Dissertation
Learning from Unstructured Data to Monitor Human Health
Washington State University
Doctor of Philosophy (PhD), Washington State University
12/2019
DOI:
https://doi.org/10.7273/000005558
Abstract
The integration of mobile devices into our daily lives has created unique opportunities to improve human health and well-being. Many of these devices such as smartphones and smartwatches allow the users to enter unstructured data such as speech. This research is focused on utilizing such data for health monitoring through development of computational algorithms and optimization strategies that process unstructured data, compute health-related markers, and provide recommendations for improved health. The applications of this research include nutrition monitoring, dietary recommendation, personality assessment, and commonsense reasoning. Diet is known as an important lifestyle factor in self-management and prevention of chronic diseases. Although mobile and wearable sensors have been used to estimate eating context, accurate monitoring of dietary intake has remained a challenging problem. New approaches based on mobile devices have been proposed to facilitate the process of food intake recording. These technologies require individuals to use mobile devices such as smartphones to record nutrition intake by either entering text or taking images of the food. These technologies are prone to measurement errors related to challenges of human memory and bias. In order to address these limitations, we introduced development and validation of two nutrition monitoring frameworks, Speech2Health and EZNutriPal, which use unstructured data along with speech processing, natural language processing (NLP), and text mining techniques to facilitate dietary assessment.Implementing strategies that improve dietary intake is also very important. A general diet behavior change framework for joint nutrition monitoring and diet planning allows continuous diet recommendations for achieving a diet goal. This research introduces a diet planning framework, called iTell-uEat, to provide diet recommendations continuously based on user's diet habits. Two approaches are proposed including a reinforcement-learning-based and a greedy-based diet planning. An optimization algorithm is proposed to construct a meaningful action space for training reinforcement learning algorithms. Moreover, a linear optimization approach is developed forgreedy diet planning. To demonstrate the potential of utilizing unstructured data in applications beyond dietary assessment, a computational framework is proposed to analyze human personality traits based on expressed texts and to use these personalities for behavior and commonsense reasoning analysis.
Metrics
2 File views/ downloads
11 Record Views
Details
- Title
- Learning from Unstructured Data to Monitor Human Health
- Creators
- Niloofar Hezarjaribi
- Contributors
- Hassan Ghasemzadeh (Chair)Assefaw Gebremedhin (Committee Member) - Washington State University, Electrical Engineering and Computer Science, School ofVenera Arnaoudova (Committee Member) - Washington State University, Electrical Engineering and Computer Science, School ofChristopher Hundhausen (Committee Member) - Washington State University, Electrical Engineering and Computer Science, School of
- Awarding Institution
- Washington State University
- Academic Unit
- Electrical Engineering and Computer Science, School of
- Theses and Dissertations
- Doctor of Philosophy (PhD), Washington State University
- Publisher
- Washington State University
- Number of pages
- 199
- Identifiers
- 99901054232301842
- Language
- English
- Resource Type
- Dissertation