The main objective of this work is to recommend a diet to different individuals. The recommender system deals with a large volume of information present by filtering the most important information based on the data provided by a user and other factors that take care of the user’s preference and interest.
Modern data have revealed that patients who follow a robust diet prescribed by a dietician or who use an Artificial Intelligent automated medical diet-based cloud system live longer, have fewer diseases, and have a higher quality of life. Medical workers, on the other hand, have yet to fully comprehend the patient-dietician reason for the recommender system. This work provides a deep learning solution for health-related datasets that automatically recognizes which meal should be provided to which patient based on disease and other characteristics such as age, gender, weight, calories, protein, fat, sodium, fiber, and cholesterol.
The goal of this research framework is to implement both machine and deep learning algorithms such as logistic regression, naive bayes, Recurrent Neural Network (RNN), Multilayer Perceptron (MLP), Gated Recurrent Units (GRU), and Long Short-Term Memory (LSTM). The medical dataset, which was gathered via the internet and hospitals, comprises of 30 patients’ data, 13 illness characteristics, and 1000 products. There are eight features in the product area. Before applying deep, machine, and learning-based protocols, the attributes of these IoMT data were evaluated and further encoded. The results show that the LSTM technique outperforms other schemes in terms of predicting accuracy, recall, precision, and F1-measures when compared to other machine learning and deep learning techniques. Using the LSTM deep learning model, we got 97.74 percent accuracy. Similarly, the allowed class achieves 98 percent precision, 99 percent recall, and 99 percent F1-measure, while the not-allowed class achieves 89 percent precision, 73 percent recall, and 80 percent F1-measure.
A recommendation system for patients/dieticians is a system that watches a user (patient/dietician) in a tailored approach towards remarkable or acceptable diets or food intake in a broad variety of possible options, and that produces the desired output. A patient/dietician recommendation system is carefully implemented with the goal of encouraging patients to adopt nutritional supplements, diets, and foods that are better suited to their health needs, taste, and dietary preferences.
Existing Model of System: –
For the required daily nutrition, an artificial bee colony algorithm was used to create a tailored meal suggestion system. The authors proposed a system that makes food and nutrition recommendations effectively using rule-based reasoning and fuzzy ontology, as well as a genetic algorithm for menu generation. Unfortunately, the system relied on the user’s Google Fit Application Programming Interface (API) for information about their daily activities and energy needs. The system also made individualized food recommendations for users based on the patient’s previous disease record, meaning that this system may not be suited for users who do not have access to a condition record. The number of users who can profit from the system is severely limited as a result of the preceding.
Proposed Model of System: –
We applied RNN, GRU, and LSTM deep learning classifiers, as well as na?ve Bayes and logistic regression machine learning classifiers, for this. A random forest classifier was used to determine which character in the dataset has the most impact. Illustrates the suggested model, which is divided into six phases. The first phase is data reading, the second phase is data pre-processing, and the third phase is data analysis and visualization of the best features. Training and testing are the fourth and fifth steps, respectively, with the assessment phase coming last.
- DATA PROCESSING
- MACHINE LEARNING CLASSIFIERS
- EVALUATION METRICS
Experiments are carried out on a Core?I3 system with Colab, which has an 8GB RAM and a 13GB Google Colab laboratory. Experiments are carried out on four CPUs, each with a 1.7 GHz processor. The training set, cross-validation set, and testing set are the three portions of the data set. Cross-validation with KFold was utilized. 70% of the dataset is used as a training set, while the remaining 30% is used for testing purposes. For both the training and testing sets, cross-validation is used.
System Requirements: –
- OS ? Windows 7, 8 and 10 (32 and 64 bit)
- RAM ? 4GB
- Anaconda navigator
- Python built-in modules