIn this prospective observational study, consecutive acutely ill medical patients (hospitalized mainly for acute medical disease as infections, neoplasm, anemia, heart failure) underwent compression ultrasonography (CUS) of proximal lower limb veins within 48 h from admission and at discharge to diagnose ADVT and SDVT. Covid-19 patients, anticoagulant therapy, surgical procedures, acute SDVT, and acute pulmonary embolism, were exclusion criteria. Biographical characteristics at hospitalization, D-Dimer (assessed by ELISA)) and DD-improve score.ResultsOf 2,100 patients (1002 females, 998 males, age 71 ± 16 years) 58 (2.7%) had proximal ADVT at admission. Logistic regression analysis showed that age, and active cancer were independently associated with ADVT at admission. The median length of hospitalization was 10 days [interquartile range: 6-15]. During the hospital stay, 6 patients (0.3%) with a negative CUS at admission experienced DVT (2 SDVT and 4 ADVT). In the subgroup of patients (n = 1118), in whom D-dimer was measured at admission, D-Dimer and IMPROVE-DD score were associated with ADVT at admission (n = 37) and with all DVT (n = 42) at discharge. ROC curve defined an IMPROVE-DD score of 2.5 as the optimal cut-off for discriminating patients with and without thrombotic events.ConclusionsWe provide evidence of early development of ADVT in unselected acutely ill medical patients suggesting the need of investigating patients by CUS immediately after hospital admission (within 48 h). Advanced age, active cancer, known thrombophilia and increased IMPROVE-DD score may identify patients at risk. The benefit of anticoagulation needs to be investigated in patients with these specific risk factors and negative CUS at admission.Trial registrationNCT03157843.

Project description:The ADA (Age-D-dimer-Albumin) score was developed to identify hospitalized patients at an increased risk for thrombosis in the coronavirus infectious disease-19 (COVID-19) setting. The study aimed to validate the ADA score for predicting thrombosis in a non-COVID-19 medically ill population from the APEX trial. The APEX trial was a multinational, randomized trial that evaluated the efficacy and safety of betrixaban vs. enoxaparin among acutely ill hospitalized patients at risk for venous thromboembolism. The study endpoints included the composite of arterial or venous thrombosis and its components. Metrics of model calibration and discrimination were computed for assessing the performance of the ADA score as compared to the IMPROVE score, a well-validated VTE risk assessment model. Among 7,119 medical inpatients, 209 (2.9%) had a thrombosis event up to 77 days of follow-up. The ADA score demonstrated good calibration for both arterial and venous thrombosis, whereas the IMPROVE score had adequate calibration for venous thrombosis (p > 0.05 from the Hosmer-Lemeshow test). For discriminating arterial and venous thrombosis, there was no significant difference between the ADA vs. IMPROVE score (c statistic = 0.620 [95% CI: 0.582 to 0.657] vs. 0.590 [95% CI: 0.556 to 0.624]; ∆ c statistic = 0.030 [95% CI: -0.022 to 0.081]; p = 0.255). Similarly, for discriminating arterial thrombosis, there was no significant difference between the ADA vs. IMPROVE score (c statistic = 0.582 [95% CI: 0.534 to 0.629] vs. 0.609 [95% CI: 0.564 to 0.653]; ∆ c statistic = -0.027 [95% CI: -0.091 to 0.036]; p = 0.397). For discriminating venous thrombosis, the ADA score was modestly superior to the IMPROVE score (c statistic = 0.664 [95% CI: 0.607 to 0.722] vs. 0.573 [95% CI: 0.521 to 0.624]; ∆ c statistic = 0.091 [95% CI: 0.011 to 0.172]; p = 0.026). The ADA score had a higher sensitivity (0.579 [95% CI: 0.512 to 0.646]; vs. 0.440 [95% CI: 0.373 to 0.507]) but lower specificity (0.625 [95% CI: 0.614 to 0.637] vs. 0.747 [95% CI: 0.737 to 0.758]) than the IMPROVE score for predicting thrombosis. Among acutely ill hospitalized medical patients enrolled in the APEX trial, the ADA score demonstrated good calibration but suboptimal discrimination for predicting thrombosis. The findings support the use of either the ADA or IMPROVE score for thrombosis risk assessment. The applicability of the ADA score to non-COVID-19 populations warrants further research.Clinical Trial Registration: http://www.clinicaltrials.gov . Unique identifier: NCT01583218.

Project description:ObjectiveTo assess the benefits and harms of different types and doses of anticoagulant drugs for the prevention of venous thromboembolism in patients who are acutely ill and admitted to hospital.DesignSystematic review and network meta-analysis.Data sourcesCochrane CENTRAL, PubMed/Medline, Embase, Web of Science, clinical trial registries, and national health authority databases. The search was last updated on 16 November 2021.Eligibility criteria for selecting studiesPublished and unpublished randomised controlled trials that evaluated low or intermediate dose low-molecular-weight heparin, low or intermediate dose unfractionated heparin, direct oral anticoagulants, pentasaccharides, placebo, or no intervention for the prevention of venous thromboembolism in acutely ill adult patients in hospital.Main outcome measuresRandom effects, bayesian network meta-analyses used four co-primary outcomes: all cause mortality, symptomatic venous thromboembolism, major bleeding, and serious adverse events at or closest timing to 90 days. Risk of bias was also assessed using the Cochrane risk-of-bias 2.0 tool. The quality of evidence was graded using the Confidence in Network Meta-Analysis framework.Results44 randomised controlled trials that randomly assigned 90 095 participants were included in the main analysis. Evidence of low to moderate quality suggested none of the interventions reduced all cause mortality compared with placebo. Pentasaccharides (odds ratio 0.32, 95% credible interval 0.08 to 1.07), intermediate dose low-molecular-weight heparin (0.66, 0.46 to 0.93), direct oral anticoagulants (0.68, 0.33 to 1.34), and intermediate dose unfractionated heparin (0.71, 0.43 to 1.19) were most likely to reduce symptomatic venous thromboembolism (very low to low quality evidence). Intermediate dose unfractionated heparin (2.63, 1.00 to 6.21) and direct oral anticoagulants (2.31, 0.82 to 6.47) were most likely to increase major bleeding (low to moderate quality evidence). No conclusive differences were noted between interventions regarding serious adverse events (very low to low quality evidence). When compared with no intervention instead of placebo, all active interventions did more favourably with regard to risk of venous thromboembolism and mortality, and less favourably with regard to risk of major bleeding. The results were robust in prespecified sensitivity and subgroup analyses.ConclusionsLow-molecular-weight heparin in an intermediate dose appears to confer the best balance of benefits and harms for prevention of venous thromboembolism. Unfractionated heparin, in particular the intermediate dose, and direct oral anticoagulants had the least favourable profile. A systematic discrepancy was noted in intervention effects that depended on whether placebo or no intervention was the reference treatment. Main limitations of this study include the quality of the evidence, which was generally low to moderate due to imprecision and within-study bias, and statistical inconsistency, which was addressed post hoc.Systematic review registrationPROSPERO CRD42020173088.

Dataset Information

Machine learning to predict venous thrombosis in acutely ill medical patients.

Publications

Machine learning to predict venous thrombosis in acutely ill medical patients.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets