Ayuda
Ir al contenido

Dialnet


Resumen de Relevant framework for social applications of internet of things by means of machine learning techniques

Xiaochen Zhang

  • The recent adaptation of wireless technologies and the widespread deployment of spatially distributed devices with embedded identification, sensing and actuation capabilities created the Internet of Things (IoT). This promising paradigm has been developing explosively in recent years. It is believed to be the next revolutionary technology by bringing the traditional sense of Internet into the physical realm. Billions of inter-connected smart things are streaming out a huge amount of data every moment and promote the world into “big data” era. Unimaginable potential value can be mined from these data supported by advanced data storage and analysis technologies, such as machine learning and cloud computing.

    With the help of advanced data mining tools, IoT can bring great benefits for various domains of the society including healthcare. The healthcare industry has been dramatically changed because of the information technology revolution that started in the last century. New technologies such as telemedicine, digital hospital and e-health have been widely applied during the past decades and now the rapidly development of IoT and machine learning is promoting healthcare from digital into intelligent.

    As an important aspect of IoT, wearable technology has also shown a surge in the past decade. Different types of wearable devices containing various embedded sensors have been introduced into the market with affordable prices. Large amounts of health-related data are generated by these wearable devices related to different daily activities. These low-cost data, supported by mobile computing and machine learning techniques, make it possible to develop Smart Decision Support Systems (SDSS) which can be beneficial to the long-term activity monitoring, remote disease diagnosis, emergency medical alerts promoting and so on. \\ As one of the most important tools to realize Artificial Intelligence (AI), machine learning has been growing explosively in the past decades with the development of Internet. Various of machine learning techniques have been widely used to implement different data mining tasks, among which, deep learning has shown outstanding performance in recent years due to the availability of big data.

    Our research aims to address the applications of IoT technology supported by advanced machine learning techniques in different social areas, especially in healthcare. A general application framework was constructed, which includes data collecting and transferring, data storage and analysis, and analysis result sharing. In order to verify the feasibility of proposed application framework, a practical human movement data collecting system was developed based on wearable technology. It includes three modules: a smart watch, a smartphone and a remote NoSQL database. The system was applied in a hospital to collect daily activity and tremor data from patients with Essential Tremor (ET). Advanced machine learning techniques, including deep learning, were adopted to realize Human Activity Recognition (HAR) and ET evaluation tasks. Through proper data preprocessing and data transformation, based on the collected acceleration data, the proposed models could recognize a series of human daily activities and classify tremor levels with high accuracy. These models could enable neurologists remotely and continuously monitor ET patients' daily activities and the corresponding tremor situation. The evaluation result could help them to improve the treatment plans. This case proved the feasibility of the presented IoT application framework and similar applications could be developed in other scenarios.

    As one of the future research directions, a personal health data sharing system based on blockchain technology was proposed in the end of the study. The aim is to protect the privacy and security during the data sharing process.


Fundación Dialnet

Dialnet Plus

  • Más información sobre Dialnet Plus