Adaptive AI-Powered Intelligent Bug Prediction and Self-Healing System for Continuous Software Testing
Year: 2025
Type: article
Abstract: Software systems need intelligent prediction systems that detect defects early through autonomous failure repairs and continuous delivery pipeline operation without human intervention to achieve both resilience and self-sustaining capacity.This paper presents REJUV-CI, a new AI-based system developed to restore and stabilize software testing processes through predictive analytics and automated self-healing capabilities.REJUV-CI uses hybrid machine learning, combining XGBoost deep neural architectures with TensorFlow-based repair models, to predict which code parts are likely to develop defects while initiating automatic correction processes.The system uses open-source bug repositories such as PROMISE Defects4J and Bugzilla to collect commit-level, semantic, and structural elements, enabling accurate bug prediction.The selfhealing engine detects problems autonomously and uses patch recommendations to assess new code segments before reintegrating fixed parts into continuous integration pipelines.The system implements continuous monitoring along with feedback mechanisms that adapt to software changes throughout the lifecycle, maintaining stability and reducing technical debt.The experimental results show that REJUV-CI achieves superior performance in accuracy, precision, recall, F1-score, and recovery time compared to traditional predictive models and manual debugging methods.REJUV-CI enables software systems to move towards autonomous development through its automated defect prediction and renewal processes, within CI/CD pipelines.
Authors: Priya Tyagi, Chunnu Lal, Himanshu Tyagi
Institution: Quantum University
Language: English
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W7202394791