Raisul Islam Khan

Raisul Islam Khan

Auto-generated profile, based on publicly available information

Raisul Islam Khan

California State Polytechnic University, Pomona, USA

Research Focus & Metrics

Academic Career Length

2Years (2025-Present)

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

Representative Publications

View All Publications
9 August 2025 in Global Research Journal of Natural Science and Technology
Verified
Journal Article
Open access
PDF file
URUzair Rahman
Uzair Rahman
HAHossain Ahmed
Hossain Ahmed
RKRaisul Islam Khan
Raisul Islam Khan
MSMuhammad Saqlain
Muhammad Saqlain
SSShahbaz Ahmed Siddiqui
Shahbaz Ahmed Siddiqui
SAShamim Akhtar
Shamim Akhtar
In the face of rapidly evolving cyber threats, traditional security systems have proven inadequate for protecting critical infrastructure environments such as Industrial Control Systems (ICS). This study investigates the integration of machine learning (ML) techniques and cybersecurity knowledge graphs (CKGs) to enhance threat detection and situational awareness in complex cyber-physical ecosystems. By evaluating advanced context-aware detection models, including LSTM-Autoencoders and hybrid SVM-based systems, the research demonstrates significant improvements in anomaly detection accuracy and interpretability. Furthermore, the study explores how knowledge graphs—constructed from heterogeneous data sources—facilitate semantic reasoning and automate threat intelligence processing. Empirical evidence highlights the value of combining behavioral attributes with semantic context to identify and mitigate sophisticated attacks that evade conventional intrusion detection systems (IDS). Despite these advancements, the research also identifies operational challenges, including system integration, computational overhead, and limited scalability of AI models in real-world deployments. Recommendations are made for future implementation strategies that prioritize interoperability, explainability, and collaboration among interdisciplinary teams. Finally, potential directions for further research are proposed, such as the adoption of blockchain for secure data provenance and the expansion of CKG frameworks into domains like finance and smart healthcare. This work contributes to the growing body of knowledge advocating for intelligent, adaptive, and semantically enriched cybersecurity systems capable of responding proactively to dynamic threat landscapes.

23 April 2026 in Discover Analytics
Verified
Journal Article
Open access
PDF file
RKRaisul Islam Khan
Raisul Islam Khan
This study investigates whether AI-powered predictive analytics can support exploratory delivery disruption detection, operationalizing supply chain resilience as an observable delivery performance outcome rather than a latent organizational capability. Random Forest and XGBoost classification models were developed and evaluated using a publicly available secondary logistics dataset hosted on Kaggle (N = 12,144), whose original organizational provenance and collection procedures could not be independently verified. Accordingly, findings are interpreted as exploratory and dataset-specific. Random Forest achieved 63.5% accuracy and XGBoost 62.0% on the held-out test set, compared with a 57.9% majority-class baseline, indicating modest predictive improvement over trivial classification. Both models performed substantially better on delayed deliveries than on on-time deliveries, suggesting limited three-class discriminative capability. Because delivery duration was derived from the same temporal fields underlying the outcome label, its predictive role is interpreted cautiously due to potential target-adjacent information. Regional comparisons are therefore presented descriptively through separate delivery indicators rather than a weighted composite resilience score. The study contributes a cautious predictive benchmark for exploratory delay-risk screening and identifies methodological priorities for future work using validated logistics datasets and leakage-resistant feature designs.

31 October 2025 in Research Square
Verified
Preprint
Open access
PDF file
RKRaisul Islam Khan
Raisul Islam Khan
The main objective of this study was to investigate how integrating AI-powered predictive analytics into supply chain management could help improve resilience and delivery performance. We developed and evaluated Random Forest and XGBoost models using a comprehensive logistics dataset (N = 15,549) to predict delivery disruptions. Shipping mode was the most important in explaining the delivery outcomes (MI = 0.179), followed by delivery duration (MI = 0.015). Analysis by region showed consistent delay frequencies (57-59%) by markets, but delivery durations differed significantly between developed and emerging areas. On the Random Forest model, we obtained 61.2% (±1.3%) for delivery outcomes prediction, slightly better than XGBoost (59.8% ±0.9%). We present empirical evidence to support the efficacy of AI in supply chain optimization and provide actionable frameworks for using predictive analytics in supply chain operations. This work helps to strengthen the theory and practice of AI-guided solutions in modern supply chain management.

7 September 2026 in Discover Artificial Intelligence
Verified
Journal Article
Open access
PDF file
MKMd Raisul Islam Khan
Md Raisul Islam Khan
AMAbdullah Al Mamun
Abdullah Al Mamun
MTMuhtasib Sarker Tahsin
Muhtasib Sarker Tahsin
Background and motivation: Drug shortages disrupt patient care and hospital operations, while public records lack inventory and procurement data. Problem statement: This study separately evaluates whether U.S. Food and Drug Administration (FDA) shortage histories can rank 90-day recurrence risk and how a status-derived risk proxy changes inventory-policy trade-offs in a synthetic simulator; it does not imply autonomous procurement. Methodology: The analysis used 1688 FDA records. Snapshot classification served only as a diagnostic. The main empirical task used a 90-day recurrence panel with historical-only features, temporal splits, leakage controls, group holdouts, and machine-learning models. Aggregate counts were compared with naive, rolling-mean, exponential-smoothing, and autoregressive integrated moving average (ARIMA) baselines. Separately, tabular Q-learning was evaluated in a disclosed synthetic inventory environment against economic order quantity (EOQ), min-max, safety-stock, threshold, and human-rule policies using stress tests, sensitivity analysis, effect sizes, and governance proxies. Findings: With 15 positive test windows, gradient boosting achieved receiver-operating-characteristic area under the curve (ROC-AUC) 0.831, precision-recall area under the curve (PR-AUC) 0.0465, and recall 0.667; logistic regression captured 40% of positives in the highest-risk 10% of observations. In simulation, Q-learning yielded a stockout rate of 0.160 versus 0.163 for min-max, with higher cost and more emergency orders. The threshold policy produced the lowest stockout rate (0.048) and greatest intervention burden. Implications: Public FDA histories are more defensible for rare-event risk ranking than high-precision prediction. Policy and governance results remain simulation-derived until validated with hospital data and qualified users.

28 July 2025 in Research Square
Verified
Preprint
Open access
PDF file
RKRaisul Islam Khan
Raisul Islam Khan
Efficient transportation is a critical factor in modern supply chains, influencing both cost management and timely product delivery. This study presents a novel approach for optimizing transportation costs through a Mixed Integer Linear Programming (MIP) model, designed to minimize the overall expense of moving goods from multiple supply nodes (warehouses) to various demand nodes (project sites). The model integrates essential supply chain factors such as freight rates, warehouse costs, and transportation capacity constraints, enabling a holistic optimization strategy. Data preprocessing techniques, including the removal of duplicates, handling of missing values, and standardization of column names, ensure the integrity and consistency of the input data. Exploratory data analysis (EDA) is employed to understand key relationships between shipping costs, service levels, and capacity distributions, which further informs the optimization model. By using MIP, the model selects the most cost-effective transportation routes while adhering to both supply and demand constraints. This study contributes to the development of a robust, data-driven decision-making tool for improving transportation logistics, offering substantial cost savings and enhanced operational efficiency for organizations. The proposed method provides an actionable framework for supply chain managers seeking to optimize logistics networks in an increasingly competitive global marketplace.

19 July 2026 in Discover Artificial Intelligence
Verified
Journal Article
Open access
PDF file
RKRaisul Islam Khan
Raisul Islam Khan
TATaycir Ben Abid
Taycir Ben Abid
MJMohammed Abu Jahed
Mohammed Abu Jahed
Agricultural supply-chain managers need early warning of unusual price movements, yet deterministic monthly forecasts remain difficult when commodity, input, fuel, and freight regimes shift. This study evaluates an explainable machine-learning workflow using 18 monthly U.S. Bureau of Labor Statistics Producer Price Index series. The supervised sample contains 189 forecast origins, 164 engineered predictors, 141 training months, and a 48-month chronological holdout. Model families and hyperparameters are selected exclusively by five-fold time-series cross-validation; the holdout is used once for evaluation. Elastic net regression is preselected for signed one-month-ahead Farm Products PPI change and achieves holdout MAE of 2.00%, RMSE of 2.72%, R2 of 0.143, and directional accuracy of 58.33%. Its squared-loss improvement over the zero-change baseline is suggestive but not significant at 5% (Harvey-Leybourne-Newbold-adjusted Diebold-Mariano p = 0.062). Extra trees regression is preselected for absolute-change forecasting and yields high-volatility screening ROC-AUC of 0.652, recall of 85.71%, precision of 25.00%, and F1 of 0.387. The exact 95% recall interval is 42.13%-99.64%, and the moving-block bootstrap ROC-AUC interval is 0.375–0.881, demonstrating substantial uncertainty. The evidence indicates limited signed-forecast skill and a high-recall but low-precision screening pattern in this holdout sample. The workflow should therefore be interpreted as a preliminary human-review risk screen, not as autonomous price prediction or a deployment-grade probability model.