Multi-modal artificial intelligence-driven code quality prediction and self-healing automation using self-modulating convolutional neural networks

Ensuring software code quality and reliability in contemporary large-scale systems is an important problem with the soaring complexity of software architecture, varied data modalities, and adaptive execution environments. This research proposes a Multi-Modal Artificial Intelligence (AI)-Driven Code Quality Prediction (CQP) and Self-Healing Automation framework using a Self-Modulating Convolutional Neural Network framework (CQP-SHA-SMCNN). The framework starts with a data gathering stage that combines diverse modalities of software artifacts such as source code repositories, static analysis reports, run logs, and developer comments. These heterogeneous sources are transformed into homogeneous feature spaces using Discrete Wavelet Transform (DWT) based feature extraction to represent the structural, semantic, and behavioral characteristics of the code. The Self-Healing Automation (SHA) Pipeline observes software systems, forecasting imminent quality degradations and runtime. In case a fault is detected, the pipeline applies the Self-Modulating Convolutional Neural Network (SMCNN) model to perform root-cause analysis and triggers self-healing, such as code patch proposals, automatic refactoring, or runtime reinit. This research proposes a Multi-Modal Artificial Intelligence-driven framework for software code quality prediction and self-healing automation, in which the implemented artificial intelligence technique is a Self-Modulating Convolutional Neural Network applied to automated software quality assessment, fault diagnosis, and self-healing maintenance. Performance is evaluated using Receiver Operating Characteristic (ROC) curves, Precision, Recall, F-measure (F1-score), Mean Absolute Error (MAE), and Root Mean Square Error (RMSE), and compared with existing methods. The study demonstrates the potential of AI-driven Deep Learning (DL) models for automated software quality prediction and intelligent self-healing maintenance in modern environments.