Digital twins: a new paradigm in oncology in the era of big data

Recent advancements in health care digitalization opened the collection and availability of big data, whose analysis requires artificial intelligence-based technologies to facilitate the development of predictive tools supporting decision making in clinical practice. In this context, the idea of constructing 'digital worlds' to evaluate the performance of such novel tools becomes more attractive. Digital twins (DTs) are 'digital objects' characterized by a bi-directional interaction with their 'real-world counterparts'. DTs aim to enhance predictions further by leveraging both the predictive capabilities of digital simulations and the continuous updating of real-life data-ideally incorporating clinical records, multiomics data, and patient-reported outcomes. DTs can potentially integrate these diverse data into virtual models applicable across pre-clinical to clinical studies. Running simulations in silico on cancer cells or cancer patients' DTs can provide valuable insights into cancer biology, clinical practice, and health care education, with the added value of reducing costs and overcoming many common limitations of current studies (limited number of variables, challenges in recruiting patients with rare tumors, lack of real-life feedback). Despite their significant potential, DTs are still in their infancy, facing numerous unsolved technical and ethical challenges that hinder their application in clinical practice.