Artificial intelligence in the oncology workflow: Applications, limitations, and future perspectives

With recent breakthroughs in the digitalization of medical records, enlargement in depth and variety of databases, novel computational technologies, and algorithmic advances in discerning complex and nonlinear relationships in multimodal data, artificial intelligence (AI) has the potential to significantly improve the current oncological clinical practice. An overview of the avant-garde applications and current status of AI methods in the oncology field is proposed from a workflow perspective. Stages from patient assessment to treatment delivery and follow-up care are surveyed in terms of opportunities and efficiency of AI solutions. A major focus is dedicated to radiation therapy (RT), whose growing complexity is associated with increasingly labor-intensive tasks and consequent variability in care quality. AI is presented as a powerful tool to improve performance standardization and to reduce interobserver variation in a time-sparing approach. Moreover, the main challenges in AI platform development, clinical implementation, and maintenance, together with currently adopted measures, are discussed. Finally, our perspective on potential solutions to ensure the delivery of equitable and unbiased care is provided.