Food Image-based diet recommendation framework to overcome PCOS problem in women using deep convolutional neural network. (October 2022)

• Record Type: Journal Article
• Authors: Kaur, Rajdeep; Kumar, Rakesh; Gupta, Meenu
• Is Part Of: Computers & electrical engineering. Volume 103(2022)
• Publication Date: 2022-10
• Subjects: PCOS -- CNN -- RF -- EfficientNet (B0-B7) -- Food image analysis -- Nutritional management systems -- Deep learning (DL) -- K-means; Computer engineering -- Periodicals; Electrical engineering -- Periodicals; Electrical engineering -- Data processing -- Periodicals; Ordinateurs -- Conception et construction -- Périodiques; Électrotechnique -- Périodiques; Électrotechnique -- Informatique -- Périodiques; Computer engineering; Electrical engineering; Electrical engineering -- Data processing; Periodicals; Electronic journals; 621.302854
• DOI: 10.1016/j.compeleceng.2022.108298
• Languages: English
• ISSNs: 0045-7906

Abstract

Highlights: EfficientNet (B0-B7) model variants are enhanced by adding four layers to classify food images and compared with the other CNN models such as VGG-16, VGG-19, ResNet-50, and ResNet-101. The number of macronutrients (fat, protein, and carbohydrates) available in the food item is identified using a nutritional information dataset collected from the USDA. The low carbohydrate diet composition with macronutrient distribution is selected for the PSOS women. The K-means algorithm is used to divide the nutritional information dataset into clusters, and the Random Forest algorithm is used to find the closest cluster based on the nutritional needs of the user to recommend a list of foods. Abstract: Polycystic Ovary Syndrome (PCOS) is a disorder that affects the reproductive, metabolic, and hormonal systems in women. The main cause behind PCOS is not exactly known by the practitioners but deficiency of the nutrients in diet may cause PCOS or its related disorders. A healthy diet with proper nutrient intake plays an important role to overcome the issue of PCOS. A novel framework is designed for managing weight with proper nutrient intake using Artificial Intelligence (AI). Pre-trained Convolutional Neural Networks (CNN) architecture (i.e., EfficientNet model variants B0-B7) is enhanced and fine-tuned to classify the food images by adding additional four layers (i.e., Augmentation layer, Dense layer, Dropout layer with dropout 0.3, and Final output classifier dense layer) on datasets derived from FOOD-101. The performance of the model is compared with the other pre-trained models such as VGG16, VGG19, ResNet50, and ResNet101 using performance evaluation metrics. After execution, 95.5% accuracy is achieved to classify the sample six food classes and 90.7% accuracy is achieved for twelve food image classes respectively. Further, K-means and Random Forest (RF) algorithms are applied to create clusters and recommend the list of foods to the PCOS patient with 97% accuracy. Graphical abstract: Image, graphical abstract