Dataset Information

New artificial intelligence prediction model using serial prothrombin time international normalized ratio measurements in atrial fibrillation patients on vitamin K antagonists: GARFIELD-AF.

ABSTRACT: Aims:Most clinical risk stratification models are based on measurement at a single time-point rather than serial measurements. Artificial intelligence (AI) is able to predict one-dimensional outcomes from multi-dimensional datasets. Using data from Global Anticoagulant Registry in the Field (GARFIELD)-AF registry, a new AI model was developed for predicting clinical outcomes in atrial fibrillation (AF) patients up to 1 year based on sequential measures of prothrombin time international normalized ratio (PT-INR) within 30 days of enrolment.

Methods and results:Patients with newly diagnosed AF who were treated with vitamin K antagonists (VKAs) and had at least three measurements of PT-INR taken over the first 30 days after prescription were analysed. The AI model was constructed with multilayer neural network including long short-term memory and one-dimensional convolution layers. The neural network was trained using PT-INR measurements within days 0-30 after starting treatment and clinical outcomes over days 31-365 in a derivation cohort (cohorts 1-3; n = 3185). Accuracy of the AI model at predicting major bleed, stroke/systemic embolism (SE), and death was assessed in a validation cohort (cohorts 4-5; n = 1523). The model's c-statistic for predicting major bleed, stroke/SE, and all-cause death was 0.75, 0.70, and 0.61, respectively.

Conclusions:Using serial PT-INR values collected within 1 month after starting VKA, the new AI model performed better than time in therapeutic range at predicting clinical outcomes occurring up to 12 months thereafter. Serial PT-INR values contain important information that can be analysed by computer to help predict adverse clinical outcomes.

SUBMITTER: Goto S

PROVIDER: S-EPMC7556811 | biostudies-literature | 2020 Sep

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

New artificial intelligence prediction model using serial prothrombin time international normalized ratio measurements in atrial fibrillation patients on vitamin K antagonists: GARFIELD-AF.

Goto Shinichi S Goto Shinya S Pieper Karen S KS Bassand Jean-Pierre JP Camm Alan John AJ Fitzmaurice David A DA Goldhaber Samuel Z SZ Haas Sylvia S Parkhomenko Alexander A Oto Ali A Misselwitz Frank F Turpie Alexander G G AGG Verheugt Freek W A FWA Fox Keith A A KAA Gersh Bernard J BJ Kakkar Ajay K AK

European heart journal. Cardiovascular pharmacotherapy 20200901 5

<h4>Aims</h4>Most clinical risk stratification models are based on measurement at a single time-point rather than serial measurements. Artificial intelligence (AI) is able to predict one-dimensional outcomes from multi-dimensional datasets. Using data from Global Anticoagulant Registry in the Field (GARFIELD)-AF registry, a new AI model was developed for predicting clinical outcomes in atrial fibrillation (AF) patients up to 1 year based on sequential measures of prothrombin time international n ...[more]

PMID: 31821482

Similar Datasets

Project description:ObjectiveThe study investigates the diagnostic and prognostic significance of the prothrombin time/international normalized ratio (PT/INR) in patients with sepsis and septic shock.BackgroundSepsis may be complicated by disseminated intravascular coagulation (DIC). While the status of coagulopathy of septic patients is represented within the sepsis-3 definition by assessing the platelet count, less data regarding the prognostic impact of the PT/INR in patients admitted with sepsis and septic shock is available.MethodsConsecutive patients with sepsis and septic shock from 2019 to 2021 were included. Blood samples were retrieved from day of disease onset (ie, day 0), as well as on day 1, 2, 4, 6 and 9 thereafter. Firstly, the diagnostic value of the PT/INR in comparison to the activated partial thromboplastin time (aPTT) was tested for septic shock compared to sepsis without shock. Secondly, the prognostic value of the PT/INR for 30-day all-cause mortality was tested. Statistical analyses included univariable t-tests, Spearman's correlations, C-statistics, Kaplan-Meier analyses and Cox proportional regression analyses.Results338 patients were included (56% sepsis without shock, 44% septic shock). The overall rate of all-cause mortality at 30 days was 52%. With an area under the curve (AUC) of 0.682 (p= .001) on day 0, the PT/INR revealed moderate discrimination of septic shock and sepsis without shock. Furthermore, PT/ INR was able to discriminate non-survivors and survivors at 30 days (AUC = 0.612; p = .001). Patients with a PT/INR >1.5 had higher rates of 30-day all-cause mortality than patients with lower values (mortality rate 73% vs 48%; log rank p = .001; HR = 2.129; 95% CI 1.494-3.033; p = .001), even after multivariable adjustment (HR = 1.793; 95% CI 1.343-2.392; p = .001). Increased risk of 30-day all-cause mortality was observed irrespective of concomitant thrombocytopenia.ConclusionThe PT/INR revealed moderate diagnostic accuracy for septic shock but was associated with reliable prognostic accuracy with regard to 30-day all-cause mortality in patients admitted with sepsis and septic shock.

Project description:BackgroundLife-long vitamin K antagonist (VKA) therapy is recommended as a standard of care in antiphospholipid syndrome (APS) patients with thrombosis. Concerns have been raised about the validity of international normalized ratio (INR) measurements in lupus anticoagulant (LA)-positive APS patients because LA may interfere with phospholipid-dependent coagulation tests and could elevate INR measurements.ObjectivesHere, we aimed to determine the interference of antigen-specific monoclonal and isolated patient antibodies with LA activity on INR measurements.MethodsPooled normal plasma and control plasma from patients on VKA (without LA) were incubated with monoclonal and isolated patient immunoglobulin G antiprothrombin and anti-beta-2-glycoprotein I antibodies that express LA activity. INR was determined before and after addition using 3 laboratory assays (Owren STA-Hepato Prest, Quick STA-NeoPTimal, and Quick STA-Neoplastine R) and 1 point-of-care test device (CoaguChek Pro II).ResultsAntiprothrombin and anti-beta-2-glycoprotein I antibodies with LA activity interfered with recombinant human thromboplastin reagents (Quick STA-Neoplastine R and CoaguChek Pro II), particularly when added to plasma of VKA-treated controls. This effect was most evident on point-of-care test INR measurements, while the recombinant Quick reagent exhibited a lesser degree of interference. In contrast, tissue-derived thromboplastin reagents (Owren STA-Hepato Prest and Quick STA-NeoPTimal) remained largely unaffected by these antibodies, both in pooled normal plasma and VKA anticoagulated control plasma. Among these reagents, the Owren INR reagent exhibited the lowest sensitivity to the influence of LA antibodies. This observed difference in sensitivity is independent of the plasma dilution factor or the presence of factor V or fibrinogen in Owren reagent.ConclusionINR reagents that utilize recombinant human thromboplastin are more sensitive to the presence of monoclonal and patient-derived antibodies with LA activity. Consequently, APS patients positive for LA should be monitored using tissue-derived thromboplastin reagents, given its reduced susceptibility to interference by LA-causing antibodies.

Dataset Information

New artificial intelligence prediction model using serial prothrombin time international normalized ratio measurements in atrial fibrillation patients on vitamin K antagonists: GARFIELD-AF.

Publications

New artificial intelligence prediction model using serial prothrombin time international normalized ratio measurements in atrial fibrillation patients on vitamin K antagonists: GARFIELD-AF.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets