On 18th and 19th April 2018, the first UK health text analytics conference (HealTAC) took place in Manchester. The main conference venue was Pendulum hotel, located on Stackville street, close to the north University of Manchester campus, the former UMIST. I had a pleasure to participate and help with a certain organisational things, as a member of local organisation committee member.
On the first day of the conference, people started arriving between 8:30-9:00 for registration. During the registration people could have some coffee and pastry for breakfast. People having posters were directed to the poster room, so they can already hang their posters. The conference started at 9:00, with welcome speech, some health and safety procedures, which was followed by a PhD forum. In total 5 PhD students presented their ongoing research in health text analytics, raging from analysis and detecting of dementia under-diagnosis, integrating structured and unstructured information sources, to sentiment analysis of patient comments and mining adverse drug reactions from social media. Each PhD presentation was followed by the questions from the panel, whose members were Prof Wendy Chapman, Prof Robert Stewart, Dr Angus Roberts, and Prof Pierre Zweigenbaum.
After the PhD forum, Prof Wendy Chapman gave an interesting talk about navigating career. Her career took a number of turns, from being linguist, moving into medical informatics, finally ending in medical text analytics and contributing with one of the most used negation detection systems – NegEx. She was comparing career with skying which was interesting. Also, concluded that there are many variables contributing to successful career, including luck, mentors, people who push you, network, and so on.
The conference continued after quite good lunch with the talk by Pierre Zweigenbaum about challenges in organizing and evaluating text mining shared task for coding death certificated in French and English. He explained a structure of a death certificates, how they differ in countries and what makes them challenging for coding.
After his talk, we were supposed to have a poster and demo session. However, about 10 minutes into the session the fire alarm sounded. While we were leaving, there was already police and fire-man in front of the building, which suggested that it was manually triggered. It turned out that there was some chemical leakage at the Manchester Institute for Biotechnology and that the venue we were in is connected with the building. For a bit we waited outside, with quite nice weather (it was 25C in Manchester), and after a bit we got a news that the conference will continue in Raynolds building. I helped collecting people who were on the conference from the mass of students and other evacuated people from other buildings and redirecting them to the new location.
Conference continued shortly in a new location with a panel about mining veterinary health records and text analytics for veterinary. Mercedes Arguello Casteleiro gave very engaging talk and very good motivating examples of why mining veterinary documents are important. Basically, there are less restrictions on obtaining animal health records than with humans, therefore big data methodologies can be tried and tested. This may, on the other hand, if successful open opportunities for human health records limitations to be lifted.
The final panel of the day was industry panel, composed of: Angus Roberts (King’s College London), Simon Thompson (CTO, SAIL Databank), Joseph Connor (Code4Health / NHS Digital), Ed Doshoki (IMO), Luca Toldo (SAP SE). Some interesting topics on engagement between academia and industry was raised.
After this, during the coffee, we got the news that the Pendulum hotel is again open, and people staying there and people who left their stuff could collect them. Almost 20 people booked the room in that hotel, therefore the situation with the chemical leakage made a bit of a problem, however, conference organizers were looking into ways to accommodate people in other places.
The second day was a bit less adventurous. It started with paper presentations. The presentation were raging from text mining veterinary health records, identifying misspelled names of drugs, to text mining of stroke outcomes.
After these talk we had a poster session, where Maksim Belousov and myself had a poster about mining adverse drug reactions from drug labels. During this session also couple of commercial demos were shown, such as Lexalytics , DeepCognito, and GATE Cloud.
On the second day Wendy Chapman gave her second talk on whether NLP can ever become out of the box. She took a case study of developing a NLP system and explained all the specificities and issues one system has to have. Also, I loved a part when she said “Can you really rely on us to get rid of us?”. Basically, will we go in automation so far that we will automate even the jobs we are doing. My guess is not really.
The conference continued with one more poster session and a panel about gaining public trust for healthcare text analytics. Public trust is quite a huge issue, because in healthcare, the data can only be obtained if the patients and hospitals trust researchers doing text analytics research. This is sensitive topic from many angles, and therefore the views on this topic was provided by people coming from different areas, such as health informatics, health text analytics, health care, philosophy, psychology, etc. (Elizabeth Ford, Lamiece Hassan, Malcolm Oswald, Jessica Stockdale, Benjamin Brown, Grace-anne McCorkle).
All in all, I have to say that this conference was a great event, bringing together many people from healthcare text analytics space. I have met people coming from Australia, Canada, USA, Europe and Asia to present their work. Therefore, it was really an international conference, even though the name may be a bit misleading.