Could digital twins be the next revolution in healthcare?

The advent of the fourth industrial revolution, known as Industry 4.0, resulted in a reinvention of the concept of simulation, giving rise to what is defined as digital twins (DT) [ 1–4 ]. The continuous growth and evolution of technologies such as the Internet of Things, artificial intelligence (AI), machine learning, and big data have enabled the collection, processing, and storage of large amounts of real-time data. These technologies are essential for the creation and operation of what we currently known as DTs. Conceptually, a DT is defined as a real-time digital replica or virtual representation of an object, process, or system [ 1 , 5 ]. Its growing use across diverse sectors highlights its potential to revolutionize the way we create, manage, and operate even the most complex systems. The health sector is recognized as a complex ecosystem that requires effective and efficient operations, optimizations, management, and control to offer reliable, economical, and quality health actions and services. Your main challenge is to provide the best possible healthcare services to patients using finite healthcare resources. The use of DT for predictive analysis, process improvement, supply capacity planning, risk management, assertiveness in diagnosis, and increased clinical safety, among others, will allow managerial and clinical decisions to improve the quality of health care offered in the future [ 6–12 ]. This work aims to compile the state of the art in the use of DT in healthcare, through a review of the scientific literature on the subject, thus making it possible to evaluate the trends and challenges of this technology for the coming years in the healthcare sector. This work expects to explore the various applications of DT in healthcare, their potential for further development, and the challenges they present. By doing so, we hope to provide valuable insights for this growing technology and its potential to revolutionize the planning, production, and management of healthcare structures and processes.

The objective of this research was to analyse scientific publications with the purpose of investigating the use of DT in healthcare, trends, and challenges in their use, providing insight into the possible paths of healthcare. This research sought to answer two guiding questions: ‘How are digital twins being used in the healthcare sector?’ and ‘What are the challenges and perspectives of using digital twins in the healthcare sector?’ PICO strategy (acronym for P: population/patients; I: intervention; C: control; O: outcome) is used to help to define what the research question should in fact specify for this paper ( Fig. 1 ). PICO strategy (Population, Intervention, Comparison, Outcome) criteria and definitions. The Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) was a great help for authors to prepare transparent accounts of their reviews, providing syntheses of the state of knowledge in a field, from which future research priorities can be identified. For this paper, a literature search was conducted in digital libraries of six academic databases: IEEE Xplore, Dimensions, Scopus, Web of Science, PubMed, and ACM. The six databases were chosen with the aim of covering the Engineering and Health research areas. IEEE Xplore and ACM serve the purpose of containing the Engineering/Technology area, whereas the PubMed database is focused on the health area. Scopus, Web of Science and Dimensions incorporate many fields of knowledge and were included to identify any other records relevant to this study that were not in the specific databases. We tried to restrict the search by using specific keywords to find publications of interest. The search strategy was based on three main terms: digital twins, digital health, and healthcare . Another filter applied in the search strategy was publications within a time range of up to 5 years (2018–23). The resulting search query is shown below ( Fig. 2 ): Search strings. At the end of applying the search strings, a total of 86 results were reported. After analysing possible duplicate references in the Zotero library, 28 duplicate results were excluded. At this point, a total of 58 publications were reached. This last material, then, went through a data survey, collection, and analysis process, divided into three stages: Pre-analysis evaluating title, abstract and keywords; Exploration of the selected material; Treatment of results and interpretations. The exclusion criteria included studies in languages other than English and articles unavailable or only partially available for download in the chosen databases. Furthermore, all articles that did not explicitly mention the use of DT in health, as well as items selected in the search, which were books or book chapters, were also excluded from the discussion. The only inclusion criterion was confluent studies that dealt with topics related to DT, digital health, and health care. The main limitations of the method employed in this work are related to the dependence on the quality of the included studies, the reliance on the availability, and accessibility of published studies, and the difficulty in addressing the heterogeneity among the studies. This issue was mitigated by dividing the studies into thematic categories. Following the pre-analysis stage, 31 results were excluded. Subsequently, during the exploration of the remaining articles, an additional 14 were excluded due to a lack of explicit focus on the application of DT in healthcare. Figure 3 illustrates the entire process. PRISMA flow diagram.

The studies included for discussion can be categorized according to the application of DT in the health sector into 2 groups: the clinical applications group, with 7 records, and the operational applications group, with 6 records. When carrying out the content analysis of the articles, we can also verify a subdivision within the groups. In the clinical applications group, we have five articles that are focused on the theme of personalized care/precision medicine, signalling the development of digital technologies based on real-time patient data for the management of specific diseases or conditions, one article addressing the reproduction of biological structures creating avatars of organs or even the human body and another that focuses on ethics issues related to using DTs in healthcare. In the operational applications group, we have a subgroup, with 5 articles, that discuss the application of DT supporting the optimization of operational processes by using real-time data integration, advanced analytics, and virtual simulations to improve patient care and another subgroup with one article that relies on the construction of virtual structures such as hospitals. The full spectrum of the articles is detailed below ( Fig. 4 ). Findings. In summary, in the sample of selected articles, the main benefits of using DT technology for healthcare predominantly included increased personalization of care, improved quality of care with the increasingly consistent use of precision medicine and gains in the operational efficiency of health facilities, equipment and services. However, some of the included studies also highlighted key challenges related to DTs in healthcare, such as interoperability, the processing of large volumes of data, patient confidentiality and data security, and listed them as the greatest obstacles at present for the large-scale implementation of this technology.

A DT can be understood as technology capable of mapping the real world to the digital world through interaction between the two in real time [ 13–16 ]. Healthcare is a field in which DT are being explored and applied. To date, most studies [ 6 , 8 , 17–26 ] have focused on virtualizing individual assets such as devices, structures, and patients, considering the point of view of a specific application, such as body parts, organs, or body systems aimed at personalizing care. Another potential for its use lies in virtualizing contexts and situations that involve several interrelated strategic assets of a healthcare organization, such as a hospital [ 27–29 ]. According to the World Health Organization (WHO), digital health encompasses the field of knowledge and practice related to the development and use of digital technologies to support health [ 30–32 ]. The ability to collect, analyse and share data effectively is enabling unprecedented personalization in healthcare, improving the assertiveness of treatments, reducing costs, and putting patients in control of their own health. However, for this to become a reality, it is necessary to combine efforts in initiatives that boost the analytical intelligence of healthcare organizations to enable the integration, mining, and interoperability of data [ 7 , 23 , 25 , 32–34 ]. Although digital health has extraordinary potential, it faces significant challenges and risks. Despite enabling increased access to healthcare, it can also be perceived as a ‘form of barrier’, where people who benefit most often face more difficulties in access due to a lack of resources or digital skills [ 35–38 ]. Established as a global priority in 2005 by the WHO [ 39 ], the digital transition has significantly impacted the health sector by creating the conditions for redefining the care model to make it more integrated, participatory, and personalized. Currently, the delivery of healthcare services continues to occur in a fragmented and timely manner for most patients. It becomes extremely difficult for healthcare professionals and patients to monitor all important events, correlations and exposures that negatively affect the provision of healthcare, both in terms of cost and quality. In this context, DT offer remarkable potential to significantly improve healthcare delivery [ 10 ]. Considering the results found in the selected literature, 4 main axes of the application of DT in healthcare are presented: The use of DT in the virtual representation of biological structures has enabled a new approach in medical education and clinical practice. The evolution of imaging technologies, such as magnetic resonance imaging and computed tomography, has allowed the creation of extremely precise 3D models of organs and tissues. These models provide an interactive representation that not only increases the understanding of anatomy but also improves knowledge retention for medical students. The ability to ‘navigate’ organs and systems in a virtual environment has a lasting impact on the training of future doctors, providing an immersive learning experience that can be difficult to achieve with traditional methods. The creation of a fully developed DT of a human is still a goal for the future. However, several companies and research institutes are exploring the development of digital replicas of body parts or physiological systems for specific purposes. For example, in the field of orthopaedics, these DT can be used to create virtual models of a patient's musculoskeletal system, including skin, bones, joints, and muscles. These DT can be employed to simulate surgical procedures, test various implants prior to surgery, and predict how a patient will respond to treatment. Building a DT for this purpose demands access to extensive healthcare data repositories, which include genetic information, medical records, imaging, histopathology, and other pertinent sources. These data must be of high quality, accurate, and comprehensive. For Viceconti et al. [ 25 ], the discussion on the creation of the Virtual Human needs technical, political, and social considerations, such as the challenge of providing fair and transparent use of data. Perhaps the greatest change is in the path that health treatments will follow, moving from being organized by a standard to being based on the genetic, phenotypic, structural, physical, and psychosocial characteristics of the individual, being referred to as precision medicine or even broader as personalized care. Essentially, patients are treated as individuals and not according to some norm or standard of care (providing the right treatment, at the right time to the right person). Digital twins are software-based replicas that simulate the dynamic functions and potential failures of engineered products and processes. In healthcare, patient-specific DT integrate established knowledge of human physiology and immunology with real-time, patient-specific clinical data. This integration enables predictive computer simulations across various types of diseases and conditions. These medical DT could prove invaluable in healthcare management, leveraging mechanistic insights, observational data, medical histories, and the capabilities of AI to optimize treatment strategies and improve patient outcomes. In the context of personal DT, prevention emerges as a pivotal focus, accompanied by the empowerment of patients. Central to this approach is ensuring that patients comprehend their health status and take responsibility for it. Currently, the strategy involves comparing patients with a comprehensive database to identify similarities and differences. However, the aim is to advance beyond this by gathering historical health data from earlier stages when the patient was healthy. These data can then be utilized to assess current health status, estimate risks, and devise a tailored prevention plan. This approach empowers patients to adjust their behaviours proactively and detect potential diseases early, leveraging straightforward information that can be extracted without direct patient involvement, according to the work of Schwartz et al. [ 4 ]. The application of DT to improve healthcare processes is driving significant advances in several areas. In oncology, e.g. DT play a fundamental role in modelling tumours and simulating treatments. This approach allows physicians and oncologists to adjust treatment strategies based on the progression of a patient's disease, resulting in more personalized and less invasive interventions. Furthermore, continuous monitoring of tumours in DT enables a more agile response to changes in the clinical picture, providing better results for patients. At the work of Wickramasinghe et al. [ 21 ], DTs can support personalized treatment for uterine cancer. Similarly, in women's health, DT are used to monitor gynaecological conditions such as endometriosis and fibroids and to manage pregnancy more effectively. The ability to simulate scenarios and monitor the progress of pregnancy in a virtual environment offers patients more personalized and safer care. This also makes it possible to identify complications early, which is essential to ensure maternal and foetal health. In the area of geriatric care, DT enable the creation of care plans adapted to the complex needs of older adults. With the aging of the population, this personalization becomes crucial for guaranteeing the quality of life of elderly people. Digital twins can model a patient's health conditions, considering comorbidities and risk factors associated with advanced age. This results in more comprehensive and effective care that improves seniors' quality of life and reduces unplanned hospital admissions. Rivera et al. [ 6 ] explained that the use of DTs to support precision medicine techniques in the context of continuous monitoring and personalized data-driven medical treatments could facilitate the management of chronical conditions. Digital twins can also improve the management of trauma and fractures, especially in the elderly population as explained by Ricci et al. [ 33 ] as well as in mental health, when used in association with virtual coaching solutions according to Gabrielli et al. [ 20 ]. A crucial point in the use of DT in personal care concerns the ethical and moral risks associated with use without any type of reflection. This has been well explored by Huang et al. [ 37 ]. Operational efficiency in healthcare is a constant concern, and DT play a crucial role in optimizing healthcare systems. One of the main benefits is the improvement in resource allocation in hospitals and other healthcare units. DT allow a detailed analysis of the use of beds, personnel, and equipment, enabling more efficient and effective planning. This not only reduces patient waiting times but also improves assertiveness in the provision of care, resulting in a more successful patient journey and a more positive experience for everyone involved in the care process. This topic was addressed both in the works of Michelfeit [ 19 ] and Cheng et al. [ 29 ]. Taking a hospital as an example, by using historical and real-time data from operations, as well as information from the surrounding environment, such as cases of notifiable diseases and traffic accidents, the DT allows the hospital unit manager to make informed decisions, identify a lack of beds, optimize team schedules, and manage room occupancy more effectively. This approach not only increases resource efficiency but also improves hospital and staff performance while reducing costs. In the works of Sun et al. [ 8 ], Vallée [ 23 ], Chaudhari et al. [ 26 ], and Rajanikanth et al. [ 28 ], DT were seen as a potentially important technology in improving the patient's journey by seeking to streamline operational and care processes. By building a DT of the patient's journey, the healthcare unit can predict patient activity and plan operational capacity according to demand, resulting in significant improvement in services delivered to the patient, in addition to increased safety, increased volume and a better patient experience. Modelling and forecasting the demand for healthcare services are also areas where DT offer great potential. By analysing real-time data and simulating scenarios, healthcare systems can make more informed decisions about resource distribution and staffing, ensuring that care is available where and when it is needed. These capabilities are critical for improving operational efficiency and ensuring that the healthcare system is prepared for future challenges. In relation to drug development, DT enable detailed simulation of molecules, pharmacological interactions, and virtual clinical trials. Throughout the drug development process, extensive amounts of data are generated and managed, which DTs’ harness to create models. Consequently, DTs can expedite clinical trials in drug research, enabling faster trials with reduced patient enrolment. The increasing use of personal health monitoring devices, such as mobile apps and integrated sensors, enables active surveillance of the user's key health parameters, such as electrocardiogram (ECG), blood pressure, heart rate, and glucose level, minimizing potential inaccuracies in data recording. Such devices can collect and transmit information anonymously to the cloud, where it can be compared with disease symptom histories or even alert competent healthcare professionals when necessary. As previously stated, DTs offer two key benefits: real-time health information and predictions, and interactive capabilities that allow communication between DTs, decision-makers, and, depending on the model, the individuals themselves. To address the ethical concerns surrounding DTs, it is important to clarify the conditions and requirements for using them to simulate a person's health. At first, each DT must correspond to a real person and be assigned a unique identifier. This identifier will allow logging into one's virtual representation, setting health parameters to monitor, and controlling who can view this information. Ensuring the DT is accurately created, and the data validated is crucial, with an authorized institution overseeing this process. To protect privacy, technologies like passwords, biometrics, and blockchain encryption should be used to prevent data tampering. Continuous data exchange between sensors, wearables, and the DT is required to ensure the DT accurately reflects changes in the person's health status. Future solutions, like blockchain technologies, might be necessary to maintain accuracy and reliability even if data flow is disrupted. It is important to specify the scope and purpose of the DTs, including the types of data they will handle, such as text (diagnosis records), numbers (weight, blood pressure), or images (ECGs), and how this data will be processed. Clearly defining the interfaces for data input and transfer is crucial. Additionally, it must be determined what kind of feedback the DT should provide. This feedback could be directed to the person being simulated, their doctor, or a designated decision-maker. The feedback, which might include predictions, suggestions, alerts, or treatment plans, should help the person and their healthcare providers make informed decisions about their health, such as improving lifestyle habits or seeking further treatment. Privacy and security must be prioritized, especially in decision-making processes, to balance detailed health information with privacy concerns. Many of these technical requirements remain unresolved, especially concerning the processing and modelling of large datasets. Finally, DT have the potential to offer significant societal benefits, such as enhancing precision in public health interventions. However, their application in personalized medicine may not be equally accessible to all individuals or communities, which could exacerbate existing disparities. Additionally, identifying patterns within a population of DTs might lead to problematic segmentation and discrimination. Therefore, it is essential to establish governance mechanisms that protect individual rights, ensure the privacy and security of personal biological data, and promote transparency and fairness in both the use of data and the distribution of its benefits at both individual and societal levels.

The use of DT mainly in process optimization and healthcare lines presents important challenges related to data integration, privacy, and interoperability. However, trends indicate great potential in personalizing treatment, prevention, remote monitoring, informed decision-making, and process management, which can result in significant improvements in quality and efficiency in healthcare. This work could, in some way, contribute to expanding discussions on the topic, opening space for new reflections. It is worth noting that, despite the sector's rising costs, increased demand for services and the already established role of primary health care (PHC) in improving the health levels of a population, there were no studies in the selected literature, on the specific use of DT. As it is the main axis of changes in health systems to improve health levels, PHC can receive due attention from the scientific community and the industrial complex as a promising field for the use of twin digital technology in healthcare and in the operation of its units. More in-depth future studies should be carried out to explore the possible consolidation of the use of DTs in health, especially in processes linked to health care and PHC, or even clarify which initiatives should be implemented or even strengthened to sustain the progress achieved thus far.