Date of Submission

Fall 2025

Supervisor

Dr. Tariq Mahmood, Professor, School of Mathematics and Computer Science (SMCS)

Co-Supervisor

Dr. Asma Sanam, Lecturer, School of Mathematics and Computer Science (SMCS)

Committee Member 1

Dr. Tariq Mahmood, Supervisor

Committee Member 2

Dr. Imran Rauf - SMCS, Institute of Business Administration (IBA), Karachi

Committee Member 3

Dr. Syed Hammad Ahmed - DHA Suffa University

Degree

Master of Science in Data Science

Department

Department of Computer Science

Faculty/ School

School of Mathematics and Computer Science (SMCS)

Keywords

Grey Wolf Optimization, Decision Trees, Swarm Intelligence, Fuzzy Logic, F1  Score

Abstract

Accurate and interpretable clinical prediction models are essential for decision-making in healthcare, where transparency and reliability directly influence patient outcomes. Decision Trees (DTs) remain one of the most interpretable machine learning models, yet their performance is often limited by greedy, locally optimal splitting strategies. To overcome these limitations, swarm intelligence algorithms have been increasingly applied for DT optimization; however, standard metaheuristics such as Genetic Algorithm (GA), Ant Colony Optimization (ACO), Particle Swarm Optimization (PSO), and the traditional Grey Wolf Optimizer (GWO) still suffer from convergence stagnation, premature exploitation, and reduced diversity, particularly on noisy, nonlinear, and imbalanced clinical datasets. In response to these challenges, this study proposes an Enhanced Grey Wolf Optimizer (EGWO) specifically designed to improve the optimization of DT structures across multiple healthcare domains, including sepsis, diabetes, cancer, heart disease, and burn injury. The proposed EGWO introduces two major methodological innovations. First, a fuzzy logic–based adaptive controller dynamically adjusts the parameter a to maintain an appropriate balance between exploration and exploitation throughout the search process. Instead of relying on GWO’s original linearly decreasing schedule, the fuzzy system incorporates hybrid membership functions, triangular and trapezoidal, to compute smooth, context-aware adjustments based on population diversity and convergence status. Second, the algorithm implements δ-wolf layer removal, simplifying the leadership hierarchy and mitigating early dominance of suboptimal leaders. This modification enhances search diversity, prevents premature convergence, and allows more wolves to contribute meaningfully to the search dynamics.

Document Type

Restricted Access

Submission Type

Thesis

Download

Share

COinS