Cancer Patient Digital Twins for Predictive Oncology: The State of the Art

Since its definition by NASA in 2010, the digital twin idea has found many innovative applications in the fields of aerospace engineering, manufacturing, construction, and automotive. Recently, the concept of a digital twin has been proposed as a more data-driven approach to healthcare in the simulation, integration, testing, and monitoring of personal health and well-being. However, the research on cancer patient digital twins is still in its infancy, with four predominant challenges. First, the human body is a complex multiscale dynamic system where many biological processes and underlying mechanisms are still unknown, and many are inferred. Secondly, the essential data (e.g., multimodal data, clinical data, experimental data, and patient-centered data) are scattered and disconnected across various systems and domains, and at levels not amenable to AI/ML analysis at scale. Thirdly, improvements are needed to address bias in data collection, algorithm development, and system training to assure long-term approaches that safeguard against health inequities, social harms, and structural injustices in the applications of AI and digital twins. Fourthly, there is a gap in evidence related to the implementation of digital twin models into clinical workflows, hence understanding the clinical utility and feasibility of digital twin models is essential.

With recent developments in data curation, multiscale modeling, and high-performance computing, there have been concurrently increasing and promising developments of AI, machine learning, and deep learning algorithms with potential applications in a cancer patient digital twin. This Research Topic is intended to present state-of-the-art science, research, and development in cancer patient digital twins in recent years. Over the long term, we would like to see digital twin research efforts transform into a national synergy among all stakeholders. Guiding a future vision, efforts in the development of cancer patient digital twin technologies would lead to enhanced care and quality of life for millions of cancer patients worldwide.

In this themed article collection, we welcome original contributions from the scientific community in the format of research articles, perspectives, and reviews. Potential topics include, but are not limited to:

• High content, high speed, and high-quality multimodal data acquisition
• Data standards and labeling for AI/ML
• Real-world data integration, curation, and management
• Multi-omics data and biomarker discovery in cancer treatments
• Mathematical, statistical, and mechanistic modeling of organs and systems
• Modeling and simulations of time-series data
• Tumor microenvironment modeling and simulations
• Ethical and trustworthy AI
• Responsible and explainable AI for cancer care
• Security and privacy in clinical AI
• Natural language processing of electronic health records and clinical notes
• Treatment response simulation and prediction
• Applications of digital twin technologies in treatment development and clinical trials
• Deployment strategies of digital twin technologies in healthcare
• AI-assisted clinical decision support
• Mobile health for patient monitoring and intervention
• AI and bioinformatics for improved quality of care
• AI-aided diagnosis and early detection of cancers
• Pharmacokinetics-Pharmacodynamics model of drug-tumor interactions
• Physics-informed machine learning
• Human-in-the-loop AI
• Knowledge representation and extraction