Periodic Reporting for period 1 - DeepPatient (Deep Understanding of Patient Experience of Healthcare from Social Media)
Okres sprawozdawczy: 2018-09-01 do 2020-08-31
The project aims to develop an automated tool to process large-scale social media data in order to understand patient feedback and provide a decision support dashboard for health care professionals and senior decision-makers to allow for timely responses to address patients’ concerns. In particular, it will extract information relating to patient feedback and experience, automatically map the extracted opinions into various aspects of health care services, discover connections between elements that result in a perception of low and high quality of service, and present results in a visual dashboard to facilitate timely interventions.
The implementation of this project includes the following key progressive objectives:
(a) Construct ontology which captures expert domain knowledge and aspects relating to health care services from data.
(b) Develop unsupervised Bayesian modelling approaches for aspect-level opinion extraction from open data sources.
(c) Develop low dimensional visual informatics of mixed high dimensional data sources containing text and temporal sequences.
(d) Validate the proposed framework via a set of demonstrator applications.
The implementation of this project includes the following key progressive objectives:
(a) Construct ontology which captures expert domain knowledge and aspects relating to health care services from data.
(b) Develop unsupervised Bayesian modelling approaches for aspect-level opinion extraction from open data sources.
(c) Develop low dimensional visual informatics of mixed high dimensional data sources containing text and temporal sequences.
(d) Validate the proposed framework via a set of demonstrator applications.
The project has collected over 300K labelled and unlablled patient reviews for model development. The keywords in the collection have been mapped to the aspects of health care services. On the constructed dataset, the project has proposed and implemented several novel approaches with 12 peer-reviewed publication in top-tier conferences or high impact factor journals including:
1) A novel framework based on reinforcement learning for topic modelling in order to improve the topic coherence measures. The experimental result on the patient reviews shows that it beats several existing topic models in generating more coherent topics. The technical training for the Fellow focused on the knowledge of reinforcement learning. The work has been published in EMNLP 2019, which is one of the top conferences in Natural Language Processing (NLP).
2) A convolutional attention-based neural sentiment model for clinical text classification. The experimental result shows that the proposed method beats several strong baselines on the collected dataset. The visualization of learned representation also gives a better interpretation of embedding space. The training in this WP for the Fellow includes data visualisation. The work has been published in IEEE ACCESS (impact factor: 3.557).
3) A joint learning method with Generative Adversarial Network (GAN) which improves the performance both in aspect extraction and sentiment classification compared with the neural topic model and neural sentiment model. It improved the performance both in aspect extraction and sentiment classification compared with the neural topic model and neural sentiment model. The work has been accepted by IEEE Transactions on Knowledge and Data Engineering (impact factor: 4.935).
4) A weakly-supervised model which can automatically discover new words describing patient experience through the statistical learning process and simultaneously capture aspect-level opinions on clinical text. It improved the performance both in aspect extraction and sentiment classification compared with the pre-trained language neural topic model and neural sentiment model. This work has been submitted to Artificial Intelligence In Medicine. (Impact factor: 4.383)
5) An Adversarial Multi-task Learning Framework to identify the aspect-invariant/dependent sentiment expressions automatically without requiring extra annotations. Experimental results on two benchmark datasets show that extending existing neural models using our proposed framework achieves superior performance. In addition, the aspect-invariant data extracted by our framework can be considered as pivot features for better transfer learning of the ABSA models on unseen aspects. This work has been accepted by CIKM 2020, which is one of the top-tier conferences in Data Mining and Information Retrieval.
All of the proposed methods above are evaluated on public dataset or project collected dataset and achieve a state of the art performance. Besides, we also implement the proposed method on relevant natural language processing and machine learning application such as emotion cause extraction, question answering, binding prediction on DNA sequence, and stance detection on social media. Most of those works appear in top-tier conferences or high-impact factor journals such as ACL, EMNLP, COLING, and Bioinformatics.
1) A novel framework based on reinforcement learning for topic modelling in order to improve the topic coherence measures. The experimental result on the patient reviews shows that it beats several existing topic models in generating more coherent topics. The technical training for the Fellow focused on the knowledge of reinforcement learning. The work has been published in EMNLP 2019, which is one of the top conferences in Natural Language Processing (NLP).
2) A convolutional attention-based neural sentiment model for clinical text classification. The experimental result shows that the proposed method beats several strong baselines on the collected dataset. The visualization of learned representation also gives a better interpretation of embedding space. The training in this WP for the Fellow includes data visualisation. The work has been published in IEEE ACCESS (impact factor: 3.557).
3) A joint learning method with Generative Adversarial Network (GAN) which improves the performance both in aspect extraction and sentiment classification compared with the neural topic model and neural sentiment model. It improved the performance both in aspect extraction and sentiment classification compared with the neural topic model and neural sentiment model. The work has been accepted by IEEE Transactions on Knowledge and Data Engineering (impact factor: 4.935).
4) A weakly-supervised model which can automatically discover new words describing patient experience through the statistical learning process and simultaneously capture aspect-level opinions on clinical text. It improved the performance both in aspect extraction and sentiment classification compared with the pre-trained language neural topic model and neural sentiment model. This work has been submitted to Artificial Intelligence In Medicine. (Impact factor: 4.383)
5) An Adversarial Multi-task Learning Framework to identify the aspect-invariant/dependent sentiment expressions automatically without requiring extra annotations. Experimental results on two benchmark datasets show that extending existing neural models using our proposed framework achieves superior performance. In addition, the aspect-invariant data extracted by our framework can be considered as pivot features for better transfer learning of the ABSA models on unseen aspects. This work has been accepted by CIKM 2020, which is one of the top-tier conferences in Data Mining and Information Retrieval.
All of the proposed methods above are evaluated on public dataset or project collected dataset and achieve a state of the art performance. Besides, we also implement the proposed method on relevant natural language processing and machine learning application such as emotion cause extraction, question answering, binding prediction on DNA sequence, and stance detection on social media. Most of those works appear in top-tier conferences or high-impact factor journals such as ACL, EMNLP, COLING, and Bioinformatics.
We expect the project to have a major impact on the following beneficiaries outside the academic research community:
- Healthcare professionals. Large numbers of health care professionals currently collate and try to respond to insights from patient experience data (including Medical Directors, Chief Nurses, patient safety leads, and so on). The insights provided by this project will directly benefit patient care and simplify/improve the contribution of health care professionals to continuing service improvement.
- Healthcare and social care organisations. This project will help healthcare and social care organisations to make better use of patient experience data and to improve the quality of decision-making and of patient care. The project will be of significant interest to other health care providers (in the community, and in the public, private and voluntary sectors), as well as to social care providers.
- Individual data analysts. The project will develop a novel aspect-level opinion mining algorithm; this is particularly crucial to sentiment analysis and opinion mining. Data analysts will be provided with tools that will enable them to be more effective in their exploration and more persuasive with the results of that analysis.
- The public. Patients and their families are repeatedly asked for their feedback on NHS services, with little clarity as to how this feedback is collated, analysed and used. This project could therefore enable a very different dialogue with patients and the public, with greater clarity as to how NHS organisations are responding to the big themes being contributed by their patients.
- Healthcare professionals. Large numbers of health care professionals currently collate and try to respond to insights from patient experience data (including Medical Directors, Chief Nurses, patient safety leads, and so on). The insights provided by this project will directly benefit patient care and simplify/improve the contribution of health care professionals to continuing service improvement.
- Healthcare and social care organisations. This project will help healthcare and social care organisations to make better use of patient experience data and to improve the quality of decision-making and of patient care. The project will be of significant interest to other health care providers (in the community, and in the public, private and voluntary sectors), as well as to social care providers.
- Individual data analysts. The project will develop a novel aspect-level opinion mining algorithm; this is particularly crucial to sentiment analysis and opinion mining. Data analysts will be provided with tools that will enable them to be more effective in their exploration and more persuasive with the results of that analysis.
- The public. Patients and their families are repeatedly asked for their feedback on NHS services, with little clarity as to how this feedback is collated, analysed and used. This project could therefore enable a very different dialogue with patients and the public, with greater clarity as to how NHS organisations are responding to the big themes being contributed by their patients.