AI is now finding its place in medical settings, where there are obviously both opportunities and risks. In this session, the focus was on real life. Sponsored by Italian AI company, Almawave, this session examined the realities affecting the use of AI. Three presentations were made: one by the Almawave company itself; another by a healthcare provider; and a third by a recently-ended European project. A number of potentially usable definitions of AI were offered at the start of the session as well as some pointers to highly readable contemporary literature on AI.
Guido Panfili of Almawave (Italy) outlined his company’s ‘one to many’ approach that enables single clinicians to offer services to increased numbers of patients. Almawave’s focus is on data: how to obtain structured data that can feed AI applications, and how to exploit the data appropriately. Rachelle Kaye of Assuta Medical Centers (Israel), speaking on behalf of her colleague, Michal Guindy, showed how the healthcare provider is working with many companies, using AI, to explore the useful application of its large volumes of data, obtained via its 100,000+ surgical interventions a year. Panagiotis Bamidis (Greece) of the Captain project reflected on how to attract and incentivise data collection when working with people, and offered some pointers towards what can enable initiatives to succeed and what to avoid. Javier Sedano of ITCL (Spain), the project coordinator of HoSmartAI, a European project that started its work in 2021, commented briefly on how the initiative will use AI in the hospital context.
✅ ePoll: Through an EHTEL ePoll, the audience members offered their opinions, on how quickly they see themselves able to use real-life data in real-life settings hooked up with AI. Their guesstimates were spread more or less equally from the year 2021 onwards, although around a quarter thought such use would only occur five years’ hence. Time before implementation and use of innovations is one concrete aspect of the AI reality wall that all people who are working and reliant on the digital health field are now facing. In this context, Crawfordwork’s, John Crawford, reinforced Assuta’s Rachelle Kaye’s desire to see work on AI focus on meaningful and actionable actions.
🗣️ Discussions: Challenges posed by data underpinned many of the ensuing discussions. Keen to ask questions and discuss AI in general, the attendees nevertheless debated many data-specific issues. The issues covered included data quality; different types of data; the need for simple and ‘explainable’ data; data sharing models; the accuracy of algorithms; and the validation of indices/indicators, especially single indicators. Organisationally, attendees’ discussions ranged over barriers to the adoption of AI; post-market surveillance and AI; professional and organisational practices when delivering healthcare using AI; appropriate design approaches that can enable the personalisation of user preferences; the overall sustainability of initiatives; and, of especial importance, ensuring the duty of care when delivering healthcare using AI.