Lebede Ngartera

Dr. Lebede Ngartera

orcid

Research Focus & Metrics

Academic Career Length

12Years (2015-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
Verified
Journal Article
Open access
PDF file
Lebede NgarteraLebede Ngartera
Lebede Ngartera
MIMAHAMAT ALI ISSAKA
MAHAMAT ALI ISSAKA
SNSaralees Nadarajah
Saralees Nadarajah
This study aims to explore the efficacy of Bayesian Neural Networks (BNNs) in enhancing predictive modeling for healthcare applications. Advancements in artificial intelligence have significantly improved predictive modeling capabilities, with BNNs offering a probabilistic framework that addresses the inherent uncertainty and variability in healthcare data. This study demonstrates the real-world applicability of BNNs through three key case studies: personalized diabetes treatment, early Alzheimer’s disease detection, and predictive modeling for HbA1c levels. By leveraging the Bayesian approach, these models provide not only enhanced predictive accuracy but also uncertainty quantification, a critical factor in clinical decision making. While the findings are promising, future research should focus on optimizing scalability and integration for real-world applications. This work lays a foundation for future studies, including the development of rating scales based on BNN predictions to improve clinical outcomes.

12 May 2025 in IEEE Access
Verified
Journal Article
Open access
Lebede NgarteraLebede Ngartera
Lebede Ngartera
MIMAHAMAT ALI ISSAKA
MAHAMAT ALI ISSAKA
SNSaralees Nadarajah
Saralees Nadarajah
Online scams continue to escalate in scale and sophistication, ranging from deceptive phishing emails to complex financial fraud schemes. These evolving threats have surpassed the capabilities of traditional detection systems, creating an urgent demand for scalable, real-time, and interpretable machine learning solutions. This study revisits the Naïve Bayes algorithm—often underestimated in modern cybersecurity—as a core building block for effective scam detection. By looking at two specific case studies on phishing emails and financial fraud detection, we look into the math behind Naïve Bayes and talk about its flaws, such as the assumption that features are independent, problems with high-dimensional data, and extreme class imbalance. We apply advanced feature engineering techniques and introduce hybrid model architectures that integrate Naïve Bayes with deep learning and ensemble methods. Our empirical results reveal that a strategically optimized Naïve Bayes model can deliver competitive accuracy, while maintaining transparency and computational efficiency—key attributes for real-world fraud prevention systems. By bridging theory and application, this research challenges the notion that simplicity implies ineffectiveness. It demonstrates that Naïve Bayes, particularly when hybridized, remains a powerful and adaptable tool in the fight against evolving cyber threats, contributing to the design of robust and responsive cybersecurity frameworks.

31 August 2024 in Journal of Mathematics Research
Verified
Journal Article
Open access
PDF file
Lebede NgarteraLebede Ngartera
Lebede Ngartera
NNNdogotar Nelio
Ndogotar Nelio
NNNgarasta Ngarkodje
Ngarasta Ngarkodje
This study explores the utilization of graph theory and optimization methods to improve the effectiveness of emergency response logistics. The study devises and evaluates multiple algorithms with the objective of enhancing the efficiency of emergency routes and resource distributions, showcasing advancements compared to conventional approaches. The findings demonstrate substantial decreases in response times, operational expenses, and improvements in resource utilization rates. This study provides reliable and powerful resources for emergency managers to enhance their ability to strategize and carry out response operations with more efficiency. Furthermore, the models introduced in this work are versatile and may be utilized in different emergency situations to guarantee the most efficient allocation of resources and response times. The results emphasize the practical significance of graph theory in emergency logistics, establishing a solid basis for additional advancement and practical implementation.

Verified
Journal Article
Open access
PDF file
Lebede NgarteraLebede Ngartera
Lebede Ngartera
SNSaralees Nadarajah
Saralees Nadarajah
RKRodoumta Koina
Rodoumta Koina
YGYoussou Gningue
Youssou Gningue
In high-stakes settings, the most consequential failure of a language model is not a wrong answer but an answer it was not entitled to give. Existing retrieval-augmented generation (RAG) pipelines retrieve context, generate text, and perhaps add citations, but they do not decide whether the evidence justifies answering, how uncertain the answer is, or at what level the system should intervene. We argue that LLMs should not only generate answers; they should be embedded inside a selective decision architecture that jointly estimates answerability, quantifies uncertainty, verifies structural validity, and chooses among direct response, escalation, abstention, or failure. We introduce BRAG (Bayesian Retrieval-Augmented Generation), a framework that operationalises this shift from answer generation to evidence-governed decision support. BRAG estimates an answerability posterior, decomposes uncertainty into epistemic and aleatoric components, and applies a structural validity gate prior to answer emission. Evaluation is conducted using controlled Monte Carlo simulation (n = 2400 queries) and a calibrated statistical pilot (N = 500), both parametric models of the pipeline’s output distribution, together with a governed operational validation that executes the full released pipeline end-to-end on independently generated MIMIC-IV-schema records (N = 100; not credentialed patient records), expert adjudication on a stratified subset (N = 200), and secondary transfer experiments on SEC EDGAR and CUAD. In simulation, BRAG reduces hallucination from 0.257 to 0.016 (93.8%) and achieves the highest coverage-adjusted utility (0.632) among five systems. In the synthetic MIMIC-IV-schema pilot, hallucination decreases from 0.292 to 0.020 (93.2%), with utility 0.538 at 89.6% coverage and an answerability AUROC of 0.692, which is moderate in absolute terms and is therefore positioned as a routing signal that operates jointly with the deterministic validity gate rather than as a stand-alone clinical classifier. Expert adjudication yields substantial agreement (Cohen’s κ = 0.778) and 93.5% concordance with BRAG decisions. Cross-domain transfer demonstrates 96–97% hallucination reduction without retriever modification, while ablation identifies the structural validity gate as the primary safety mechanism and the answerability posterior as the primary coverage and routing-precision mechanism. These results show that combining answerability estimation with structural validity enforcement can substantially reduce unsupported outputs. All findings are methodological rather than clinical: every evaluation tier uses synthetic or schema-conformant data, and validation on credentialed de-identified patient records remains necessary before any clinical deployment.

Verified
Journal Article
Open access
PDF file
Lebede NgarteraLebede Ngartera
Lebede Ngartera
SNSaralees Nadarajah
Saralees Nadarajah
RKRodoumta Koina
Rodoumta Koina
YGYoussou Gningue
Youssou Gningue
Each year, inventory decisions made under demand uncertainty generate substantial economic losses, reflecting a persistent disconnect between forecasting models and the operational decisions they are intended to support. This paper addresses this gap by proposing a Decision Intelligence Framework that unifies three components typically treated in isolation: probabilistic demand forecasting via gradient boosting quantile regression, constrained newsvendor optimization under capacity and budget constraints, and coherent tail risk evaluation using Conditional Value-at-Risk (CVaR95). We establish a central theoretical result showing that calibrated quantile forecasts are mathematically equivalent to optimal newsvendor solutions, providing a rigorous decision-theoretic foundation linking probabilistic forecasting and inventory control. The framework is evaluated on the UCI Online Retail dataset (2010–2011), aggregated to daily demand at the country–SKU level and densified to a daily panel by treating missing transaction days as zero demand. Relative to median-based (P50) policies, P90 policies reduce tail risk (CVaR95) by 26.7% under empirical residual bootstrap, increase cycle service levels from 44.4% to 89.5%, and reduce mean cost by 48.7% (non-overlapping bootstrap CIs for CVaR95). A lognormal stress test shows larger reductions (72.3%), and a CV sweep confirms monotone gains in this setting.

28 October 2015 in International Journal of Applied Mathematics
Verified
Journal Article
Open access
PDF file
Lebede NgarteraLebede Ngartera
Lebede Ngartera
SDS. Diagne
S. Diagne
YGY. Gningue
Y. Gningue
Contents IJAM: Volume 28, No. 5 (2015)