Modern medicine has broadened its approach to cancer care and now exploits the use of Artificial Intelligence (AI) and its applications in health imaging and oncology. The AI for Health Imaging (AI4HI), a network comprised of six EU-funded research projects, targets the construction of robust AI infrastructures to advance cancer care. AI4HI involves clinicians, legal and technical researchers. The application of AI within practical medicine gives rise to ethical concerns as it can exacerbate systematic biases and health inequities. This study focused on assessing researchers’ awareness around AI-driven biases and their impact on health equity.

A qualitative research study was conducted among 24 stakeholders from 6 countries to explore their perceptions on AI-driven biases and measures to overcome them. Participants were recruited from the AI4HI network group through purposive sampling. The developed interview guide was based on an existing framework for the ethical development and application of AI/ML. Descriptive statistics were performed using SPSS. The qualitative data was analysed using NVivo 12 software, and thematic analysis was performed. Ethical approval was obtained for this study.

Most of the participants were female (58.3%) and from Greece (41.7%). A lack of shared understanding was noted among the professionals on the definition of bias in AI. Limited diverse patient representation was identified as a main source of bias. A multidisciplinary approach and the involvement of legal experts to navigate through data protection laws and ensure broader data inclusion, was also highlighted.

The findings call for an urgent need to align researchers’ understanding of the concept of AI-driven Bias. Additionally, there is a need to proactively design and implement mitigation measures to limit various forms of bias and promote equitable AI in healthcare.