An Optimized Hyperparameter of Convolutional Neural Network Algorithm for Bug Severity Prediction in...

Area of Science:

• Software Engineering
• Medical Informatics
• Artificial Intelligence

Background:

• Software is integral to healthcare, from patient management to advanced medical devices.
• Internet of Things (IoT) medical devices, used for monitoring conditions like Alzheimer's and Parkinson's, increasingly rely on complex embedded software.
• Software bugs in these critical devices pose significant risks, including patient harm and delayed care.

Purpose of the Study:

• To develop a robust model for predicting the severity of software bugs in healthcare systems and IoT medical devices.
• To address the critical need for accurate bug severity prediction in high-stakes medical technology.

Main Methods:

• A hybrid bug severity prediction model was proposed, integrating Convolutional Neural Network (CNN) with Harris Hawk Optimization (HHO).
• The HHO algorithm was used to optimize CNN hyperparameters, including batch size, learning rate, and activation functions.
• A dataset of bugs from healthcare systems and IoT medical devices was curated and preprocessed for model training and evaluation.

Main Results:

• The proposed hybrid CNN-HHO model achieved a high accuracy of 96.21% in predicting bug severity.
• A 10-fold cross-validation was performed to rigorously evaluate the model's performance.
• The model demonstrated effectiveness in classifying bug severity for critical healthcare software.

Conclusions:

• The hybrid CNN-HHO model offers a promising solution for enhancing the safety and reliability of software in healthcare IoT devices.
• Accurate bug severity prediction is crucial for mitigating risks associated with software failures in medical technology.
• This research contributes to improving patient safety through advanced AI-driven software quality assurance in healthcare.