Evaluation of Synthetic Categorical Data Generation Techniques for Predicting Cardiovascular Diseases and Post-Hoc Interpretability of the Risk Factors

Machine Learning (ML) methods have become important for enhancing the performance of decision-support predictive models. However, class imbalance is one of the main challenges for developing ML models, because it may bias the learning process and the model generalization ability. In this paper, we consider oversampling methods for generating synthetic categorical clinical data aiming to improve the predictive performance in ML models, and the identification of risk factors for cardiovascular diseases (CVDs). We performed a comparative study of several categorical synthetic data generation methods, including Synthetic Minority Oversampling Technique Nominal (SMOTEN), Tabular Variational Autoencoder (TVAE) and Conditional Tabular Generative Adversarial Networks (CTGANs). Then, we assessed the impact of combining oversampling strategies and linear and nonlinear supervised ML methods. Lastly, we conducted a post-hoc model interpretability based on the importance of the risk factors. Experimental results show the potential of GAN-based models for generating high-quality categorical synthetic data, yielding probability mass functions that are very close to those provided by real data, maintaining relevant insights, and contributing to increasing the predictive performance. The GAN-based model and a linear classifier outperform other oversampling techniques, improving the area under the curve by 2%. These results demonstrate the capability of synthetic data to help with both determining risk factors and building models for CVD prediction.

Antimalarials exert a cardioprotective effect in lupus patients: Insights from the Spanish Society of Rheumatology Lupus Register (RELESSER) analysis of factors associated with heart failure.

BACKGROUND/OBJECTIVES: Factors associated with chronic heart failure (CHF) in patients with systemic lupus erythematosus (SLE) have received little attention. Recent data on the use of hydroxychloroquine in the treatment of SARS-CoV-2 infection have cast doubt on its cardiac safety. The factors associated with CHF, including therapy with antimalarials, were analyzed in a large multicenter SLE cohort. METHODS: Cross-sectional study including all patients with SLE (ACR-1997 criteria) included in the Spanish Society of Rheumatology Lupus Register (RELESSER), based on historically gathered data. Patients with CHF prior to diagnosis of SLE were excluded. A multivariable analysis exploring factors associated with CHF was conducted. RESULTS: The study population comprised 117 patients with SLE (ACR-97 criteria) and CHF and 3,506 SLE controls. Ninety percent were women. Patients with CHF were older and presented greater SLE severity, organ damage, and mortality than those without CHF. The multivariable model revealed the factors associated with CHF to be ischemic heart disease (7.96 [4.01-15.48], p < 0.0001), cardiac arrhythmia (7.38 [4.00-13.42], p < 0.0001), pulmonary hypertension (3.71 [1.84-7.25], p < 0.0002), valvulopathy (6.33 [3.41-11.62], p < 0.0001), non-cardiovascular damage (1.29 [1.16-1.44], p < 0.000) and calcium/vitamin D treatment (5.29 [2.07-16.86], p = 0.0015). Female sex (0.46 [0.25-0.88], p = 0.0147) and antimalarials (0.28 [0.17-0.45], p < 0.000) proved to be protective factors. CONCLUSIONS: Patients with SLE and CHF experience more severe SLE. Treatment with antimalarials appears to confer a cardioprotective effect.

Data and Network Analytics for COVID-19 ICU Patients: A Case Study for a Spanish Hospital.

The COVID-19 pandemic presents unprecedented challenges to the healthcare systems around the world. In 2020, Spain was among the countries with the highest Intensive Care Unit (ICU) hospitalization and mortality rates. This work analyzes data of COVID-19 patients admitted to a Spanish ICU during the first wave of the pandemic. The patients in our study either died (deceased patients) or were discharged from the ICU (non-deceased patients) and underwent the following landmarks: beginning of symptoms; arrival at the emergency department; beginning of the hospital stay; and ICU admission. Our goal is to create a graph-based data-science methodology to find associations among patients' comorbidities, previous medication, symptoms, and the COVID-19 treatment, and to analyze their evolution across landmarks. Towards that end, we first perform a hypothesis test based on bootstrap to identify discriminative features among deceased and non-deceased patients. Then, we leverage graph-based representations and network analytics to determine pairwise associations and complex relations among clinical features. The descriptive statistical analysis confirms that deceased patients exhibit multiple comorbidities with stronger levels of association and are treated with a wider range of drugs during the ICU stay. We also observe that the most common treatment was the simultaneous administration of lopinavir/ritonavir with hydroxychloroquine, regardless of the patients' outcome. Our results illustrate how graph tools and representations yield insights on the relations among comorbidities, drug treatments, and patients' evolution. All in all, the approach puts forth a new data-analysis tool for clinicians that can be applied to analyze (post-COVID) symptom/patient evolution.

An integrated emergency department/hospital at home model in mild COVID-19 pneumonia: feasibility and outcomes after discharge from the emergency department.

To evaluate the effectiveness of an integrated emergency department (ED)/hospital at home (HH) medical care model in mild COVID-19 pneumonia and evaluate baseline predictors of major outcomes and potential savings. Retrospective cohort study with patients evaluated for COVID-19 pneumonia in the ED, from March 3 to April 30, 2020. All of them were discharged home and controlled by HH. The main outcomes were ED revisit and the need for deferred hospital admission (protocol failure). Outcome predictors were analyzed by simple logistic regression model (OR; 95% CI). Potential savings of this medical care model were estimated. Of the 377 patients attended in the ED, 109 were identified as having mild pneumonia and were included in the ED/HH medical care model. Median age was 50.0 years, 52.3% were males and 57.8% had Charlson index ≥ 1. The median HH stay was 8 (IQR 3.7-11) days. COVID-19-related ED revisit was 19.2% (n = 21) within 6 days (IQR 3-12.5) after discharge from ED. Overall protocol failure (deferred hospital admission) was 6.4% (n = 7), without ICU admission. The ED/HH model provided potential cost savings of 77% compared to traditional stay, due to the costs of home care entails 23% of the expenses generated by a conventional hospital stay. 789 days of hospital stay were avoided by HH, rather than hospital admission. An innovative ED/HH model for selected patients with mild COVID-19 pneumonia is feasible, safe and effective. Less than 6.5% of patients requiring deferred hospital admission and potential savings were generated due to hospitalization.