1 Background

Care for patients is complex and requires a multidisciplinary team of healthcare professionals to collaborate in order to optimize patient outcomes and avoid potential errors. In the best of circumstances and despite the wide availability of advanced monitoring and life support capabilities, cognitive lapses of the clinical team, which constitutes a joint cognitive system within a natural work domain, do occur. Such lapses are more likely in resource limited or “austere” circumstances – as in lower income countries or high-income countries during mass casualty events. These lapses result in less than optimal patient care, ineffective utilization of resources, and, in some circumstances, patient harm. While the field of telemedicine is over 50 years old [1], new and emerging technologies offer previously unavailable capabilities for clinicians to better utilize computer systems to decrease cognitive load, enhance decision making, and, hopefully, improve patient outcome. The military has invested significant research funding in developing clinical decision support (CDS)/artificial intelligence (AI), telemedicine/virtual health, and autonomous technologies to enhance combat casualty care throughout the evacuation continuum (Fig. 1) [2, 3]. For the last 20 years of conflict, the US Military has experienced air superiority and freedom of movement across the active battlespace, allowing for quick evacuation casualties to advanced medical support. However, the military anticipates that in future conflicts, access to advanced medical decision-making and surgical stabilization may be restricted or significantly delayed due to distance or adversary denial to freedom of movement. Consequently, there will be increased need to access the capabilities, particularly the expertise, of the multidisciplinary medical team in more austere, severely resource limited, pre-hospital environments at or near the point of injury [4,5,6].

Fig. 1.
figure 1

Military vs. Civilian Care continuum. AIREVAC, Air evacuation; CASEVAC, Casualty Evacuation; CONUS, Continental United States; CCAT, Critical Care Air Transport; ED, emergency department; ICU, Intensive care unit; MEDCEN, Medical Center; MEDEVAC, medical evacuation; OR, Operating room; R1–4, Role 1–4.

Recent natural disasters and mass casualty events across the United States and abroad spotlight the challenges of resource limitations associated with austere medicine. Austerity is not a location, but rather a clinical context in which resources or expertise are severely limited or, in some cases, absent. It is in this context that the military is focusing efforts to develop telemedical, autonomous, and robotic systems to support local caregivers by providing enhanced medical capabilities, confidence, knowledge, expertise, and, when possible, new technical skills. Consequently, these technologies have potential importance even in well-developed healthcare systems when certain resources, like experience with a new surgical approach or a new medication, are limited.

Examples of such innovation are already apparent in the civilian market following natural disasters [7,8,9,10]. In the wake of Hurricane Harvey, Dallas-based Children’s Health set up pediatric telemedicine consults for displaced patients in shelters and companies like American Well, Doctor on Demand, and MDLive offered free remote consults to patients in affected areas [11]. After Hurricane Maria’s landfall in Puerto Rico, the US Army’s 47th Combat Support Hospital enabled telemedical consultation using satellite signal to areas with limited power, water, and even telephone lines (personal communications, authors JCP, KLD, DK).

Through these examples, it is possible to envision a future where military and civilian partnerships might enable medical aid in areas devastated by nature, disease, or even war. Virtual access to the military’s robust and experienced trauma centers during peacetime also has the potential to provide enhanced clinical care to places without access to their own resources, providing the possibility for better clinical outcomes and reducing the costs of establishing such trauma/inter-disciplinary teams in every civilian facility.

CDS/AI systems, telemedicine, autonomous medical devices, and robotics will contribute capabilities to the current and future care of casualties in austere environments. These solutions provide comprehensive medical care in austere environments by bringing medical expertise and new capabilities to the point of need. Experts require information and particular resources to deliver their care, but not necessarily physical presence. Clinical decision support and telemedicine are current means to deliver expertise to the point of need; however, the local caregiver who physically delivers it requires specific training and equipment for it to be effective. Automation and robotics may alleviate some of the physical demands of the local caregiver and AI/deep machine learning are expected to advance CDS to the point where remote consultation with a telemedicine provider may become unnecessary for at least some clinical scenarios. In light of these anticipated needs, the Army has created three new Science & Technology research task areas: (1) Virtual Health, (2) Medical Robotics, and (3) Medical Autonomous & Unmanned Capabilities. This paper discusses the clinical challenges and capability gaps of providing comprehensive medical support in these environments, identifies some of the tools the military is developing to address them, and outline their potential benefits for the civilian healthcare.

2 Solutions

2.1 Telemedicine

Telemedicine, also referred to as virtual health, involves the use of telecommunication and information technologies to provide health assessments, treatments, consultation, and other services across distances [12]. There are four well-recognized types of telemedicine: synchronous, asynchronous, remote patient monitoring, and mobile health. In synchronous telemedicine, the remote expert uses bidirectional communications technology to conduct or direct medical care in real time. This form of telemedicine requires the largest bandwidth, however offers the highest fidelity of information. Asynchronous telemedicine, also called “store and forward” telemedicine, involves the transmission of recorded health information to a remote specialist who then renders care outside of real-time interactions. In remote patient monitoring, personal health and medical data is collected via electronic communication technology, and then transmitted back to a provider (in a different location) to aid in medical decision making. Lastly, mobile health involves the use mobile communication devices, such as cell phones, tablets, or computers, typically in order to conduct public health practice and education.

By leveraging communication technology, telemedicine has the ability to connect less-experienced and sometimes untrained, medical care providers with medical specialists in order to enhance point of care clinical decision making and medical interventions. Optimal telemedical care, especially for complicated patient conditions, requires access to real-time bio-physiologic patient data (i.e. monitors, ventilator, infusion pumps, etc.). Unfortunately, transmission of this information across secure military networks in a standardized fashion for review by the remote expert is not always possible. Telemedicine also requires connectivity between the remote expert and local care-giver or patient. The amount of bandwidth required will vary based on the type of telemedicine encounter being performed, number of casualties, and the software that is being used to conduct it (Fig. 2). Given the bandwidth-constrained nature of most military operational environments, and likely any civilian natural disaster at least in its early stages, this requirement for connectivity is the principle limitation for current forms of telemedicine [7, 11, 13]. While the military has a long history of utilizing and advancing telemedicine and telemedical technologies [13,14,15,16,17], recent language in the National Defense Authorization Act of 2017 [18] will promote and advance this technology solution even further.

Fig. 2.
figure 2

Telemedical capabilities, number of casualties, and relative network requirements. Not shown is jitter or packet errors, both of which need to be low for higher capabilities.

One current project, the ADvanced VIrtual Support for OpeRational forces (ADVISOR) is worth highlighting. ADVISOR is a low cost, low bandwidth, highly reliable pilot program funded by the Telemedicine and Advanced Technologies Research Center, Fort Detrick, Maryland. ADVISOR facilitates telephone communications using an automatic call distribution system to connect local caregivers with remote experts [4]. Phone calls are augmented by e-mail messages that include background casualty information and photographs to provide remote consultant consultation. If and when network capabilities allow, and local hardware is available, phone calls may be “escalated” to real-time video teleconferencing (VTC) with or without use of peripheral, video assisted physical exam equipment. ADVISOR brings medical expertise to a patient’s side anywhere and at any time to optimize outcomes and reduce costs by avoiding unnecessary deployments of providers and evacuations of patients. Lessons learned strongly suggest that training local caregivers and remote experts with the new technologies is as essential for their adoption in the austere care setting as is the native usability of the technology itself [19]. This low-cost system also seeks to integrate and evolve technologies to optimize combat casualty care across the evacuation spectrum.

As of January 2018, the ADVISOR system has supported over 70 real and simulated patient care encounters, primarily with virtual critical care support, but also with general surgery, orthopedic surgery, toxicology, and infectious disease. In both real world and training scenarios, the vas majority clinical support was provided with telephone calls only or telephone calls combined with e-mails (>60%). Video tele-consultation was use in <5% of cases, primarily due to poor network capabilities and low bandwidth. Real world cases have sought urgent recommendations about management for ocular and periocular infection, possible weaponized chemical/biologic exposure, fracture management, sepsis evaluation and management, wound care, surgical decision making, and evacuation guidance. Training cases primarily involve burn care, trauma, hemorrhage, shock, and respiratory failure management.

2.2 Clinical Decision Support Systems

One approach to mitigate the need to project clinical expertise to the point of need is the use of computer and information technologies to assist providers in delivering expert patient care when there are no experts available. This includes situations ranging from delivering care in a facility when there is no local expert available to assist in treating patients to delivering care by non-expert providers in geographically remote environments. Software or information technology-based solutions used to assist in the delivery of care are considered Computer Decision Support (CDS) Systems. In very resource constrained environments, the use of CDS could augment local caregiver capabilities – which in some cases may be no more than lay person rescue – by providing simple tools to allow providers to easily navigate and document patient care based on a clinical practice guideline to more complex systems that can monitor, diagnose, and make treatment recommendations for the optimal care of the patient [20,21,22,23].

One example of a military CDS system developed by the military is the Cooperative Communication System (CCS) [21, 24, 25]. The goal of the CCS system is to provide users with a tools that enhances patient care by (1) improving and optimizing the display of information presented to the user based on the status of the patient and the user role, and (2) leveraging machine learning technologies to assist providers in managing patients by matching the current patient’s condition with historical patients that have been analyzed and matched a multivariate model of patient variables and characteristics. The CCS system is an example of a CDS implementation that leverages existing data sets to provide users with enhanced capability tailored to the needs to the patient and providers through the use of enhanced displays and data analysis/modeling.

Once a treatment plan is required, a CDS can also be used to optimize and manage the care of the patient by providing tools to assist both experts and non-experts through the treatment phase for the patient. The Burn NavigatorTM system [26] is an example of a CDS that provides the user with recommendations and treatment options for a patient who has suffered a major burn. The system is an FDA cleared CDS to help manage the fluid resuscitation of a patient during the initial 24 to 48 h after a major burn injury. Using a set of mathematical models, the system provides the user with hourly fluid rate recommendations to optimize the resuscitation needs of the patient. In addition, the system provides an enhanced display that allows providers to better visualize the patient status and treatment effects and increase situation awareness.

Tools such as these have been shown to increase provider efficiency and improve patient outcomes by harnessing the power of information technology through several ways. First, providing users with enhanced displays and interfaces, these systems increase provider situational awareness by converting data into useful information through the use of trends, graphs, and other advanced display technologies. Second, the use of these enhanced visualization tools reduces the cognitive load for providers and increases their efficiency during patient care [27]. This is especially critical in combat casualty care when patients may need to be cared for by less educated or less experienced providers. Third, CDS systems that provide treatment recommendations and options may further improve the management of the patient by assisting the provider in guiding the appropriate recommendation in cases where the provider may not be aware of the appropriate course of action, or by reinforcing what the expert provider knows by validating a specific course of action.

One common characteristic of CDS systems is the need to maintain a “human in the loop” concept. CDS capabilities are used to enhance patient care by providing users with diagnostics and/or treatment recommendations to optimize patient care. However, once a CDS has been fully validated and shown to provide effective care, the natural evolution of these is as a “closed loop” system to allow for the CDS to fully automate the care of the patient [28,29,30,31]. Models and algorithms driving CDS systems can form the basis for developing fully automated interventional system that require little or no human interface. Bridging the concepts of CDS and full automation will provide users in austere and possibly modern care environments with additional capabilities for managing patients in difficult situations. Use of fully automated systems in scenarios with extended patient care, exceptional resources limitations, mass casualties, or extended transport times is necessary to fully optimize patient management when providers may not have all the necessary resources to properly care for these patients.

2.3 Autonomy and Robotics

The emerging fields of autonomy and robotics represent areas of significant possibility to solve some of the great medical challenges faced in austere care environments [32]. Remote surgery enabled by tele-robotics currently exists using dedicated fiber optic networks. The daVinci surgical robot is currently used in military and civilian hospitals for minimally invasive surgery [33]. This same type of technology could be leveraged, when combined with advanced physiologic sensors, computer vision and other autonomous or semiautonomous systems (i.e. autonomous anesthesia [31]) to deliver effective casualty assessment, triage, and surgical intervention in the absence of local experts and timely evacuation even on unmanned ground and air systems (drones and robots) [6,7,8, 34]. Conceptually, one might imagine that a drone ambulance, fully equipped with a robotic attendant and interactive video communications with a distant emergency medical provider, could respond to an emergency on a remote mountaintop or deep in a jungle.

“Military funded research has demonstrated that surgical robotic systems can be successfully deployed to extreme environments and wirelessly operated via microwave and satellite platforms [35]. However, employment of these capabilities on the battlefield have not yet progressed beyond experimental proofs of concept, are not ruggedized, and are tele-operated component capabilities at best. Significant additional research is required to develop supervisory controlled autonomous robots that can overcome the operational communication challenges of limited bandwidth, latency, and loss of signal in the deployed combat environment. Addressing acute and life-threatening injuries such as major non-compressible vascular injury requires development of new surgical robots that move beyond stereoscopic, bimanual tele-manipulators and leverage advances such as computer vision and application of directed energy technologies already used in non-medical military robotic systems.”

Additionally, since the successful surgery to install fully implantable artificial cardiac pacemakers occurred in 1958, society has slowly become more accustomed to “closed-loop” medical care in which machines perform their function without need for human involvement. However, reliable closed-loop intervention remains an unsolved Artificial Intelligence challenge. Current AI systems do not have the capacity for judgment in the way that medical caregivers do. While clearly a science, medicine remains, in many ways, an art form. In addition to teaching the science of medicine, medical training is an extended apprenticeship that provides vast amounts of experience about how to interpret and make judgment calls in the context of uncertainty [3, 4]. Computers today lack this capacity. Additionally, while some success has been documented with closed loop systems in controlled settings, the reality in the field is more complex and current systems do not have the ability to account for such variability effectively enough to be proven clinically safe [3, 4]. Interestingly, telemedical technologies may offer a means to “bridge the gap” between current computer capabilities and human judgment in the context of uncertainty and might allow semi-autonomous systems to be utilized in the near term (Fig. 3) [36]. Telemedicine provides opportunities to record data that could help teach computers about human decisions: the data that remote human experts require to make informed decisions during telemedical encounters is the same data that autonomous systems will need to make similar “decisions” in the future. This real-time data includes information from interoperable, cyber secured medical devices (i.e. patient monitors, medication pumps, ventilators, robotics) and the data analytics to make these signals standardized, synchronized, and salient. As robotic capabilities continue to evolve, autonomous and tele-operated semiautonomous robotic patient support systems could enable closed-loop patient monitoring and triage as well as robotic intervention.

Fig. 3.
figure 3

Man vs. Machine, the progression of automation and how telemedicine can start to bridge the gap between artificial intelligence and expertise in medicine. Telemedicine brings the expert to the point of need in the absence of or in addition to the effective clinical decision making by inexperienced clinicians aided by automated systems and clinical decision support systems (CDS).

Transitioning between human in-the-loop systems and completely autonomous, closed loop medical care systems represent a major technical hurdle facing current medical care. In anticipation of combat casualty care in the future; the military is investing heavily in research portfolios to develop solutions to bridge this gap. Near-term prototypes will involve a hybrid of human and autonomous care models that shift between more human-intervention and more autonomous-care based on the clinical, technical, and logistical needs of a given situation. The military has identified the following as the most realistic research initiatives based on the complexities of the technology involved, and the current state of the medical robotic and autonomous systems (Fig. 4):

Fig. 4.
figure 4

Source: Joint Program Committee 1; Medical Simulation and Information Sciences, US Army Medical Research and Materiel Command, Fort Detrick MD and the Defense.

Medical Device Interoperability Research Objectives to Support Telecritical Care.

  1. (1)

    Development of an autonomous closed loop critical care systems-of-systems based on autonomous clustering and intelligent agents interacting with each other to provide care for multiple polytrauma patients. This type of system would act as a medical force multiplier by increasing medical care capacity during prolonged field care or mass casualty situations.

  2. (2)

    Support continuous performance improvement and enable self-learning systems that can rapidly adapt to changing scenarios.

  3. (3)

    Investigating artificial intelligence and machine perception systems for accurate detection and modeling of the human body to enable semi-autonomous robotic casualty extraction from complex environments and robot procedural intervention (e.g. placement of interosseous needle, needle thoracostomy, cricothyrotomy); these applications require high fidelity mapping of the human body in near-real time for safe physical contact with the casualty.

2.4 The Medical Fusion Center

Optimization of the complex care environment for both military operations and civilian emergency response requires a coordinated, real-time common operating platform, much like the control room for space flight. Alternatively, a virtual control room can be developed allowing interactions from around the globe. Real-time visibility of all medical assets in support of both civilian and military conventional and special operations forces, particularly in kinetic multi-domain battlefields or disaster relief operations will facilitate optimal use of resources during evacuation, transport, patient care, tele-support/mentoring, and resupply. Ultimately the goal is a comprehensive “System of Systems” to support multiple end-users with medical fusion centers of global medical capabilities to facilitate a coordinated medical system and support all dimensions of healthcare. Key components of this type of center include: (1) real-time visualization of the area of operations; (2) a “medical intelligence” platform comprised of tele-support, decision support and analytics to facilitate clinical care, operational and logistic requirements, and evacuation/patient movement; and (3) robust reporting features for continuous performance improvement processes.

The primary aims of a functioning medical fusion center are to: (1) Send the appropriate resource (personnel, platform, and capabilities) to the right place at the right time to meet the specific patient requirements, (2) Support continuous performance improvement and enable self-learning systems, (3) Inform algorithms and decision support systems for tasking personnel assignments, medical treatment facility placement, and patient evacuation/movement (“Intelligent Tasking”), (4) Facilitate Virtual Health: tele support, just-in-time mentoring, remote patient mentoring, and reduced medical error rates, (5) Interface with allied military services, civilian and Government emergency response organizations.

Fundamental challenges to such a comprehensive, interconnected system include adequate connectivity and bandwidth. Assumptions of interruption of communications in a military or emergency setting are understood so optimizing local decision support tools will be essential. Streaming real-time information may not always be possible so AI will be needed to supplement decision support algorithms.

3 Future Directions

Medicine is woefully behind other industries, such as aviation and automotive industries, in adopting advanced semi and fully autonomous operations [37, 38]. The opportunity to provide high quality, reproducible, and effective monitoring for patients that informs local and remote care providers is an important, and achievable, goal for healthcare; it is a requirement for a military that needs to support isolated patients in austere locations or casualty evacuation in unmanned, pilotless systems. To this end, the Defense Health Agency and the military services, along with the other federal agencies such as the Food and Drug Administration and the National Institute of Standards and Technology, are moving toward investing in a comprehensive program of research. The overall objectives of this program are to develop an open, standards-based technical architecture and reference model for medical device data, sensors and actuators, communications enablers, algorithms, and knowledge representation that is suitable for informing autonomous, closed-loop, human-computer integrated CDS (Fig. 5).

Fig. 5.
figure 5

Source: Joint Program Committee 1; Medical Simulation and Information Sciences, US Army Medical Research and Materiel Command, Fort Detrick MD and the Defense Health Agency.

Future Vision of Medical Device Interoperability and Telecritical Care.

4 Conclusion

Rapid advancements in both medical and non-medical technologies offer the potential for improving clinical performance of both novice and advanced medical teams in resource limited contexts. Additionally, the correct application and integration of new medical combat casualty care doctrine with proven technologies offers significant improvements in both readiness and access to care while reducing unnecessary and risky evacuations. Beyond the potential for optimizing casualty care in the military, these efforts should be applicable to civilian healthcare systems in the future. Military human-computer models that support telemedicine and autonomous care in austere environments may help shape similar civilian healthcare solutions in similar environments. An open, standards-based infrastructure of medical devices could revolutionize healthcare by reducing research and development time, facilitating integration of new devices into the medical ecosystem, reducing costs, increasing healthcare reliability and safety, improving documentation of care, enabling more effective training, making logistical support more efficient, and, ultimately, improving patient outcomes. Efforts to develop and evaluate these solutions require continued military, academic, industry, cross -government, and international collaboration and technical ingenuity to be successful now and in the future.