A Study Of College Students' Lifestyle Regularity Based On Wearable Devices And Deep Learning

The group of college students is the backbone of the country's construction and development,and the main force of the country's future progress.Different lifestyles have different effects on the health of college students.Therefore,a regular lifestyle is particularly important to the physical and mental development of college students.However,in recent years,the analysis of the physical health of college students in my country shows that the health level of college students is showing a downward trend.Statistics show that colleges are currently living with various health problems.For example,some college students often suffer from staying up late,giving up on themselves,and indulging in games,and some college students often overeating and lack of physical exercise.There is no doubt that this part of college students needs to arouse our extensive attention.In recent years,with the improvement of living standards and the development of wearable technology,more and more people have begun to pay attention to health issues,and smart wearable devices containing various sensors have begun to be widely used in medical care and daily life.According to the survey,there are many researches on the regularity of college students' life.In the existing research,there are mainly the following problems:(1)Most of the research on the regularity of college students' life is still at the stage of questionnaire survey and literature research.Few smart wearable devices are used in the research on the regularity of life,and the experimental results are more subjective;(2)In the experiments that use wearable devices to study the regularity of college students' daily life,most of the experiments are carried out with equipment that is bulky,expensive,and interferes with students' daily life.Inability to follow student status on a large scale.Therefore,the experimental results have certain limitations.Based on the above problems,in this work,we monitor the regularity of college students' life through use a wrist-type wearable device that is relatively light and will not affect the daily life of the students-the Xiaomi bracelet.We established a deep learning neural network model to classify the collected volunteer data,and proposed a method of combining wearable devices with the mixture of experts model MOE(Mixture of Experts)and transfer learning(Transfer learning)to improve the accuracy of model classification.The experimental results show that the classification accuracy of using MOE can be improved by 11% compared with that without MOE;by using MOE+transfer learning,the classification accuracy of the model can be further improved by 7%.And we conducted experiments on public data sets to prove that the MOE method used in this article can indeed improve the accuracy of model classification,and the MOE+ transfer learning method proposed in this article can further improve the accuracy of regular classification of students.