Last week the annual European Conference on Computer-Supported Cooperative Work (ECSCW) took place in Coimbra, Portugal. This annual conference focuses on practice-centred computing and the design of cooperation technologies and is committed to grounding technological development and systems design in an understanding of the specifics of practical, situated action. I always wanted to go to the CSCW or ECSCW conference, and I was looking forward to actually go this year, especially presenting a position paper in the workshop on Exploring Human-Centered AI in Healthcare: Diagnosis, Explainability, and Trust and to connect with people I met online over the last few years. Unfortunately, the pandemic got to me and I was still recovering, so I had to prioritise my health and cancel the flight. Luckily, the workshop organisers had decided from the beginning to have a hybrid meeting, so I could join nevertheless.
Together with my inspiring colleagues at RRD, Eline te Braake, Marian Hurmuz and Stephanie Jansen-Kosterink, I submitted a position paper titled “Balancing data-hungriness of AI and the workload of manual data collection”. The paper is based on what we learned from patients and healthcare professionals during the first year of the RE-SAMPLE project. In the position paper, we acknowledged that artificial intelligence systems need big data sets for the development of models or machine learning algorithms. This data has to be collected, which is not always possible to do automatically. Manual data collection in healthcare is often performed by patients or their caregivers, which adds workload to their existing disease burden. We reported on a qualitative study with Dutch patients and healthcare professionals focusing on data collection of patients living with chronic obstructive pulmonary disease. Both groups were concerned about additional workload for patients when asking them to fill in questionnaires on a daily basis. We concluded that manual data collection by patients creates a response burden that needs to be balanced against their disease burden and the limited energy patients have at their disposal. When automatic data collection via sensors is not possible, the data collection instrument needs to be designed in a way that pays attention to multiple factors that affect the response burden.
The workshop took place in the afternoon on 27th and 28th June 2022. Five position papers and two field works were presented, during which all participants could add questions on a Miro board. After each presentations, there were really interesting discussion on human-centred AI, processes to incorporate the AI systems in daily care and trust that people put in the technology. After our presentation, there were interesting discussions around the fact, that some benefits of AI technologies only transpire later (e.g., predictions based on data collected) and how to communicate this to patients. We also discussed that while AI systems do provide value that is realised more long-term, we should ensure that patients also benefit in the short term and are not ask to merely collect data. In RE-SAMPLE these short term benefits are related to coaching, self-management and peer-to-peer support that are currently being conceptualised. The workshop was very insightful and inspiring and we are looking forward to future collaborations.
