Machine Learning Improves the Identification of Individuals With Higher Morbidity and Avoidable Health Costs After
Luiz Sérgio Fernandes de Carvalho1, Silvio Gioppato2, Marta Duran Fernandez3
1Clarity Healthcare Intelligence, Jundiaí, SP, Brazil; Cardiology Department, State University of Campinas (Unicamp), Campinas, SP, Brazil; Laboratory of Data for Quality of Care and Outcomes Research, Institute of Strategic Management in Healthcare Brasília, DF, Brazil; Escola Superior de Ciências da Saúde, Brasília, DF, Brazil.
Insights
Machine learning models accurately predict long-term cardiovascular risks and associated healthcare costs in acute coronary syndrome (ACS) patients. This approach identifies high-cost individuals, enabling better resource allocation and potentially reducing avoidable expenses.
Area of Science:
- Cardiovascular Medicine
- Health Economics
- Machine Learning in Healthcare
Background:
- Traditional risk scores for acute coronary syndrome (ACS) are limited in predicting long-term individual risks and healthcare expenditures.
- Existing methods for directly predicting ACS costs from clinical data have shown restricted success.
- Novel approaches are essential for accurately forecasting cardiovascular risk and health spending in ACS patients.
Purpose of the Study:
- To develop and validate machine learning (ML) models for predicting major/minor adverse cardiovascular events (MACE) and associated healthcare costs in ACS individuals.
- To compare the performance of ML models against traditional risk scores and logistic regression in predicting MACE.
- To identify factors contributing to high healthcare costs in ACS patients and assess the potential for cost reduction.
Main Methods:
- A two-step approach was employed: (1) predicting MACE using ML (gradient-boosting machine) and logistic regression (LR), compared with existing scores; (2) deriving costs linked to non-cardiovascular deaths, dialysis, ambulatory-care-sensitive hospitalizations (ACSH), strokes, and MACE.
- A 29-variable model incorporating socioeconomic, clinical/laboratory, and coronarography data was trained on 80% and tested on 20% of 1089 consecutive ACS patients using 4-fold cross-validation.
- Individual costs were estimated from a healthcare system perspective, utilizing cause-specific hospitalization data.
Main Results:
- The gradient-boosting machine model achieved a superior area under the curve (AUROC) of 0.891, significantly outperforming the Syntax Score II (AUROC=0.635) for MACE prediction.
- High-risk individuals (>90th percentile) exhibited elevated HbA1c and LDL-C levels and incurred 4.96-fold higher per capita costs, largely due to avoidable hospitalizations (ACSH).
- The two-step ML approach proved more effective in identifying individuals with high healthcare costs than direct cost prediction models.
Conclusions:
- Machine learning models can effectively predict long-term cardiovascular risks and avoidable healthcare costs following acute coronary syndrome.
- This predictive capability aids in identifying high-cost patients, potentially leading to more targeted interventions and cost savings.
- The study highlights the value of advanced analytics in managing cardiovascular disease and optimizing healthcare resource allocation.
Objectives:
Traditional risk scores improved the definition of the initial therapeutic strategy in acute coronary syndrome (ACS), but they were not designed for predicting long-term individual risks and costs. In parallel, attempts to directly predict costs from clinical variables in ACS had limited success. Thus, novel approaches to predict cardiovascular risk and health expenditure are urgently needed. Our objectives were to predict the risk of major/minor adverse cardiovascular events (MACE) and estimate assistance-related costs.
Methods:
We used a 2-step approach that: (1) predicted outcomes with a common pathophysiological substrate (MACE) by using machine learning (ML) or logistic regression (LR) and compared with existing risk scores; (2) derived costs associated with noncardiovascular deaths, dialysis, ambulatory-care-sensitive-hospitalizations (ACSH), strokes, and MACE. With consecutive ACS individuals (n = 1089) from 2 cohorts, we trained in 80% of the population and tested in 20% using a 4-fold cross-validation framework. The 29-variable model included socioeconomic, clinical/lab, and coronarography variables. Individual costs were estimated based on cause-specific hospitalization from the Brazilian Health Ministry perspective.
Results:
After up to 12 years follow-up (mean = 3.3 ± 3.1; MACE = 169), the gradient-boosting machine model was superior to LR and reached an area under the curve (AUROC) of 0.891 [95% CI 0.846-0.921] (test set), outperforming the Syntax Score II (AUROC = 0.635 [95% CI 0.569-0.699]). Individuals classified as high risk (>90th percentile) presented increased HbA1c and LDL-C both at <24 hours post-ACS and 1-year follow-up. High-risk individuals required 33.5% of total costs and showed 4.96-fold (95% CI 3.71-5.48, P < .00001) greater per capita costs compared with low-risk individuals, mostly owing to avoidable costs (ACSH). This 2-step approach was more successful for finding individuals incurring high costs than predicting costs directly from clinical variables.
Conclusion:
ML methods predicted long-term risks and avoidable costs after ACS.
Related Concept Videos
01:30Coronary Artery Disease I: Introduction
01:30Acute Coronary Syndrome III: Diagnostic Studies
01:28Acute Coronary Syndrome IV: Interprofessional Care
01:26Coronary Artery Disease IV: Preventive Measures
01:29Cardiomyopathy III: Hypertrophic Cardiomyopathy
01:27Coronary Artery Disease V: Interprofessional Care