Research Article
A Framework for Healthcare Data Integration Based on Model-as-a-Service
@INPROCEEDINGS{10.1007/978-3-031-32443-7_11, author={Yujie Fang and Chao Gao and Xinyue Zhou and Jianmao Xiao and Zhiyong Feng}, title={A Framework for Healthcare Data Integration Based on Model-as-a-Service}, proceedings={Mobile Networks and Management. 12th EAI International Conference, MONAMI 2022, Virtual Event, October 29-31, 2022, Proceedings}, proceedings_a={MONAMI}, year={2023}, month={5}, keywords={IoT Healthcare Model-as-a-Service Workflow Machine Learning Data Analysis}, doi={10.1007/978-3-031-32443-7_11} }- Yujie Fang
Chao Gao
Xinyue Zhou
Jianmao Xiao
Zhiyong Feng
Year: 2023
A Framework for Healthcare Data Integration Based on Model-as-a-Service
MONAMI
Springer
DOI: 10.1007/978-3-031-32443-7_11
Abstract
More and more IoT detection devices are entering into healthcare domain. They collect remote data through the MQTT protocol. Along with the chaos of server subscription, data format, message structure, and content parsing, how to realize the end-to-end “model-as-a-service” in the healthcare scenario is an issue worthy of further study. This paper designs and implements a healthcare data integration framework that integrates the whole process from detected data subscription to model training deployment and data analysis automatically based on workflow, and provides users with a low-code workflow configuration method. First, this paper defined a custom description language for the workflow of the integration problem. Next, fully considering the situation of message parsing and storage of different devices, we build an end-to-end healthcare integration framework that realizes the dynamic management of access data and subscription clients. In addition, it provides customization and AutoML-based automation options to select machine learning models and parameters. Finally, the experiment shows that the framework completes the dynamic subscription, parsing, storage, model training and deployment, and data analysis of various device messages. This framework can further integrate techniques such as streaming data analysis and deep learning automation to perform complex tasks in different scenarios like real-time data analysis of elderly care and medical diagnosis.
