When I read this article, I thought, “wow, I don’t understand a word of this, AND I think it may be really important”. Sooo, I asked Jorge Galvez, the Kugler Vonderfecht Professor, Vice Chair of Pediatric Anesthesiology, University of Nebraska Medical Center, and the chair of the Society for Pediatric Anesthesia’s Biomedical Informatics Special Interest Group to review this and to become the PAAD’s primary IT editor/reviewer. He chose Drs. Nelson and Simpao to assist him in today’s PAAD. Myron Yaster MD
The evolution of information science and the capacity to store, process and transfer data have advanced at a remarkable pace over the last few decades. Moore’s Law was introduced to the world of computer science in 1965, stating that ‘the number of transistors on a microchip doubles every two years, though the cost of computers is halved.’(1) Based on this law, we can expect the speed and capability of computers to increase every couple of years and we will (hopefully) see decreasing costs.
Putting this law into clinical context, anesthesia care has evolved drastically since the introduction of smartphones 15 years ago, marked by the introduction of the first iPhone in 2007. Anesthesiologists and anesthetists relied on landline phones and pagers for communication and documented anesthetics on paper anesthesia records. It was rare to see a computer near an anesthesia machine, let alone a computer workstation incorporated into the anesthesia workspace!
In present day, anesthesiologists routinely use smartphones to review a patient’s medical record and view intraoperative vital signs in real time. Today, we have an opportunity to explore the ever-advancing realm of technological possibilities in an established concept applied in a novel way: the digital twin.
Original Article:
Lonsdale H, Gray GM, Ahumada LM, Yates HM, Varughese A, Rehman MA. The Perioperative Human Digital Twin. Anesthesia & Analgesia. 2022 Apr 1;134(4):885-92. PMID: 35299215
Dr. Lonsdale and colleagues present a well-established concept, the digital twin, and propose its application and relevance in perioperative care.(2) Digital twins have been used in engineering and manufacturing for decades.(3) A digital twin is essentially a virtual replica of a complex system that can be evaluated in a computer simulation environment. In the manufacturing world, digital twins can be used to design cars and then tweak and optimize the design before paying the cost of manufacturing the product. Creating a high-fidelity digital model of a complex system requires vast amounts of accurate data about the system as well as the environment in which the system will be evaluated.
Dr. Lonsdale and colleagues describe the human digital twin as a ‘mathematical model of a system constructed from all available information.’(2) The goal of developing a digital human twin is to tailor health care interventions at the individual level. The conceptual model has myriad potential applications ranging from guiding anesthetic technique to optimizing the perioperative period and recovery.
We have just begun to see the synergistic application of data generated by a patient’s own wearable devices and data obtained in the clinical setting. Notable examples include smartwatches with continuous heart rate and oxygen saturation monitoring capability and continuous glucose monitors and insulin pumps. The prospect of integrating ALL available data in a mathematical model of a human is intriguing.
The authors describe obstacles to developing and implementing the human digital twin project. Data and device interoperability is essential. Integrating genomic data, electronic health records, personal devices and environmental factors is a daunting process. There are countless technical and regulatory challenges that need to be addressed at institutional, state, national and international levels. In the US, the Food and Drug Administration issued an action plan with a regulatory framework for regulating artificial intelligence and machine learning technologies in medical devices.(4) The framework classifies algorithms that guide or automate treatments as high risk, in contrast with low-risk algorithms that provide information without automating or directing care. Tools like the human digital twin model could be construed as high or low risk depending on how they are implemented. For example, a tool that displays a score predicting a successful wean off mechanical ventilation might be considered low risk, whereas a “high risk” algorithm would automate the decision to transition to spontaneous respiration from mechanical ventilation.
There are potential drawbacks to the human digital twins. Equitable access is a concern. Developing an accurate digital twin requires comprehensive, high-quality data. Access to wearable devices, internet connectivity, and health systems with electronic health record systems are not universal in the U.S.A. This imbalance of access could lead to under-representation of vulnerable populations in these models, a concept known as the digital health divide.(5) Children are particularly interesting, because they do not typically have devices that continuously monitor their activity. Families with multiple children face the added challenges of managing all the connected devices and the costs of any subscriptions to services. Lonsdale and colleagues mention concerns about data privacy and lack of confidence that the data will be used to benefit the individual as potential barriers to adoption. Human digital twin models could be used to characterize an individual’s risks for illness and disease and influence insurance coverage and eligibility.
Innovative ideas create challenges and new opportunities, while technology continues to advance at a staggering pace. In the next ten years, will our patients receive individualized care tailored in a low-risk, digital environment? We hope that the long-held promise of leveraging technology to improve patient care and human health comes to fruition. Time will tell.
Olivia Nelson, MD, Allan F. Simpao, MD MBI, Jorge A. Gálvez, MD MBI
References:
1. Mack CA. Fifty years of Moore's law. IEEE Transactions on semiconductor manufacturing. 2011 Jan 20;24(2):202-7.
2. Lonsdale H, Gray GM, Ahumada LM, Yates HM, Varughese A, Rehman MA. The Perioperative Human Digital Twin. Anesthesia & Analgesia. 2022 Apr 1;134(4):885-92.
3. Chryssolouris G, Mavrikios D, Papakostas N, Mourtzis D, Michalos G, Georgoulias K. Digital manufacturing: History, perspectives, and outlook. Proceedings of the Institution of Mechanical Engineers, Part B: Journal of Engineering Manufacture. 2009;223(5):451-462. Doi:10.1243/09544054JEM1241
5. Jenkins CL, Imran S, Mahmood A, Bradbury K, Murray E, Stevenson F, Hamilton FL. Digital Health Intervention Design and Deployment for Engaging Demographic Groups Likely to Be Affected by the Digital Divide: Protocol for a Systematic Scoping Review. JMIR Research Protocols. 2022 Mar 18;11(3):e32538.