Akksatcha Duangsuphasin.. The development of a decision support system using deep learning methods to select passion fruit for the aging society. Doctoral Degree(Industrial Engineering). King Mongkut's University of Technology North Bangkok. Central Library. : King Mongkut's University of Technology North Bangkok, 2024.
The development of a decision support system using deep learning methods to select passion fruit for the aging society
Abstract:
As the aging population grows, chronic health issues are increasing, necessitating proper nutritional guidance. Passion fruit, rich in nutrients, can help alleviate these conditions, but recommended daily intake for the elderly is not well-established. This study aims to classify passion fruit and suggest appropriate daily intake for elderly individuals using a Decision Support System. Three models and one software are compared: the Network in Network (NiN), the Multilayer Perceptron Neural Network (MLPNN), a hybrid MLPNN integrated with the Tunicate Swarm Algorithm (TSA), and the EaglAI software. Passion fruit image is photographed in a studio box with a white background and LED lighting to ensure consistent image quality. The NiN model achieved the highest accuracy, with 96.76% on the training set and 95.89% on the validation set, outperforming EaglAI (84.6%). The MLPNN model attained 61.9% training accuracy and 63.7% on the test set, while the MLPNN hybrid TSA showed slight improvement (65.0% test accuracy). The NiN model's sensitivity analysis highlights the blue channel as most impactful, suggesting targeted optimization. For the MLPNN model, Parameter 4-Woman (age 60-69) is critical, with Parameters 6-Woman, age 80 years ago and Parameter 8-Hypertension also influencing outcomes. The benefit of this research is that Thai elderly people in each age group who have chronic diseases can consume an appropriate amount of passion fruit nutrients per day, according to a result of the NiN model based upon sugar nutrients.