Dataset Information

Improving dynamic stroke risk prediction in non-anticoagulated patients with and without atrial fibrillation: comparing common clinical risk scores and machine learning algorithms.

ABSTRACT:

Aims

Diversified cardiovascular/non-cardiovascular multi-morbid risk and efficient machine learning algorithms may facilitate improvements in stroke risk prediction, especially in newly diagnosed non-anticoagulated atrial fibrillation (AF) patients where initial decision-making on stroke prevention is needed. Therefore the aims of this article are to study common clinical risk assessment for stroke risk prediction in AF/non-AF cohorts together with cardiovascular/ non-cardiovascular multi-morbid conditions; to improve stroke risk prediction using machine learning approaches; and to compare the improved clinical prediction rules for multi-morbid conditions using machine learning algorithms.

Methods and results

We used cohort data from two health plans with 6 457 412 males/females contributing 14,188,679 person-years of data. The model inputs consisted of a diversified list of comorbidities/demographic/ temporal exposure variables, with the outcome capturing stroke event incidences. Machine learning algorithms used two parametric and two nonparametric techniques. The best prediction model was derived on the basis of non-linear formulations using machine learning criteria, with the highest c-index was obtained for logistic regression [0.892; 95% confidence interval (CI) 0.886-0.898] with consistency on external validation (0.891; 95% CI 0.882-0.9). These were significantly higher than those based on the conventional stroke risk scores (CHADS2: 0.7488, 95% CI 0.746-0.7516; CHA2DS2-VASc: 0.7801, 95% CI 0.7772-0.7831) and multi-morbid index (0.8508, 95% CI 0.8483-0.8532). The machine learning algorithm had good internal and external calibration and net benefit values.

Conclusion

In this large cohort of newly diagnosed non-anticoagulated AF/non-AF patients, large improvements in stroke risk prediction can be shown with cardiovascular/non-cardiovascular multi-morbid index and a machine learning approach accounting for dynamic changes in risk factors.

SUBMITTER: Lip GYH

PROVIDER: S-EPMC9382661 | biostudies-literature | 2022 Aug

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Improving dynamic stroke risk prediction in non-anticoagulated patients with and without atrial fibrillation: comparing common clinical risk scores and machine learning algorithms.

Lip Gregory Y H GYH Tran George G Genaidy Ash A Marroquin Patricia P Estes Cara C Landsheft Jeremy J

European heart journal. Quality of care & clinical outcomes 20220801 5

<h4>Aims</h4>Diversified cardiovascular/non-cardiovascular multi-morbid risk and efficient machine learning algorithms may facilitate improvements in stroke risk prediction, especially in newly diagnosed non-anticoagulated atrial fibrillation (AF) patients where initial decision-making on stroke prevention is needed. Therefore the aims of this article are to study common clinical risk assessment for stroke risk prediction in AF/non-AF cohorts together with cardiovascular/ non-cardiovascular mult ...[more]

PMID: 33999139

Similar Datasets

Project description:BackgroundDespite direct oral anticoagulants (DOACs) being safer than warfarin for stroke prevention in atrial fibrillation (AF), major bleeding concerns persist. Most bleeding risk scores predate DOAC approval.ObjectivesThis study aimed to compare the Age, history of Bleeding, and non-bleeding related Hospitalisation [ABH] score's performance-derived for DOAC-treated patients-with those of 5 other scores (Anticoagulation and Risk Factors in Atrial Fibrillation [ATRIA], Hypertension, Abnormal renal/liver function, Stroke, Bleeding history or predisposition, Labile international normalized ratio, Elderly [>65 years], Drugs/alcohol concomitantly [HAS-BLED], Hepatic, Hepatic or Renal Disease, Ethanol Abuse, Malignancy, Older Age, Reduced Platelet Count or Function, Re-Bleeding, Hypertension, Anemia, Genetic Factors, Excessive Fall Risk and Stroke [HEMORR2HAGES], Outcomes Registry for Better Informed Treatment of Atrial Fibrillation [ORBIT-AF], and Congestive heart failure, Hypertension, Age ≥75 [doubled], Diabetes, Stroke [doubled]-Vascular disease, Age 65-74, Sex category [CHA2DS2-VASc]) in predicting DOAC-related major bleeding in patients with AF.MethodsIn this retrospective study of 2364 patients with nonvalvular AF on rivaroxaban or apixaban (median age, 68.3 years; 32.1% women), International Society on Thrombosis and Haemostasis-defined major bleeding (incidence, 4.1%; n = 97) was analyzed. C-statistics from time-dependent receiver operating characteristic (ROC) curves for continuous risk scores were the primary comparison metric, but other metrics, such as decision curves, were also compared.ResultsAt 100 days, C-statistics were highest for ORBIT-AF and ATRIA (0.62 and 0.61, respectively, with other scores having an area under the ROC curve of <0.60); some significant differences favored ORBIT-AF. At 1100 days, C-statistics remained highest for ORBIT-AF and ATRIA (0.62 and 0.61, respectively, with other scores having an area under the ROC curve of <0.60 again), and ORBIT-AF had significantly higher C-statistics than those for all other risk scores (P < .05), except for ATRIA. At 2100 days, all C-statistics were <0.60 with no significant differences. Decision curves showed the greatest net benefit for ORBIT-AF and ATRIA at 100 days and for ATRIA at 1100 days, with no discernible net benefit for any of the scores at 2100 days.ConclusionORBIT-AF and ATRIA provided the best bleeding risk prediction within the first 1100 days. None of the 6 bleeding risk scores provided predictive benefit over 2100 days of DOAC treatment.

Project description:Background Patients with atrial fibrillation (AF) treated with oral anticoagulants may be exposed to an increased risk of bleeding events. The HAS-BLED (Hypertension, Abnormal renal and liver function, Stroke, Bleeding, Labile INRs, Elderly, Drugs or alcohol) score is a simple, well-established, clinical bleeding-risk prediction score. Recently, a new algorithm-based score was proposed, the GARFIELD-AF (Global Anticoagulant in the Field-AF) bleeding score. We compared HAS-BLED and GARFIELD-AF scores in predicting adjudicated bleeding events in a clinical trial cohort of patients with AF taking anticoagulants, in the first external comparative validation of both scores. Methods and Results We analyzed patients from the SPORTIF (Stroke Prevention Using an Oral Thrombin Inhibitor in Patients With AF) III and V trials. All patients assigned to the warfarin arm with information to calculate the scores were considered. Outcomes were major, major/clinically relevant nonmajor, and any bleeding. A total of 3550 warfarin-treated patients were available for analysis. Of these patients, 2519 (71.0%) had a HAS-BLED score ≥3, whereas based on GARFIELD-AF median value, 2056 (57.9%) were categorized as "high score." Both HAS-BLED and GARFIELD-AF C-indexes showed modest predictive value (C-index [95% confidence interval] for major bleeding, 0.58 [0.56-0.60] and 0.56 [0.54-0.57], respectively); however, GARFIELD-AF was not predictive of any bleeding. The GARFIELD-AF bleeding score had a significantly lower sensitivity and a negative reclassification for any bleeding compared with HAS-BLED, assessed by integrated discrimination improvement and net reclassification improvement (both P<0.001). HAS-BLED showed a 5% net benefit for any bleeding occurrence. Conclusions The algorithm-based GARFIELD-AF bleeding score did not show any significant improvement in major and major/clinically relevant nonmajor prediction compared with the simple HAS-BLED score. For clinical usefulness in prediction of any bleeding, the HAS-BLED score showed a significant net benefit compared with the GARFIELD-AF.

Dataset Information

Improving dynamic stroke risk prediction in non-anticoagulated patients with and without atrial fibrillation: comparing common clinical risk scores and machine learning algorithms.

Aims

Methods and results

Conclusion

Publications

Improving dynamic stroke risk prediction in non-anticoagulated patients with and without atrial fibrillation: comparing common clinical risk scores and machine learning algorithms.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets