Adaptive multi-task positive-unlabeled learning for joint prediction of multiple chronic diseases using online shopping behaviors

作者:

Highlights:

• Joint prediction of multiple chronic diseases using online shopping behaviors.

• Propose a novel multi-task positive-unlabeled learning paradigm.

• Open-source a new synthetic dataset from a real e-commerce site.

• Achieve high predictive performance for depression and diabetes simultaneously.

• Provide innovative insights into chronic disease surveillance.

摘要

•Joint prediction of multiple chronic diseases using online shopping behaviors.•Propose a novel multi-task positive-unlabeled learning paradigm.•Open-source a new synthetic dataset from a real e-commerce site.•Achieve high predictive performance for depression and diabetes simultaneously.•Provide innovative insights into chronic disease surveillance.

论文关键词:Chronic disease prediction,Online shopping behaviors,Depression,Diabetes,Multi-task learning,Positive-Unlabeled learning

论文评审过程:Received 14 August 2021, Revised 3 October 2021, Accepted 12 November 2021, Available online 27 November 2021, Version of Record 4 December 2021.

论文官网地址:https://doi.org/10.1016/j.eswa.2021.116232