Multi-Task Time Series Analysis applied to Drug Response Modelling
This work addresses the need for personalized medical models to enhance drug response predictions for patients, though it appears incremental as it adapts existing multi-task learning to time series.
The paper tackled the problem of personalizing time series models for individual patients while maintaining statistical power, using multi-task learning, and demonstrated improved predictive accuracy and uncertainty estimation in drug response modeling.
Time series models such as dynamical systems are frequently fitted to a cohort of data, ignoring variation between individual entities such as patients. In this paper we show how these models can be personalised to an individual level while retaining statistical power, via use of multi-task learning (MTL). To our knowledge this is a novel development of MTL which applies to time series both with and without control inputs. The modelling framework is demonstrated on a physiological drug response problem which results in improved predictive accuracy and uncertainty estimation over existing state-of-the-art models.