Development of personalized diet and exercise recommender system based on clinical data

Title

Development of personalized diet and exercise recommender system based on clinical data

Creator

Sheshrao, Vairale Vaishali

Contributor

Shukla, Samiksha

Description

The present health scenario indicates that thyroid diseases are a common challenge experienced by most individuals. According to the statistics in India, one out of eight women suffer from thyroid-related conditions. Hyperthyroid, hypothyroid, or thyroid cancer are categories of thyroid disorder. It is imperative to maintain optimum levels of secretion of the thyroid hormones as the imbalance could lead to thyroid diseases. Therefore, thyroid patients must be vigilant regarding their iodine intake and follow a customized daily diet and exercise plan. The diet plan, along with balanced iodine levels, must also be able to meet the patient's nutritional needs. A personalized diet plan could help thyroid patients to be more aware and focused on their body metabolism. Existing recommender systems usually provide generic diet recommendations, and unfortunately, it may not be beneficial to patients suffering from a specific disease. Content-based Neighborhood-Conditional RBM (CB-NCRBM) model has posited to recommend Top-3 diet and exercise plans for thyroid patients. The proposed model considers the joint probability distribution of different scores using the user profile. Similarly, preference and health scores are estimated based on content features. The model feeds these scores as visible units to conditional RBM. The proposed model also integrates several content-based features such as users' physiological profiles, thyroid disease information, food, and exercise preferences. The proposed recommender model validates the experimental results using recommendation error and classification accuracy metrics. The proposed hybrid model outperforms several popularly used recommendation models, such as collaborative filtering, content-based, and pure RBM models. The system also provides a feedback loop to enhance the quality of the recommended diet and exercise plans based on user experience.

Source

Author's Submission

Date

2020-01-01

Publisher

Christ(Deemed to be University)

Subject

Computer Science and Engineering

Rights

Open Access

Relation

61000134

Format

PDF

Language

English

Type

PhD

Identifier

http://hdl.handle.net/10603/313848