Programmes

HealTAC 2000 will have three local “hubs” (see the list below) where participants can get together face-to-face locally to follow the conference. Coffee and light lunch will be provided. Please contact the local organiser below if you would like to join your local hub.

Pre-conference: afternoon of 14th June 2022

Start End Event Chair
13:00 15:00 Tutorial 1: Patient and Public Involvement and Engagement (PPIE): Hands on Guidance for Clinical Text Analytics Ms Natalie Fitzpatrick
&
Ms Jenny Robertson
Click to expand details...
Short description
Patient and public involvement and engagement (PPIE) plays a vital role in helping to improve research and is now a crucial part of funding applications. As an early career researcher, one might find it difficult to take the first step of PPIE and to navigate through options and practices. In this tutorial, our PPIE veterans, Jenny Robertson and Natalie Fitzpatrick, and experienced public contributor, Colin Wilkinson, will guide you through a full journey of PPIE in relation to research projects on clinical text analytics. Hands-on guidance on how to develop and cost PPIE sections in a grant application and what to do next when a project is funded will be provided. By the end of the tutorial, you will have the conceptual understanding of PPIE and know the options, resources and tools available to embark on your PPIE journey.

(BYOP!!) We highly recommend you to Bring Your Own Project for the PPIE clinic. You will benefit from the direct advice on your project from our PPIE experts.
15:00 15:30 Break
15:30 17:30 Tutorial 2: De-identification of clinical and medical texts using MASK and MedCAT Drs Nikola Milosevic
&
Zeljko Kraljevic
Click to expand details...
Short description:
This session will introduce de-identification of medical and clinical documents as one of the enablers for data sharing in biomedicine. We will introduce benefits and challenges of automated de-identification using machine learning. We will introduce two tools that allow users to utilize machine learning-based de-identification of medical and clinical texts, namely MASK and MadCAT. We will start by describing components of these systems and will present how users can train, fine-tune and use the named systems. Attendees by the end will have been:
  • Introduced to de-identification of medical and clinical documents, its benefits and challenges
  • Familiarised with the architecture of MASK framework and its plugin-based architecture for components
  • Introduced to Transformers (BERT) and how they can be used to de-identify clinical documents
  • Given the opportunity to train or fine-tune a model for de-identification of clinical documents using MASK and MedCAT
  • Shown how to validate the created model on a subset of documents
  • Instructed on how to utilize MASK/MedCAT for larger scale de-identification of documents

Day 1: 15th June 2022

Start End Event Chair
10:00 10:10 Welcome Dr Bea Alex
Opening by Prof Dave Robertson, Head of College of Science & Engineering, University of Edinburgh
10:10 10:50 Accepted Paper Session 1 Dr Anoop Shah
10:10-10:30 | A Quantitative and Qualitative Analysis of Suicide Ideation Detection using Deep Learning.
Siqu Long, Rina Carines Cabral, Josiah Poon and Soyeon Han

10:30-10:50 | AI-driven dermatology triage: deep learning and knowledge guided approaches.
Minhong Wang and Honghan Wu
10:50 11:00 Break Dr Anoop Shah
11:00 11:50 PhD Forum Session 1 Prof Paul Rayson
11:00-11:25 | Dynamic biomedical corpus generation and machine annotation a case study of autism spectrum disorders.
Antoine Lain and Ian Simpson

11:25-11:50 | Adverse Childhood Experiences Identification from Clinical Notes with Ontologies and NLP.
Jinge Wu, Rowena Smith and Honghan Wu
11:50 12:00 Break Dr Anoop Shah
12:00 13:00 Panel session 1: How does PPIE add value in text analytics research? Ms Natalie Fitzpatrick
Click to expand details... Chair: Prof William Dixon, Director, Centre for Epidemiology Versus Arthritis, University of Manchester

In this special panel discussion, four different stakeholders (a public involvement expert, a patient by experience expert, a NLP researcher and a representative from an NHS funding body) will discuss how patient and public involvement adds value to research using health care text analytics and natural language processing. Members of the audience will then be invited to discuss the issues presented and ask questions of the expert panel.

Panel membership:
Sinduja Manohar, HDR UK
Colin Wilkinson, HDR UK and Versus Arthritis
Prof Rob Stewart, King’s College London
Jim Elliott, NHS Health Research Authority.

13:00 13:15 Open forum Dr Honghan Wu
Pitches to the community (3-5 minutes each)
13:15 14:00 LUNCH Dr Honghan Wu
14:00 14:45 Keynote Talk Dr Bea Alex
Speaker: Prof James Teo
Abstract: In this talk, Prof James Teo will outline some real-world use of text analytics in healthcare settings at Kings College Hospital and Guys & St Thomas Hospital. This will cover not just research use-cases, but use in operational aspects of the hospital including safety alerting, clinical audits, dashboards for population health and covid-situational-awareness.
14:45 15:00 Break Dr Bea Alex
15:00 15:50 Poster and demo session 1 Dr Anoop Shah
Click to expand details... Sample Size in Healthcare Natural Language Processing.
Jaya Chaturvedi, Diana Shamsutdinova, Sumithra Velupillai, Daniel Stahl, Robert Stewart and Angus Roberts

Using MedCAT for cohort identification with patient-level contexts
Jack Wu, Zeljko Kraljevic, Richard Dobson and Daniel Bean

A systematic review deep learning approaches to automatic radiology report generation.
Yuxiang Liao, Hantao Liu and Irena Spasic

Medical Named Entity Recognition for French Language.
Fouad Aouinti, Stefan Bornhofen and Aymeric Histace

A Hybrid Approach for the Extraction of Diagnoses from Out-patient Letters.
Judith Jeyafreeda Andrew, Warren Del-Pinto, Goran Nenadic, Alfredo Madrid García, Lifeng Han,
Caitlin Bullen, Ghada Alfattni, Meghna Jani and William G Dixon.

iPOF: Improving Peer Online Forums.
Fiona Lobban, Paul Rayson and Matthew Coole

Demonstration of NLP in a patient-centric viewer to aid clinical decision support for acute stroke
Hannah I. Watson, Cameron Brown, Keith W. Muir and Alexander Weir
15:50 16:50 Panel (vet) PJ Noble
16:50 17:00 Closing marks for day 1 Dr Bea Alex

Day 2: 16th June 2022

Start End Event Chair
10:00 10:10 Welcome Dr Honghan Wu
Recap of day 1 and Introduction to day 2
10:10 10:50 Accepted Paper Session 2 Dr Ian Simpson
10:10-10:30 | Using free text electronic health records from Sussex mental health services to implement a risk calculator to identify people at risk of psychosis.
Elizabeth Ford, James Stone, Benjamin Fell, Gergely Bartl, Dominic Oliver, Paolo Fusar-Poli and Kathryn Greenwood

10:30-10:50 | Automated Clinical Coding: What, Why, and Where We Are?
Hang Dong, Matúš Falis, William Whiteley, Beatrice Alex, Shaoxiong Ji, Jiaoyan Chen and Honghan Wu
10:50 11:00 Break Dr Ian Simpson
11:00 11:50 PhD Forum Session 2 Dr Arlene Casey
11:00-11:25 | Multi-Label Extraction from Radiology Reports in Practice: Predictions on a Large Uncurated Corpus.
Patrick Schrempf, William Clackett, Antanas Kascenas, Hannah I. Watson, David Harris-Birtill and Alison Q. O'Neil

11:25-11:50 | EyeBERT: a Large Language Model for Clinical Narratives in Ophthalmology.
Quang Nguyen, Honghan Wu and Nikolas Pontikos
11:50 12:00 Break Dr Arlene Casey
12:00 13:00 Panel session 2: How can NLP enable personalised medicine? Dr Alison O'Neil
&
Dr William Clackett
Click to expand details... Chair: Alison O’Neil and William Clackett

Topic discussion: How can NLP enable personalised medicine?

Description: With the expansion of electronic health records in inpatient care, a considerable amount of unstructured text data has become available for researchers to leverage through advances in natural language processing. While analysis of structured clinical data such as physical observations, patient demographics and lab tests is commonly used to tailor clinical decision making alongside other data types such as genetics and images, automated analysis of unstructured text data in delivering precision medicine remains comparatively novel. This topic aims to explore the perspective of healthcare technology providers towards natural language processing and its role in facilitating personalised medicine.
13:00 13:15 Open forum Dr Bea Alex
Pitches to the community (3-5 minutes each)
13:15 14:00 LUNCH Dr Bea Alex
14:00 14:45 Keynote Talk Dr Honghan Wu
Speaker: Prof Ozlem Uzuner
Abstract: National NLP Clinical Challenges (n2c2) have been sharing de-identified clinical notes with the research community since 2006. In this time, many NLP tasks have been addressed and benchmarked against n2c2 corpora. Ozlem Uzuner will provide an outline of the n2c2 shared task data and share example solutions generated for addressing the research questions encompassed by these data.
14:45 15:00 Break Dr Honghan Wu
15:00 15:50 Poster and demo session 2 Dr Anoop Shah
Click to expand details... Diagnosis certainty and progression: a natural language processing approach to enable characterisation of the evolution of diagnoses in clinical notes.
Alfredo Madrid, Caitlin Bullen, Lifeng Han, Judith Jeyafreeda Andrew, Warren Del-Pinto, Ghada Alfattni, Oswaldo Solarte Pabón, Ernestina Menasalvas Ruiz, Luis Rodriguez Rodriguez, Meghna Jani, Goran Nenadic and William Dixon

Developing annotated corpora for training and evaluating biomedical text mining algorithms.
Meiqi Wang, Thomas Rowlands, Nazanin Faghih Mirzaei, Joram Posma and Tim Beck

Natural Language Processing in Barrett’s Oesophagus.
Xinyue Zhang, Angus Roberts and Sebastian Zeki

A Transformer-based Machine Learning Framework using Conditional Random Fields as Decoder for Clinical Text Mining.
Lifeng Han, Valerio Antonini, Ghada Alfattni, Alfredo Madrid, Warren Del-Pinto, Judith Andrew, William Dixon, Meghna Jani and Goran Nenadic

From Unstructured to Structured Triage Complaints: Predicting ED Disposition and Severity Level.
Haya Elayan and Owen Johnson

Predictions for COVID-19 Transmission Trend with Probabilistic FastText model.
Siyue Song, Tianhua Chen, Sean Knox and Grigoris Antoniou

[Demo] DeepCognito DEID Platform: A Scalable Architecture for Trusted Research Environments.
Azad Dehghan
15:50 16:00 Closing marks Drs Honghan Wu & Bea Alex