Dataset Information

A Machine Learning Model for Predicting In-Hospital Mortality in Chinese Patients With ST-Segment Elevation Myocardial Infarction: Findings From the China Myocardial Infarction Registry.

ABSTRACT:

Background

Machine learning (ML) risk prediction models, although much more accurate than traditional statistical methods, are inconvenient to use in clinical practice due to their nontransparency and requirement of a large number of input variables.

Objective

We aimed to develop a precise, explainable, and flexible ML model to predict the risk of in-hospital mortality in patients with ST-segment elevation myocardial infarction (STEMI).

Methods

This study recruited 18,744 patients enrolled in the 2013 China Acute Myocardial Infarction (CAMI) registry and 12,018 patients from the China Patient-Centered Evaluative Assessment of Cardiac Events (PEACE)-Retrospective Acute Myocardial Infarction Study. The Extreme Gradient Boosting (XGBoost) model was derived from 9616 patients in the CAMI registry (2014, 89 variables) with 5-fold cross-validation and validated on both the 9125 patients in the CAMI registry (89 variables) and the independent China PEACE cohort (10 variables). The Shapley Additive Explanations (SHAP) approach was employed to interpret the complex relationships embedded in the proposed model.

Results

In the XGBoost model for predicting all-cause in-hospital mortality, the variables with the top 8 most important scores were age, left ventricular ejection fraction, Killip class, heart rate, creatinine, blood glucose, white blood cell count, and use of angiotensin-converting enzyme inhibitors (ACEIs) and angiotensin II receptor blockers (ARBs). The area under the curve (AUC) on the CAMI validation set was 0.896 (95% CI 0.884-0.909), significantly higher than the previous models. The AUC for the Global Registry of Acute Coronary Events (GRACE) model was 0.809 (95% CI 0.790-0.828), and for the TIMI model, it was 0.782 (95% CI 0.763-0.800). Despite the China PEACE validation set only having 10 available variables, the AUC reached 0.840 (0.829-0.852), showing a substantial improvement to the GRACE (0.762, 95% CI 0.748-0.776) and TIMI (0.789, 95% CI 0.776-0.803) scores. Several novel and nonlinear relationships were discovered between patients' characteristics and in-hospital mortality, including a U-shape pattern of high-density lipoprotein cholesterol (HDL-C).

Conclusions

The proposed ML risk prediction model was highly accurate in predicting in-hospital mortality. Its flexible and explainable characteristics make the model convenient to use in clinical practice and could help guide patient management.

Trial registration

ClinicalTrials.gov NCT01874691; https://clinicaltrials.gov/study/NCT01874691.

SUBMITTER: Yang J

PROVIDER: S-EPMC11322712 | biostudies-literature | 2024 Jul

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

A Machine Learning Model for Predicting In-Hospital Mortality in Chinese Patients With ST-Segment Elevation Myocardial Infarction: Findings From the China Myocardial Infarction Registry.

Yang Jingang J Li Yingxue Y Li Xiang X Tao Shuiying S Zhang Yuan Y Chen Tiange T Xie Guotong G Xu Haiyan H Gao Xiaojin X Yang Yuejin Y

Journal of medical Internet research 20240730

<h4>Background</h4>Machine learning (ML) risk prediction models, although much more accurate than traditional statistical methods, are inconvenient to use in clinical practice due to their nontransparency and requirement of a large number of input variables.<h4>Objective</h4>We aimed to develop a precise, explainable, and flexible ML model to predict the risk of in-hospital mortality in patients with ST-segment elevation myocardial infarction (STEMI).<h4>Methods</h4>This study recruited 18,744 p ...[more]

PMID: 39079111

Similar Datasets

Project description:ObjectivesCompared with ST-segment elevation myocardial infarction (STEMI) patients, non-STEMI (NSTEMI) patients have more comorbidities and extensive coronary artery disease. Contemporary comparative data on the long-term prognosis of stable post-myocardial infarction subtypes are needed.DesignLong-Term rIsk, clinical manaGement and healthcare Resource utilisation of stable coronary artery dISease (TIGRIS) was a multinational, observational and longitudinal cohort study.SettingPatients were enrolled from 350 centres, with >95% coming from cardiology practices across 24 countries, from 19 June 2013 to 31 March 2017.ParticipantsThis study enrolled 8277 stable patients 1-3 years after myocardial infarction with ≥1 additional risk factor.Outcome measuresOver a 2 year follow-up, cardiovascular events and deaths and self-reported health using the EuroQol 5-dimension questionnaire score were recorded. Relative risk of clinical events and health resource utilisation in STEMI and NSTEMI patients were compared using multivariable Poisson regression models, adjusting for prognostically relevant patient factors.ResultsOf 7752 patients with known myocardial infarction type, 46% had NSTEMI; NSTEMI patients were older with more comorbidities than STEMI patients. NSTEMI patients had significantly poorer self-reported health and lower prevalence of dual antiplatelet therapy at hospital discharge and at enrolment 1-3 years later. NSTEMI patients had a higher incidence of combined myocardial infarction, stroke and cardiovascular death (5.6% vs 3.9%, p<0.001) and higher all-cause mortality (4.2% vs 2.6%, p<0.001) compared with STEMI patients. Risks were attenuated after adjusting for other patient characteristics. Health resource utilisation was higher in NSTEMI patients, although STEMI patients had more cardiologist visits.ConclusionsPost-NSTEMI chronic coronary syndrome patients had a less favourable risk factor profile, poorer self-reported health and more adverse cardiovascular events during long-term follow-up than individuals post STEMI. Efforts are needed to recognise the risks of stable patients after NSTEMI and optimise secondary prevention and care.Trial registration numberNCT01866904.

Project description:ImportanceChallenges to improving ST-segment elevation myocardial infarction (STEMI) care are formidable in low- to middle-income countries because of several system-level factors.ObjectiveTo examine access to reperfusion and percutaneous coronary intervention (PCI) during STEMI using a hub-and-spoke model.Design, setting, and participantsThis multicenter, prospective, observational study of a quality improvement program studied 2420 patients 20 years or older with symptoms or signs consistent with STEMI at primary care clinics, small hospitals, and PCI hospitals in the southern state of Tamil Nadu in India. Data were collected from the 4 clusters before implementation of the program (preimplementation data). We required a minimum of 12 weeks for the preimplementation data with the period extending from August 7, 2012, through January 5, 2013. The program was then implemented in a sequential manner across the 4 clusters, and data were collected in the same manner (postimplementation data) from June 12, 2013, through June 24, 2014, for a mean 32-week period.ExposuresCreation of an integrated, regional quality improvement program that linked the 35 spoke health care centers to the 4 large PCI hub hospitals and leveraged recent developments in public health insurance schemes, emergency medical services, and health information technology.Main outcomes and measuresPrimary outcomes focused on the proportion of patients undergoing reperfusion, timely reperfusion, and postfibrinolysis angiography and PCI. Secondary outcomes were in-hospital and 1-year mortality.ResultsA total of 2420 patients with STEMI (2034 men [84.0%] and 386 women [16.0%]; mean [SD] age, 54.7 [12.2] years) (898 in the preimplementation phase and 1522 in the postimplementation phase) were enrolled, with 1053 patients (43.5%) from the spoke health care centers. Missing data were common for systolic blood pressure (213 [8.8%]), heart rate (223 [9.2%]), and anterior MI location (279 [11.5%]). Overall reperfusion use and times to reperfusion were similar (795 [88.5%] vs 1372 [90.1%]; P = .21). Coronary angiography (314 [35.0%] vs 925 [60.8%]; P < .001) and PCI (265 [29.5%] vs 707 [46.5%]; P < .001) were more commonly performed during the postimplementation phase. In-hospital mortality was not different (52 [5.8%] vs 85 [5.6%]; P = .83), but 1-year mortality was lower in the postimplementation phase (134 [17.6%] vs 179 [14.2%]; P = .04), and this difference remained consistent after multivariable adjustment (adjusted odds ratio, 0.76; 95% CI, 0.58-0.98; P = .04).Conclusions and relevanceA hub-and-spoke model in South India improved STEMI care through greater use of PCI and may improve 1-year mortality. This model may serve as an example for developing STEMI systems of care in other low- to middle-income countries.

Dataset Information

A Machine Learning Model for Predicting In-Hospital Mortality in Chinese Patients With ST-Segment Elevation Myocardial Infarction: Findings From the China Myocardial Infarction Registry.

Background

Objective

Methods

Results

Conclusions

Trial registration

Publications

A Machine Learning Model for Predicting In-Hospital Mortality in Chinese Patients With ST-Segment Elevation Myocardial Infarction: Findings From the China Myocardial Infarction Registry.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets