以多目標深度學習建構台灣體適能資料中身心健康預測模型之研究

Taipei Medical University Institutional Repository > 管理學院 > 大數據科技及管理研究所 > 博碩士論文 > Item 987654321/62892

請使用永久網址來引用或連結此文件: http://libir.tmu.edu.tw/handle/987654321/62892

題名:	以多目標深度學習建構台灣體適能資料中身心健康預測模型之研究 A study of Constructing a Multi-Objective Deep Learning Model for Predicting Physical and Mental Health in Taiwanese
作者:	黃士軒 HUANG, SHIH-HSUAN
貢獻者:	大數據科技及管理研究所碩士班張詠淳
關鍵詞:	多目標學習;深度學習;機器學習;體適能資料;身理健康;快樂程度;睡眠困難;神經網路 Multi-Objective;Deep Learning;Machine Learning;Physical Fitness;Health Status;Happiness;Sleep Status;Deep Neural Network;Contextual Long Short-Term Memory
日期:	2023-06-20
上傳時間:	2023-09-21 14:27:32 (UTC+8)
摘要:	隨著國人對自我健康意識抬頭，越來越多人開始重視自我健康與生活狀況，為了滿足人們對自我狀況的了解，在不必耗費大量的時間等待醫療檢測或是浪費醫療資源的情況下，以非醫學性、非檢驗性報告等結構化資料，進行實驗與研究，從模型預測中掌握自身的狀況，則顯得特別有意義。本研究運用輔仁大學所收集的全國體適能資料等結構化資料進行實驗，本研究的體適能資料可以分為七大部分的資料類型，這幫助我們從不同的面相與角度切入問題，針對自我現況進行審核，我們透過多目標深度學習方法，使用CLSTM模型，對睡眠品質、自我健康狀況、自我快樂程度進行預測，經由實驗證實，相較於一般的模型，本實驗所設計的(MUSCLE : multi-task learning CLSTM model)模型，在預測表現上則顯得更為出色，這幫助我們更準確的分析出個人身心狀況，本研究的目的，在於提早發現可能存在的身心健康或是睡眠困難等因素，而後規劃出相對應的解決方案與方法，從中改善可能造成其影響的變因，以幫助受測者的身心、生活更加健康，進而提升國人健康水準，強化國人對自我健康的意識與認識，造就更加健康與快樂的生活。 With the rising awareness of self-health in Taiwan, more and more people are beginning to value their own well-being and lifestyle. In order to meet people's need, people understand their own conditions without having to spend a significant amount of time waiting for medical tests or wasting on healthcare resources, researching with structured data such as non-medical, non-diagnostic reports has become particularly meaningful and useful in Taiwan. This study utilized structured data, including nationwide fitness data collected by Fu Jen Catholic University, for experimentation. The fitness data in this study can be divided into seven major categories, allowing us to approach the problem from different perspectives, and conduct self-evaluations based on individual circumstances. We use multi-objective deep learning methods, and employed the CLSTM model to predict sleep quality, health status, and personal happiness. Experimental results have demonstrated that our designed MUSCLE model outperforms machine learning models in terms of prediction performance. This method helps us to analyze personal physical and mental conditions more accurately. The purpose of this research is to detect potential factors related to physical and mental health or sleep quality at the early stage. We plan appropriate solutions to improve the identified influential factors, helping individuals achieve better physical and mental health. Ultimately, this contributes to enhancing the overall health level of the Taiwan population and strengthening their awareness of self-health.
描述:	碩士指導教授：張詠淳委員：張詠淳委員：何健章委員：蘇家玉
資料類型:	thesis
顯示於類別:	[大數據科技及管理研究所] 博碩士論文

文件中的檔案:

檔案	描述	大小	格式	瀏覽次數
index.html		0Kb	HTML	59	檢視/開啟

在TMUIR中所有的資料項目都受到原著作權保護.

TAIR相關文章

資料載入中.....