Machine learning enhances the performance of short and long-term mortality prediction model in non-ST-segment elevation myocardial infarction

Lee, Woojoo; Lee, Joongyub; Woo, Seoung Il; Choi, Seung Kang; Bae, Jang Whan; Jung, Seungpil; Jeong, Myung Ho; Lee, Won Kyung

doi:10.1038/s41598-021-92362-1

Cited by 28 publications

(23 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…But this can be alleviated by advancing the increasing use of digital strategies aimed at enhancing the participation of patients and healthcare providers in integrated healthcare management. [13][14][15][16][17] Incorporation of such ML models into electronic health records may help the scalability of prevention strategies and improved management in terms of healthcare cost savings and better quality of care.…”

Section: Discussionmentioning

confidence: 99%

Incident and recurrent myocardial infarction (MI) in relation to comorbidities: Prediction of outcomes using machine‐learning algorithms

Lip

Genaidy

Tran³

et al. 2022

Eur J Clin Investigation

View full text Add to dashboard Cite

Background To date, incident and recurrent MI remains a major health issue worldwide, and efforts to improve risk prediction in population health studies are needed. This may help the scalability of prevention strategies and management in terms of healthcare cost savings and improved quality of care. Methods We studied a large‐scale population of 4.3 million US patients from different socio‐economic and geographical areas from three health plans (Commercial, Medicare, Medicaid). Individuals had medical/pharmacy benefits for at least 30 months (2 years for comorbid history and followed up for 6 months or more for clinical outcomes). Machine‐learning (ML) algorithms included supervised (logistic regression, neural network) and unsupervised (decision tree, gradient boosting) methodologies. Model discriminant validity, calibration and clinical utility were performed separately on allocated test sample (1/3 of original data). Results In the absence of MI in comorbid history, the overall incidence rates were 0.442 cases/100 person‐years and in the presence of MI history, 0.652. ML algorithms showed that supervised formulations had incrementally higher discriminant validity than unsupervised techniques (e.g., for incident MI outcome in the absence of MI in comorbid history: logistic regression “LR” – c index 0.921, 95%CI 0.920–0.922; neural network “NN” – c index 0.914, 95%CI 0.913–0.915; gradient boosting “GB” – c index 0.902, 95%CI 0.900–0.904; decision tree “DT” – c index 0.500, 95%CI 0.495–0.505). Calibration and clinical utility showed good to excellent results. Conclusion ML algorithms can substantially improve the prediction of incident and recurrent MI particularly in terms of the non‐linear formulation. This approach may help with improved risk prediction, allowing implementation of cardiovascular prevention strategies across diversified sub‐populations with different clusters of complexity.

show abstract

Section: Discussionmentioning

confidence: 99%

Incident and recurrent myocardial infarction (MI) in relation to comorbidities: Prediction of outcomes using machine‐learning algorithms

Lip

Genaidy

Tran³

et al. 2022

Eur J Clin Investigation

View full text Add to dashboard Cite

show abstract

“…By developing ML solutions as hands-on approaches to precisely predict in-real time the disease prognosis, this direction empowers robust personalized medicine decisions for specific populations, or individual patients, in the clinical setting. Previous efforts aiming to advance the application of ML prediction models in CVDs have been previously executed in patients suffering from ST-elevated myocardial infarction (STEMI) ( 30 ) and non-ST-segment elevation myocardial infarction (non-STEMI) ( 31 ).…”

Section: Discussion and Future Workmentioning

confidence: 99%

A Risk-Stratification Machine Learning Framework for the Prediction of Coronary Artery Disease Severity: Insights From the GESS Trial

Mittas

Chatzopoulou

Kyritsis

et al. 2022

Front. Cardiovasc. Med.

View full text Add to dashboard Cite

Our study aims to develop a data-driven framework utilizing heterogenous electronic medical and clinical records and advanced Machine Learning (ML) approaches for: (i) the identification of critical risk factors affecting the complexity of Coronary Artery Disease (CAD), as assessed via the SYNTAX score; and (ii) the development of ML prediction models for accurate estimation of the expected SYNTAX score. We propose a two-part modeling technique separating the process into two distinct phases: (a) a binary classification task for predicting, whether a patient is more likely to present with a non-zero SYNTAX score; and (b) a regression task to predict the expected SYNTAX score accountable to individual patients with a non-zero SYNTAX score. The framework is based on data collected from the GESS trial (NCT03150680) comprising electronic medical and clinical records for 303 adult patients with suspected CAD, having undergone invasive coronary angiography in AHEPA University Hospital of Thessaloniki, Greece. The deployment of the proposed approach demonstrated that atherogenic index of plasma levels, diabetes mellitus and hypertension can be considered as important risk factors for discriminating patients into zero- and non-zero SYNTAX score groups, whereas diastolic and systolic arterial blood pressure, peripheral vascular disease and body mass index can be considered as significant risk factors for providing an accurate estimation of the expected SYNTAX score, given that a patient belongs to the non-zero SYNTAX score group. The experimental findings utilizing the identified set of important risk factors indicate a sufficient prediction performance for the Support Vector Machine model (classification task) with an F-measure score of ~0.71 and the Support Vector Regression model (regression task) with a median absolute error value of ~6.5. The proposed data-driven framework described herein present evidence of the prediction capacity and the potential clinical usefulness of the developed risk-stratification models. However, further experimentation in a larger clinical setting is needed to ensure the practical utility of the presented models in a way to contribute to a more personalized management and counseling of CAD patients.

show abstract

“…In 2021, Lee et al constructed machine learning methods by logistic regression with regularization, random forest, support vector machine, and extreme gradient boosting to construct 3-month, 12-month, and in-hospital mortality prediction models for AMI patients. The results showed that machine learning outperformed traditional prediction models [ 46 ]. In 2022, Xiao et al employed six machine learning methods to construct predictive models for the occurrence of major adverse cardiovascular events (MACEs) in AMI patients.…”

Section: Discussionmentioning

confidence: 99%

A simple APACHE IV risk dynamic nomogram that incorporates early admitted lactate for the initial assessment of 28-day mortality in critically ill patients with acute myocardial infarction

Song

Yan

et al. 2022

BMC Cardiovasc Disord

View full text Add to dashboard Cite

Background Early risk stratification is important for patients with acute myocardial infarction (AMI). We aimed to develop a simple APACHE IV dynamic nomogram, combined with easily available clinical parameters within 24 h of admission, thus improving its predictive power to assess the risk of mortality at 28 days. Methods Clinical information on AMI patients was extracted from the eICU database v2.0. A preliminary XGBoost examination of the degree of association between all variables in the database and 28-day mortality was conducted. Univariate and multivariate logistic regression analysis were used to perform screening of variables. Based on the multifactorial analysis, a dynamic nomogram predicting 28-day mortality in these patients was developed. To cope with missing data in records with missing variables, we applied the multiple imputation method. Predictive models are evaluated in three main areas, namely discrimination, calibration, and clinical validity. The discrimination is mainly represented by the area under the receiver operating characteristic curve (AUC), net reclassification improvement (NRI) and integrated discrimination improvement (IDI). Calibration is represented by the calibration plot. Clinical validity is represented by the decision curve analysis (DCA) curve. Results A total of 504 people were included in the study. All 504 people were used to build the predictive model, and the internal validation model used a 500-bootstrap method. Multivariate analysis showed that four variables, APACHE IV, the first sample of admission lactate, prior atrial fibrillation (AF), and gender, were included in the nomogram as independent predictors of 28-day mortality in AMI. The prediction model had an AUC of 0.819 (95%CI 0.770–0.868) whereas the internal validation model had an AUC of 0.814 (95%CI 0.765–0.860). Calibration and DCA curves indicated that the dynamic nomogram in this study were reflective of real-world conditions and could be applied clinically. The predictive model composed of these four variables outperformed a single APACHE IV in terms of NRI and IDI. The NRI was 16.4% (95% CI: 6.1–26.8%; p = 0.0019) and the IDI was 16.4% (95% CI: 6.0–26.8%; p = 0.0020). Lactate accounted for nearly half of the total NRI, which showed that lactate was the most important of the other three variables. Conclusion The prediction model constructed by APACHE IV in combination with the first sample of admission lactate, prior AF, and gender outperformed the APACHE IV scoring system alone in predicting 28-day mortality in AMI. The prediction dynamic nomogram model was published via a website app, allowing clinicians to improve the predictive efficacy of the APACHE IV score by 16.4% in less than 1 min.

show abstract

Machine learning enhances the performance of short and long-term mortality prediction model in non-ST-segment elevation myocardial infarction

Cited by 28 publications

References 32 publications

Incident and recurrent myocardial infarction (MI) in relation to comorbidities: Prediction of outcomes using machine‐learning algorithms

Incident and recurrent myocardial infarction (MI) in relation to comorbidities: Prediction of outcomes using machine‐learning algorithms

A Risk-Stratification Machine Learning Framework for the Prediction of Coronary Artery Disease Severity: Insights From the GESS Trial

A simple APACHE IV risk dynamic nomogram that incorporates early admitted lactate for the initial assessment of 28-day mortality in critically ill patients with acute myocardial infarction

Contact Info

Product

Resources

About