Telemonitoring is the use of electronic devices to remotely monitor patients. Taking the Parkinson's disease (PD) as an example, the use of at-home testing device (AHTD) enables remote, internet-based measurement of PD vocal symptoms. Translating AHTD measurement into a unified PD rating scale (UPDRS) through predictive analytics enables cost-effective, convenient, and close tracking of PD progression. Building a predictive model between AHTD measurement and UPDRS is not straightforward because PD patients are highly heterogeneous, which requires patient-specific models. Learning a patient-specific model faces the challenge of limited data. Transfer learning (TL) tackles this challenge by leveraging other patients' information to make up the data shortage when modeling a target patient. Among different TL methods, the category of parameter transfer methods is more appropriate for the telemonitoring application because it transfers patient-specific model parameters but not patients' data. However, existing parameter transfer methods fall short because not every other patient's information is helpful and blind transfer causes the problem of negative transfer. To tackle this limitation, we propose a positive TL (PTL) method. We provide an in-depth theoretical study on the risk and condition for negative transfer to happen, which further drive the development of novel PTL algorithms that are robust to negative transfer. We apply PTL to predict UPDRS of 42 PD patients using their AHTD vocal measurement. PTL achieves significantly better accuracy compared with single learning and one-model-fits-all approaches. Note to Practitioners-This paper was motivated by the growing use of telemonitoring devices to facilitate remote, close tracking of disease progression. These devices are usually paired up with predictive analytics to translate the measurement signals into a clinical indicator of disease progression. This is a challenging task because the predictive model needs to be patient specific in order to account for patient heterogeneity. This paper proposes a TL method called PTL that can leverage other patients' information when building a predictive model for a target patient. The unique feature of PTL is that it can intelligently select which patients to transfer from, and thus preventing negative transfer. PTL is a just-in-time development for the emerging healthcare paradigm of telemonitoring and telemedicine by providing predictive analytics models and algorithms to address the unique challenges and contributing to the use of these remote devices to improve patient care.
|Number of pages||12|
|Journal||IEEE Transactions on Automation Science and Engineering|
|Publication status||Published - 2019 Jan|
Bibliographical noteFunding Information:
Manuscript received August 29, 2018; accepted October 1, 2018. Date of publication November 2, 2018; date of current version January 4, 2019. This paper was recommended for publication by Associate Editor H. Yang and Editor J. Li upon evaluation of the reviewers’ comments. This work was supported by NSF CAREER under Grant 1149602. (Corresponding author: Jing Li.) The authors are with the Industrial Engineering Program, School of Computing, Informatics, and Decision Systems Engineering, Arizona State University, Tempe, AZ 85281 USA (e-mail: firstname.lastname@example.org; email@example.com).
Dr. Li is a member of IISE and INFORMS. She is a recipient of the NSF CAREER Award.
© 2004-2012 IEEE.
All Science Journal Classification (ASJC) codes
- Control and Systems Engineering
- Electrical and Electronic Engineering