Ensemble Of Bayesian Networks Research Articles

Emerging trends in antimicrobial resistance in bloodstream infections: multicentric longitudinal study in India (2017–2022)

BackgroundAntimicrobial resistance (AMR) has escalated to pandemic levels, posing a significant global health threat. This study examines the patterns and trends of AMR in Bloodstream Infections (BSIs) across India, aiming to inform better surveillance and intervention strategies. MethodsSix-year data from 21 tertiary care centers in the Indian Council of Medical Research's AMR Surveillance Network (IAMRSN) were retrospectively analyzed to estimate cluster-robust trends in resistance. Time-series analysis was used to discern lead/lag relationships between antibiotic pairs and the directional influence of resistance in community and hospital-acquired BSIs(CA/HA BSIs). A data-driven Bayesian network ensemble averaged over 301 bootstrap samples was modelled to uncover systemic associations between AMR and Sustainable Development Goals (SDGs). FindingsOur findings indicate significant (p<0.001) monthly increases in Imipenem and Meropenem resistance for Klebsiella, E. coli, and Acinetobacter BSIs. Importantly, Carbapenem resistance in HA-BSIs preceded that in CA-BSIs for Klebsiella and Acinetobacter (p<0.05). At a national level, Cefotaxime resistance emerged as a potential early indicator for emerging Carbapenem resistance, proposing a novel surveillance marker. In Klebsiella BSIs, states with higher achievement of SDG3 goals showed lower Imipenem resistance. A model-based AMR scorecard is introduced for focused interventions and continuous monitoring. InterpretationThe identified spatiotemporal trends and drug resistance associations offer critical insights for AMR surveillance aligning with WHO GLASS standards.The escalation of carbapenem resistance in BSIs demands vigilant monitoring and may be crucial for achieving SDGs by 2030. Implementing the proposed framework for data-driven evidence can help nations achieve proactive AMR surveillance. FundingNo specific funding was received for this analysis.

COVID-19 Risk Stratification and Mortality Prediction in Hospitalized Indian Patients: Harnessing clinical data for public health benefits.

The variability of clinical course and prognosis of COVID-19 highlights the necessity of patient sub-group risk stratification based on clinical data. In this study, clinical data from a cohort of Indian COVID-19 hospitalized patients is used to develop risk stratification and mortality prediction models. We analyzed a set of 70 clinical parameters including physiological and hematological for developing machine learning models to identify biomarkers. We also compared the Indian and Wuhan cohort, and analyzed the role of steroids. A bootstrap averaged ensemble of Bayesian networks was also learned to construct an explainable model for discovering actionable influences on mortality and days to outcome. We discovered blood parameters, diabetes, co-morbidity and SpO2 levels as important risk stratification features, whereas mortality prediction is dependent only on blood parameters. XGboost and logistic regression model yielded the best performance on risk stratification and mortality prediction, respectively (AUC score 0.83, AUC score 0.92). Blood coagulation parameters (ferritin, D-Dimer and INR), immune and inflammation parameters IL6, LDH and Neutrophil (%) are common features for both risk and mortality prediction. Compared with Wuhan patients, Indian patients with extreme blood parameters indicated higher survival rate. Analyses of medications suggest that a higher proportion of survivors and mild patients who were administered steroids had extreme neutrophil and lymphocyte percentages. The ensemble averaged Bayesian network structure revealed serum ferritin to be the most important predictor for mortality and Vitamin D to influence severity independent of days to outcome. The findings are important for effective triage during strains on healthcare infrastructure.

Abstract 150: Machine Learning Methodology Predicts Comorbidities are Associated With Increased Total Healthcare Costs Among Patients With Severe Peripheral Artery Disease

Introduction: Peripheral artery disease (PAD) is manifested over a continuum of severity with comorbidities that may significantly increase healthcare costs (HC). Little research has been completed to understand the healthcare resource use (HRU) and HC in this population, specifically among severe PAD patients. We sought to understand the economic burden in this population. Methods: We identified severe PAD patients (rest pain, gangrene or ulceration) from an integrated administrative claims and electronic medical records database (Optum + Humedica 2007-15) with over 7 million patients. The first PAD diagnosis was the index date. Patients were required to be age ≥50 at index date, have clinical activity and continuous enrollment in the 6-month pre-index and 12-month or until death post-index periods. Patients with history of intracranial hemorrhage, stroke and transient ischemic attack were removed. We assessed HRU and all-cause annual total HC in the post-index period or until death, and descriptive analyses, means and SDs. Reverse Engineering and Forward Simulation (REFS TM ) models, an ensemble of Bayesian networks, were machine learned to examine baseline demographic and clinical characteristics and their association with post-index natural log all-cause annual total HC among living patients. We assessed effect estimates across the ensemble using Mean Percentage Change in Costs (MPCC) with SD. Results: The final study sample included 3,189 severe PAD patients. Mean number of all-cause hospitalizations was 1.3 (SD 1.8) and the mean length of stay was 8.0 days (SD 18.2). The mean all-cause annual total HC per patient was $56,973 (SD $91,523). Highly predictive factors associated with increased costs were (MPCC, SD): chronic ulcer of leg or foot (1.9, 0.1), chronic kidney diseases (CKD; 1.9, 0.2), cellulitis and abscess (1.8, 0.2), hypertension (1.6, 0.1), and carditis and cardiomyopathy (1.2, 0.1). Conclusion: In this study, the presence of chronic ulcers in the lower extremities and CKD were two factors most predictive of increased all-cause total HC in a geographically diverse population of severe PAD patients.

Automatic Emergence Detection in Complex Systems

Complex systems consist of multiple interacting subsystems, whose nonlinear interactions can result in unanticipated (emergent) system events. Extant systems analysis approaches fail to detect such emergent properties, since they analyze each subsystem separately and arrive at decisions typically through linear aggregations of individual analysis results. In this paper, we propose a quantitative definition of emergence for complex systems. We also propose a framework to detect emergent properties given observations of its subsystems. This framework, based on a probabilistic graphical model called Bayesian Knowledge Bases (BKBs), learns individual subsystem dynamics from data, probabilistically and structurally fuses said dynamics into a single complex system dynamics, and detects emergent properties. Fusion is the central element of our approach to account for situations when a common variable may have different probabilistic distributions in different subsystems. We evaluate our detection performance against a baseline approach (Bayesian Network ensemble) on synthetic testbeds from UCI datasets. To do so, we also introduce a method to simulate and a metric to measure discrepancies that occur with shared/common variables. Experiments demonstrate that our framework outperforms the baseline. In addition, we demonstrate that this framework has uniform polynomial time complexity across all three learning, fusion, and reasoning procedures.

Feature deduction and ensemble design of intrusion detection systems

Current intrusion detection systems (IDS) examine all data features to detect intrusion or misuse patterns. Some of the features may be redundant or contribute little (if anything) to the detection process. The purpose of this study is to identify important input features in building an IDS that is computationally efficient and effective. We investigated the performance of two feature selection algorithms involving Bayesian networks (BN) and Classification and Regression Trees (CART) and an ensemble of BN and CART. Empirical results indicate that significant input feature selection is important to design an IDS that is lightweight, efficient and effective for real world detection systems. Finally, we propose an hybrid architecture for combining different feature selection algorithms for real world intrusion detection.