Multitask Gaussian processes for multivariate physiological time-series analysis.
Gaussian process (GP) models are a flexible means of performing nonparametric Bayesian regression. However, GP models in healthcare are often only used to model a single univariate output time series, denoted as single-task GPs (STGP). Due to an increasing prevalence of sensors in healthcare setti...
| Main Authors: | , , , , |
|---|---|
| Format: | Journal article |
| Language: | eng |
| Published: |
IEEE
2015
|
| _version_ | 1797054668311887872 |
|---|---|
| author | Dürichen, R Pimentel, M Clifton, L Schweikard, A Clifton, D |
| author_facet | Dürichen, R Pimentel, M Clifton, L Schweikard, A Clifton, D |
| author_sort | Dürichen, R |
| collection | OXFORD |
| description | Gaussian process (GP) models are a flexible means of performing nonparametric Bayesian regression. However, GP models in healthcare are often only used to model a single univariate output time series, denoted as single-task GPs (STGP). Due to an increasing prevalence of sensors in healthcare settings, there is an urgent need for robust multivariate time-series tools. Here, we propose a method using multitask GPs (MTGPs) which can model multiple correlated multivariate physiological time series simultaneously. The flexible MTGP framework can learn the correlation between multiple signals even though they might be sampled at different frequencies and have training sets available for different intervals. Furthermore, prior knowledge of any relationship between the time series such as delays and temporal behavior can be easily integrated. A novel normalization is proposed to allow interpretation of the various hyperparameters used in the MTGP. We investigate MTGPs for physiological monitoring with synthetic data sets and two real-world problems from the field of patient monitoring and radiotherapy. The results are compared with standard Gaussian processes and other existing methods in the respective biomedical application areas. In both cases, we show that our framework learned the correlation between physiological time series efficiently, outperforming the existing state of the art. |
| first_indexed | 2024-03-06T19:00:24Z |
| format | Journal article |
| id | oxford-uuid:13522376-845d-41e8-9fc8-81a17d33e2d6 |
| institution | University of Oxford |
| language | eng |
| last_indexed | 2024-03-06T19:00:24Z |
| publishDate | 2015 |
| publisher | IEEE |
| record_format | dspace |
| spelling | oxford-uuid:13522376-845d-41e8-9fc8-81a17d33e2d62022-03-26T10:13:24Z Multitask Gaussian processes for multivariate physiological time-series analysis. Journal articlehttp://purl.org/coar/resource_type/c_dcae04bcuuid:13522376-845d-41e8-9fc8-81a17d33e2d6engSymplectic Elements at OxfordIEEE2015Dürichen, RPimentel, MClifton, LSchweikard, AClifton, D Gaussian process (GP) models are a flexible means of performing nonparametric Bayesian regression. However, GP models in healthcare are often only used to model a single univariate output time series, denoted as single-task GPs (STGP). Due to an increasing prevalence of sensors in healthcare settings, there is an urgent need for robust multivariate time-series tools. Here, we propose a method using multitask GPs (MTGPs) which can model multiple correlated multivariate physiological time series simultaneously. The flexible MTGP framework can learn the correlation between multiple signals even though they might be sampled at different frequencies and have training sets available for different intervals. Furthermore, prior knowledge of any relationship between the time series such as delays and temporal behavior can be easily integrated. A novel normalization is proposed to allow interpretation of the various hyperparameters used in the MTGP. We investigate MTGPs for physiological monitoring with synthetic data sets and two real-world problems from the field of patient monitoring and radiotherapy. The results are compared with standard Gaussian processes and other existing methods in the respective biomedical application areas. In both cases, we show that our framework learned the correlation between physiological time series efficiently, outperforming the existing state of the art. |
| spellingShingle | Dürichen, R Pimentel, M Clifton, L Schweikard, A Clifton, D Multitask Gaussian processes for multivariate physiological time-series analysis. |
| title |
Multitask Gaussian processes for multivariate physiological time-series analysis.
|
| title_full |
Multitask Gaussian processes for multivariate physiological time-series analysis.
|
| title_fullStr |
Multitask Gaussian processes for multivariate physiological time-series analysis.
|
| title_full_unstemmed |
Multitask Gaussian processes for multivariate physiological time-series analysis.
|
| title_short |
Multitask Gaussian processes for multivariate physiological time-series analysis.
|
| title_sort | multitask gaussian processes for multivariate physiological time series analysis |
| work_keys_str_mv | AT durichenr multitaskgaussianprocessesformultivariatephysiologicaltimeseriesanalysis AT pimentelm multitaskgaussianprocessesformultivariatephysiologicaltimeseriesanalysis AT cliftonl multitaskgaussianprocessesformultivariatephysiologicaltimeseriesanalysis AT schweikarda multitaskgaussianprocessesformultivariatephysiologicaltimeseriesanalysis AT cliftond multitaskgaussianprocessesformultivariatephysiologicaltimeseriesanalysis |