Symposium synopsis
On Friday the 22nd of March, the Cypriot research and innovation community spend an afternoon with a selection of esteemed speakers from Cyprus, Europe and the US discussing the most recent advances in artificial intelligence for oncology diagnosis and clinical practice. The event was hosted by the German Oncology Center (GOC) at the premises of the Cyprus University of Technology (CUT). Researchers and oncologists in the field of AI discussed the challenges for developing a tool that is both accurate and acceptable for use into clinical practice. Prof Nicolas Tsapatsoulis and Prof Andreas Anayiotos from CUT, Dr Constantinos Zamboglou, the Vice Medical Director of the GOC and Ms Georgina Kyriakoudes the CEO of DCentric health welcomed the guests by introducing Agora3.0, a new research project funded by the Cypriot Research & Innovation Foundation (RIF).
During the first part of the symposium, attendees heard a presentation from Dr Alexander Thieme from Stanford University on the development of apps and the rising need for artificial intelligence based processing of surmounting data for therapeutic purposes. Prof Andreas Charalambous from CUT spoke about the formation of the European Office on AI and how the landscape in EU is changing to include regulations on the use of AI into the clinic. Another presentation was followed by Dr Jan Peeken from the Technical University of Munich who spoke about the training of AI models for tumour diagnosis and the challenges in the developing algorithms.
For the second part of the symposium Dr Paul Doolan and Dr Yiannis Roussakis from the GOC hosted a panel discussion on the pressing psychological and ethical dilemmas in using AI in oncological clinical practice.
Mr Nikolas Philippou (PASYKAF) kicked off the discussion on patient consent concerning the use of patient data for the training of models towards developing algorithms. The importance of including patients at this earlier stage of development can promote transparency and communication with patients about how their data are going to be used. Acceptability of an AI tool in oncology, is another multifaced aspect that Ms Andrea Kkoufou from the Anticancer Society emphasised. Ms Kkoufou provided examples where patients do not comprehend the implication of their diagnosis for treatment and follow-up, nevertheless the meaning and impact of the use of AI for the clinical management of their condition.
Prof Constantine Dovrolis from the Cyprus Institute emphasized how a well-designed AI app for the US population may not work so well in the Cypriot population due to genetic differences or other variations that introduce bias in the final prediction outcome. In particular, AI can be used in a clinical environment where it identifies clinical patients at high risk of lethal cancer – or it predicts severe treatment-related adverse effects. However, there are additional layers of ethics and consent that need to be considered, including surmounting data to train these AI models. Prof Anastasia Constantinidou from the University of Cyprus, noted that there is unpredictability of using such an AI tool and that is also part of the process. While AI in oncology is still at its infancy including patients and their data during this process of developing and adopting models for use into the clinic is the way forward.
The discussion ended with further aspects that researchers and oncologists in the field of AI need to consider for developing a tool that is both accurate and acceptable for use into clinical practice. Mr Theodoros Loukaidis from RIF concluded the event by congratulating the speakers and panellists for introducing such a multidisciplinary approach to tackle the future challenges of AI in modern oncology.
View more on the event programme
