Decision making in cancer treatment increasingly relies on AI

AI systems provide crucial support for personalized cancer care but are only as helpful as their interaction with physicians allows. In a recent study conducted at two hospitals involving 81 lung and liver cancer evaluations, AI helped with dose adjustments in radiotherapy. Analysis showed that trust in AI recommendations plays a significant role; 76% of lung cancer decisions were adjusted by AI to optimize both tumor control and tissue damage. However, doubt about helpful dose increases persists, especially without convincing clinical evidence. Criticism from physicians highlights the need to place AI recommendations in a clinical context. Although AI has improved consistency among physicians, human judgment remains indispensable.

These findings suggest that AI is transforming the dynamics of oncology care, but the physician's role remains crucial for successful implementation.

The impact of AI on clinical decisions

AI support in cancer treatment leads to significant adjustments in clinical decision-making. In non-small cell lung cancer (NSCLC), 57% of treatment decisions were modified after AI consultation, compared with 47% in liver cancer (HCC). Remarkably, 88% of all evaluations included at least two adjustments after AI input. Confidence in AI recommendations varies widely among physicians, with adjustment rates ranging from 0% to 100% for NSCLC and 22% to 67% for HCC. In the Netherlands, we see similar developments, with hospitals such as Antoni van Leeuwenhoek actively implementing AI in their oncology care pathways. This integration represents a crucial step in the modernization of cancer care.

Treatment optimization through AI assistance

The impact of AI on treatment outcomes is significant. UMC Utrecht has launched four new research projects to further investigate the effectiveness of AI in cancer treatment. One significant development is that AI has accelerated the diagnosis of brain tumors during surgery from a week to just 1.5 hours, significantly improving the efficiency of treatments.

The study shows a clear correlation between physicians' confidence in AI and their willingness to follow AI recommendations. For HCC, a strong correlation was found between trust and acceptance of AI advice. Dutch experts, such as Jonas Teuwen of the Antoni van Leeuwenhoek, emphasize that AI will affect every aspect of care. However, as Wouter Kroese of Pacmed warns, implementing AI in healthcare is sometimes thought of too lightly. This underscores the importance of careful integration, where the physician's expertise remains central.

Dutch innovations

Dutch healthcare institutions are taking the lead in AI innovation for oncology. TNO has developed a solution for secure data analysis where privacy-sensitive information remains protected. The Amsterdam UMC is using AI to predict treatment outcomes and side effects. IKNL is working on AI models for the treatment of various cancer types. These developments join the international trend in which AI systems are becoming increasingly sophisticated in supporting clinical decision-making. Azam Nurmohamed of Amsterdam UMC sums up the development for the moment, “AI will not replace the employee, but will become a co-pilot in healthcare.”