In prostate cancer, AI has been widely utilised such as fusion of imaging modalities and cancer detection prediction. Healthcare is increasingly becoming data intensive, and AI has a potential role in recognising and extracting information from clinical data. Most clinicians without AI, data science and software development backgrounds will need to work with AI/data scientists to form an AI model. Our team consisting of urology clinicians without such backgrounds and AI scientists recently worked together to build an AI model for prostate cancer. We reflect upon the challenges and benefits in working in an interdisciplinary team.

A literature review was performed utilising keywords “AI” “artificial intelligence” “interdisciplinary team” and “medicine”. Reflections and learning points were collated from the research team.

There was limited published research on this topic. One publication reported low visibility of interdisciplinary collaboration between medical clinicians and AI scientists. Another study reported the benefits of collaborative research allowed teams to extend research into a variety of areas, with challenges identified as scheduling conflicts and enhancing project management. We found the main challenge of working with diverse backgrounds was the “language barrier” for the exchange of technical knowledge and concepts of each member’s discipline. Sometimes this knowledge divide was quite significant with clinicians having limited understanding of AI concepts and technicalities involved in handling datasets and complex machine learning algorithms. On the other hand, AI scientists had limited clinical acumen and exposure which could limit data preparation and translation of outcomes from AI models into clinical context and downstream applications. AI scientists also lack the domain knowledge to further extend their work or recognise limitations of the models they have developed. The strengths of interdisciplinary teams were diverse backgrounds allowed improved contextualisation of the model into real-world practice. There were broader perspectives and better problem-solving as different perspectives provided alternate approaches to challenges. As well, clinicians were able to curate and prepare data to be more representative and therefore reduce bias. The commonality between disciplines were foundations of statistics although different analytical methods were applied in AI programming. Our recommendations to clinicians interested in AI research would include skill development which can occur through education and industry engagement. Both AI scientist and medical professionals should work together to develop a common language to bridge the gaps of technical knowledge in our respective fields.

AI is a rapidly evolving field with significant potential to advance clinical research. As medical professionals, there is large scope for involvement in AI-based research and clinical application with interdisciplinary collaboration with AI specialists. Diversity of teams in AI related to medicine may improve model adaptability, clinically relevant results which may lead to more robust innovations.