bebi 20: e3

Research Article

Well-being Forecasting using a Parametric Transfer-Learning method based on the Fisher Divergence and Hamiltonian Monte Carlo

Download8 downloads
  • @ARTICLE{10.4108/eai.16-10-2020.166661,
        author={Eirini Christinaki and Tasos Papastylianou and Sara Carletto and Sergio Gonzalez-Martinez and Luca Ostacoli and Manuel Ottaviano and Riccardo Poli and Luca Citi},
        title={Well-being Forecasting using a Parametric Transfer-Learning method based on the Fisher Divergence and Hamiltonian Monte Carlo},
        journal={EAI Endorsed Transactions on Bioengineering and Bioinformatics},
        volume={},
        number={},
        publisher={EAI},
        journal_a={BEBI},
        year={2020},
        month={10},
        keywords={Transfer Learning, MCMC, Bayesian Inference, Well-being Prediction, Personalised Modelling, NEVERMIND},
        doi={10.4108/eai.16-10-2020.166661}
    }
    
  • Eirini Christinaki
    Tasos Papastylianou
    Sara Carletto
    Sergio Gonzalez-Martinez
    Luca Ostacoli
    Manuel Ottaviano
    Riccardo Poli
    Luca Citi
    Year: 2020
    Well-being Forecasting using a Parametric Transfer-Learning method based on the Fisher Divergence and Hamiltonian Monte Carlo
    BEBI
    EAI
    DOI: 10.4108/eai.16-10-2020.166661
Eirini Christinaki1, Tasos Papastylianou1, Sara Carletto2, Sergio Gonzalez-Martinez3, Luca Ostacoli4, Manuel Ottaviano3, Riccardo Poli1, Luca Citi1,*
  • 1: School of Computer Science and Electronic Engineering, University of Essex, Colchester, UK
  • 2: Department of Neuroscience “Rita Levi Montalcini”, Università degli Studi di Torino, Turin, Italy
  • 3: Life Supporting Technologies, Universidad Politécnica de Madrid, Madrid, Spain
  • 4: Department of Clinical and Biological Sciences, Università degli Studi di Torino, Turin, Italy
*Contact email: lciti@essex.ac.uk

Abstract

INTRODUCTION: Traditional personalised modelling typically requires sufficient personal data for training. This is a challenge in healthcare contexts, e.g. when using smartphones to predict well-being.

OBJECTIVE: A method to produce incremental patient-specific models and forecasts even in the early stages of data collection when the data are sporadic and limited.

METHODS: We propose a parametric transfer-learning method based on the Fisher divergence, where information from other patients is injected as a prior term into a Hamiltonian Monte Carlo framework. We test our method on the NEVERMIND dataset of self-reported well-being scores.

RESULTS: Out of 54 scenarios representing varying training/forecasting lengths and competing methods, our method achieved overall best performance in 50 (92.6%) and demonstrated a significant median difference in45 (83.3%).

CONCLUSION: The method performs favourably overall, particularly when long-term forecasts are required given short-term data.