Neural-signature methods for structured EHR prediction
Abstract Models that can effectively represent structured Electronic Healthcare Records (EHR) are central to an increasing range of applications in healthcare. Due to the sequential nature of health data, Recurrent Neural Networks have emerged as the dominant component within state-of-the-art archit...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
BMC
2022-12-01
|
| Series: | BMC Medical Informatics and Decision Making |
| Subjects: | |
| Online Access: | https://doi.org/10.1186/s12911-022-02055-6 |
| _version_ | 1811301613922418688 |
|---|---|
| author | Andre Vauvelle Paidi Creed Spiros Denaxas |
| author_facet | Andre Vauvelle Paidi Creed Spiros Denaxas |
| author_sort | Andre Vauvelle |
| collection | DOAJ |
| description | Abstract Models that can effectively represent structured Electronic Healthcare Records (EHR) are central to an increasing range of applications in healthcare. Due to the sequential nature of health data, Recurrent Neural Networks have emerged as the dominant component within state-of-the-art architectures. The signature transform represents an alternative modelling paradigm for sequential data. This transform provides a non-learnt approach to creating a fixed vector representation of temporal features and has shown strong performances across an increasing number of domains, including medical data. However, the signature method has not yet been applied to structured EHR data. To this end, we follow recent work that enables the signature to be used as a differentiable layer within a neural architecture enabling application in high dimensional domains where calculation would have previously been intractable. Using a heart failure prediction task as an exemplar, we provide an empirical evaluation of different variations of the signature method and compare against state-of-the-art baselines. This first application of neural-signature methods in real-world healthcare data shows a competitive performance when compared to strong baselines and thus warrants further investigation within the health domain. |
| first_indexed | 2024-04-13T07:11:56Z |
| format | Article |
| id | doaj.art-13b46dc6cd0942beb2a6cbc8b94365dd |
| institution | Directory Open Access Journal |
| issn | 1472-6947 |
| language | English |
| last_indexed | 2024-04-13T07:11:56Z |
| publishDate | 2022-12-01 |
| publisher | BMC |
| record_format | Article |
| series | BMC Medical Informatics and Decision Making |
| spelling | doaj.art-13b46dc6cd0942beb2a6cbc8b94365dd2022-12-22T02:56:50ZengBMCBMC Medical Informatics and Decision Making1472-69472022-12-0122111210.1186/s12911-022-02055-6Neural-signature methods for structured EHR predictionAndre Vauvelle0Paidi Creed1Spiros Denaxas2Institute of Health Informatics, University College LondonBenevolentAIInstitute of Health Informatics, University College LondonAbstract Models that can effectively represent structured Electronic Healthcare Records (EHR) are central to an increasing range of applications in healthcare. Due to the sequential nature of health data, Recurrent Neural Networks have emerged as the dominant component within state-of-the-art architectures. The signature transform represents an alternative modelling paradigm for sequential data. This transform provides a non-learnt approach to creating a fixed vector representation of temporal features and has shown strong performances across an increasing number of domains, including medical data. However, the signature method has not yet been applied to structured EHR data. To this end, we follow recent work that enables the signature to be used as a differentiable layer within a neural architecture enabling application in high dimensional domains where calculation would have previously been intractable. Using a heart failure prediction task as an exemplar, we provide an empirical evaluation of different variations of the signature method and compare against state-of-the-art baselines. This first application of neural-signature methods in real-world healthcare data shows a competitive performance when compared to strong baselines and thus warrants further investigation within the health domain.https://doi.org/10.1186/s12911-022-02055-6Machine learningElectronic healthcare recordsSignature methods |
| spellingShingle | Andre Vauvelle Paidi Creed Spiros Denaxas Neural-signature methods for structured EHR prediction BMC Medical Informatics and Decision Making Machine learning Electronic healthcare records Signature methods |
| title | Neural-signature methods for structured EHR prediction |
| title_full | Neural-signature methods for structured EHR prediction |
| title_fullStr | Neural-signature methods for structured EHR prediction |
| title_full_unstemmed | Neural-signature methods for structured EHR prediction |
| title_short | Neural-signature methods for structured EHR prediction |
| title_sort | neural signature methods for structured ehr prediction |
| topic | Machine learning Electronic healthcare records Signature methods |
| url | https://doi.org/10.1186/s12911-022-02055-6 |
| work_keys_str_mv | AT andrevauvelle neuralsignaturemethodsforstructuredehrprediction AT paidicreed neuralsignaturemethodsforstructuredehrprediction AT spirosdenaxas neuralsignaturemethodsforstructuredehrprediction |