FAZLE KARIM

Smart-agriculture research using UAVs and an AlexNet CNN to detect crop diseases in Bangladesh, with blockchain integrated to improve supply-chain management and crop-protection data.

Compared six ML algorithms and a deep-learning sequential neural network on NASA's PROMISE dataset for software bug prediction. SVM outperformed other models, and feature selection meaningfully improved accuracy.

Explores interpretable machine learning models for predicting heart failure readmissions while maintaining high predictive accuracy. Evaluates multiple algorithms and explainability techniques to support reliable, data-driven clinical decision making.

Explores consumer sentiment and behavior trends through social media data across key US platforms (X-Twitter, Facebook, Instagram, TikTok). Leverages official Graph and microblogging APIs alongside machine learning models (Logistic Regression, Random Forest, XGBoost) to extract real-time business insights and consumer perception metrics.

Focuses on applying artificial intelligence toward optimal energy consumption and sustainable resource management across Southern California. Analyzed multi-year hourly electricity records across residential, commercial, and industrial facilities with environmental indicators.

Investigated large-scale social media interactions and microblogging datasets to quantify consumer sentiment trajectories in the US marketplace, applying strategic metrics (accuracy, precision, recall, F1-score) to evaluate classifier performance.

Designed a fine-grained aspect-based sentiment analysis (ABSA) framework for Amazon customer reviews of consumer electronics. Combined TF-IDF and word n-grams with Support Vector Machines (SVM), achieving 91% accuracy while addressing real-world class imbalance.

Applied machine learning analytics to evaluate US low-carbon technology import and export trade flows, assessing macroeconomic contributions, trade pattern shifts, and the efficacy of sustainable economic policies.

Investigated how AI-powered technologies, predictive analytics, and Dynamic Capabilities Theory enhance supply chain flexibility and resilience, analyzing real-world enterprise operational strategies.

Deployed machine learning and deep learning models to detect incipient indicators of Alzheimer's Disease from MRI neuroimaging scans in the Open Access Series of Imaging Studies (OASIS) dataset, enabling early therapeutic intervention.

Formulated a real-time, multi-modal AI detection framework fusing Medicare claims, electronic health records (EHR), and provider behavioral signals. Utilizes autoencoders, deep classifiers, and Graph Neural Networks (GNNs) with an estimated ROC-AUC of 0.96.

Investigated how enterprise AI integration, robotic process automation (RPA), and predictive modeling drive performance benchmarks, cost reduction, and operational resilience across digital procurement and logistics networks.

Comprehensive training in machine learning applications for healthcare data, clinical diagnosis models, healthcare epiphenomena analysis, and prognostic AI frameworks.

Professional certification covering business requirement analysis, enterprise data visualization, interactive Power BI executive dashboards, and statistical decision-making.