Qasem Al-Tashi

Research

Research interests & current projects

AI-Driven Biomarker Discovery

Precision medicine through machine learning models for biomarker identification.

Swarm Intelligence & Metaheuristics

Optimization for feature selection and model tuning (GWO, PSO, and hybrids).

Causal Inference & Subgroup Identification

Treatment effect estimation for precision oncology.

Computer Vision

Deep learning for industrial and medical imaging applications.

LLMs & RAG

Retrieval-augmented generation for intelligent decision support systems.

Survival Analysis

Prognostic modeling in oncology using deep survival networks.

Current research projects

AI-Powered CT Verification System

Computer Vision

Dual-camera computer vision system for automated Current Transformer connection verification in manufacturing QA, using OCR, Hough-circle detection, and SVM/KNN classifiers — achieving 100% leave-one-out accuracy.

Enterprise RAG & LLM Decision Support

LLMs / GenAI

Retrieval-augmented generation architectures for natural language interfaces to structured databases, document retrieval, and conversational AI for industrial maintenance diagnostics.

Causal AI for Cancer Immunotherapy

Computational Medicine

Causal AI-based clinical and radiomic analysis for optimizing patient selection in combined immunotherapy and SABR in early-stage NSCLC (Journal for Immunotherapy of Cancer, 2025).

Swarm Intelligence for Deep Survival Networks

Swarm Intelligence

Swarm-based hyperparameter optimization and feature selection for deep survival networks applied to prognostic radiomics across multiple solid cancers.

Industrial IoT Analytics

Applied ML

ML pipelines for industrial equipment monitoring using Gaussian Mixture Model-based automatic threshold computation for regime detection and utilization analytics.

Open-source software

SwarmDeepSurv

Python package for swarm intelligence-enhanced deep survival networks for prognostic radiomics signatures in cancer research.

Hybrid-Binary-GWO-FS

MATLAB package for binary optimization using hybrid Grey Wolf Optimization for feature selection.

BMOGWO-S

MATLAB package for binary multi-objective Grey Wolf Optimizer for feature selection in classification.