Front. Disaster Emerg. Med. Frontiers in Disaster and Emergency Medicine Front. Disaster Emerg. Med. 2813-7302 Frontiers Media S.A. 10.3389/femer.2026.1756349 Original Research Digital cross-sectoral training with dynamic patient and situation simulation in the context of mass casualty incidents—insights from the research projects D2PuLs and D2PuLs PRO Konrad Robert 1 * ‡ † Conceptualization Resources Validation Writing – original draft Pfadenhauer Tim 2 † Conceptualization Resources Validation Writing – original draft Eckenweber Luca 1 3 † Conceptualization Validation Writing – original draft Software Visualization Banahovschi Alexei 3 Software Writing – review & editing Braun Christopher 3 Software Writing – review & editing Reuter Thomas 4 Resources Validation Writing – review & editing Data curation Popp Steffen 4 Resources Validation Writing – review & editing Data curation Engelen Christian 5 ‡ Resources Validation Writing – review & editing Data curation Baierl Lukas 1 Software Visualization Writing – review & editing Soy Mustafa 3 Writing – review & editing Funding acquisition Project administration Supervision Bauer Christian 1 3 ‡ Writing – original draft Writing – review & editing Funding acquisition Project administration Supervision Institute of Rescue, Emergency and Disaster Management (IREM), Technical University of Applied Sciences Würzburg-Schweinfurt (THWS), Nuremberg, Germany Malteser, Educational Center for Emergency Medical Services, Wetzlar, Germany nVista technologies GmbH, Nuremberg, Germany Nuremberg Hospital, Campus South, Nuremberg, Germany Emergency Department Lauf Hospital, Lauf an der Pegnitz, Germany Correspondence: Robert Konrad, robert.konrad@thws.de

ORCID: Robert Konrad orcid.org/0000-0001-9082-0237; Christian Engelen orcid.org/0009-0000-2647-1183; Christian Bauer orcid.org/0000-0002-2267-3009

These authors have contributed equally to this work and share first authorship

 12 02 2026 2026 4 1756349 28 11 2025 11 01 2026 16 01 2026 Copyright © 2026 Konrad, Pfadenhauer, Eckenweber, Banahovschi, Braun, Reuter, Popp, Engelen, Baierl, Soy and Bauer. 2026 Konrad, Pfadenhauer, Eckenweber, Banahovschi, Braun, Reuter, Popp, Engelen, Baierl, Soy and Bauer https://creativecommons.org/licenses/by/4.0/ This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms. Introduction

Managing Mass Casualty Incidents (MCI) demands complex, cross-sectoral competencies. However, training these skills using traditional full-scale exercises is often limited by high resource requirements, low frequency, and a lack of patients (patho-)physiological dynamics. Against this backdrop, the research projects D2PuLs and D2PuLs PRO (Digital Dynamic Patient and Scene Simulation) developed a modular and adaptive digital simulation environment designed for interdisciplinary and cross-sectoral disaster response training.

 Methods

A core element of this system is a dynamic physiology model that simulates (patho-)physiological changes and therapeutic interventions in real time, replacing static vital signs with realistic, intervention-dependent progressions. The platform facilitates configurable scenarios ranging from individual training on mobile devices to distributed large-scale exercises incorporating external situation simulators and physical simulation manikins.

 Results

Preliminary results of initial field tests and subsequent surveys of Emergency Medical Services (EMS) personnel and clinical teams showed the fundamental technical and didactic applicability of the system across the entire rescue chain—from initial care and triage to transport and trauma room management. The results provide solid, albeit preliminary, indications that the system can be successfully utilized to train decision-making, triage, and resource management across sector boundaries, while providing digital data foundation for debriefings. Although the complexity of the available medical interventions initially placed high demands on user operation, the modular architecture proved effective in adapting scenarios to specific learning objectives.

 Discussion

The study concludes that dynamic digital simulations represent a scalable and valid supplement to practical exercises, serving to sustainably strengthen operational confidence in disaster medicine.

 customizable simulation platform digital mass casualty incident (MCI) training emergency room training interdisciplinary disaster response simulation scene simulation triage training virtual patients Bundesministerium für Forschung und Technologie 10.13039/501100004937 13N15503 13N17018 The author(s) declared that financial support was received for this work and/or its publication. Funding was received by the Federal Ministry for Research, Technology, and Space (BMFTR) under the program “Research for Civil Security” (Grant numbers: 13N15503 and 13N17018) and supported by the publication fund of the Technical University of Applied Sciences Würzburg-Schweinfurt. section-at-acceptance Disaster Medicine Introduction

Simulation training has long been an integral component of education, advanced training, and continuous professional development in emergency medical services (EMS) and emergency medicine. It facilitates experience-based and case-oriented learning within a controlled environment. Its deployment in EMS and medical education aims to develop and consolidate practical competencies through simulation before they are applied in real-life emergency missions. Errors in decision-making or execution do not result in actual harm during simulation training. Consequently, simulation training is regarded as the gold standard for conveying action-oriented, practical operational competencies (1).

The preparation, execution, and technical setup of classical, non-digital simulation training for major incidents involving mass casualties are highly resource-intensive and demanding. As a result, the workflows and performance of the emergency forces cannot be captured in sufficient detail for fact-based evaluation and debriefing, often rendering the training only partially effective. Furthermore, these exercises are typically associated with a high degree of artificiality.

Moreover, such exercises are often limited in terms of scalability and systematic evaluation. Due to the high organizational and significant financial effort involved, full-scale exercises are conducted only at long intervals. From the perspective of sustainable experience and competence acquisition, they effectively occur too infrequently. Participating organizations, such as regular EMS providers, are often involved only occasionally and to a limited extent, as maintaining regular service operations takes priority (2). Consequently, in practice, specific units usually train only isolated concepts—such as “first arriving ambulance” or “triage in an Managing Mass Casualty Incidents (MCI) situation”—separately or via (online) courses (3, 4). Volunteer emergency forces generally receive their primary training through basic helper courses, specialized modules, and practice evenings, while incident command personnel are trained via scenario-based exercises which, compared to full-scale exercises, possess a higher degree of artificiality (5).

A significant disadvantage of live field exercises is that training success can rarely be reliably tracked, as the learning objectives for individual actors are often inadequately formulated or cannot be reliably traced (6). A detailed description of learning objectives, down to the level of granular goals for specific roles within a live exercise, could concretize the horizon of expectations. Providing a sufficient number of trained and technically proficient exercise observers for the respective roles is another frequently challenging task in the preliminary planning of live exercises. Even if this task is successful, observers must be provided with detailed evaluation forms to limit their scope for subjective interpretation. The quality of data collection is then subject to the individual limitations of the observer during the live exercise, which in turn complicates a uniform and systematically structured evaluation. Consequently, cross-actor, cross-interface, and cross-sectoral training domains are often not adequately represented, despite their critical importance in the casualty care process (7).

Despite the steadily increasing availability of digital training environments on the market, these systems generally are limited to specific training objectives and scenarios, offering insufficient adaptability to the needs and operational constraints of the trainees.

Studies by Bauer and Loose (1), García Ulerio et al. (8), and Zhang et al. (9) demonstrate how simulation approaches are establishing themselves as effective teaching and learning methods in the field of emergency medicine. This development is notably reflected in the revised Training and Examination Ordinance for Emergency Paramedics (Notfallsanitäter) in Germany (10). In its initial 2014 version, this regulation permitted simulation-based training solely in the field of gynecology. With the 2023 revision, topic-specific simulations are now permissible in the fields of anesthesia and intensive care medicine, accounting for up to 25% of the required hours (11).

Complementing classical, non-digital simulation approaches, digitalization offers opportunities to provide more cost-efficient training variants, particularly through modular and hybrid approaches (9, 12). Specifically, regarding the practice of Mass Casualty Incidents, very few offerings are currently known in the German market. Computer-based tabletop exercises for various disaster and accident scenarios are currently found primarily on the US market (e.g., Virtual Heroes). In the realm of Virtual (VR) and Augmented Reality (AR), the DRK Rheinhessen-Nahe (German Red Cross) presented MANV3D in 2018, a pilot project featuring a VR environment specifically designed for MCI situations. Furthermore, a solution tailored specifically to emergency situations in a military-tactical environment (known as Ersthelfer Alpha) was developed for the German Armed Forces (Bundeswehr). However, this high-investment solution is not yet available for civilian use.

Within the domain of digital exercise environments, a distinction can be made between computer-based tabletop exercises and Virtual Reality (VR) or Augmented Reality (AR) based training environments. In both categories, systems exist or are emerging predominantly for general EMS and emergency medical training. Examples include XVR Simulation for tactical training of emergency services, i:medtasim for VR-supported pre-hospital and intra-hospital emergency scenarios, or UbiSim from Switzerland, which focuses on nursing education. A non-exhaustive overview of further available systems, including their training focus, can be found in Table 1.

Overview of selected digital training platforms for pre-hospital and clinical emergency training.

Provider Weblink Training focus
EMERGE (DE) emerge-game.com Virtual emergency room simulation for medical students; training in clinical decision-making and organizational skills under time pressure
FwESi (DE) fwesi.de Training simulation for fire departments; training within tactical units according to German standards (FwDV)
i:medtasim (DE) tricat.net/imedtasim Multi-user VR training for pre-hospital and in-hospital emergency care; emergency medical services personnel and healthcare professionals
MediTrain VR (DE) meditrainvr.com VR-based medical training applications; multi-user scenarios for emergency medicine
ONEBONSAI (BE) onebonsai.com/cases/projects/vr-paramedical-training Paramedic training for preclinical emergency situations; VR first aid and emergency training
Oxford Medical Simulation (UK) oxfordmedicalsimulation.com Multi-user VR simulations for nursing, medicine, and allied health; clinical training and assessment
UbiSim (CH/USA) ubisimvr.com VR nursing simulation for nursing education; development of clinical judgment and communication skills
Virtual Heroes (USA) virtualheroes.com Immersive training solutions and serious games for the military and healthcare sectors; focus on realistic physiology simulation and tactical medicine
VRPatients (USA) vrpatients.com No-code platform for customizable VR/MR patient simulations; preclinical and clinical training with AI-supported patients
XVR Simulation (NL) xvrsim.com Tactical VR training for fire departments, police, and emergency medical services; incident command and resource management in major disaster situations

Looking at Germany, the largest group of potential simulation users consists of primarily volunteer personnel (approximately 90%) within the civil protection framework (13). These forces are distributed across the four major aid organizations—the German Red Cross (Deutsches Rotes Kreuz), the Order of Malta Volunteers (Malteser Hilfsdienst), St. John Accident Assistance (Johanniter-Unfall-Hilfe), and the Workers' Samaritan Federation (Arbeiter-Samariter-Bund)—as well as the Federal Agency for Technical Relief (Technisches Hilfswerk), fire departments, and other institutions involved in civil protection. None of the current digital training concepts consistently integrates hospitals. Emergency departments and hospital command staff, in particular, would benefit significantly from a digital system offering cross-interface and cross-sectoral training, as intra-hospital care and resource utilization are currently neglected within that context.

 Materials and methods: training with the D2PuLs digital simulation environment

The fundamental concept behind “Digital Dynamic Patient and Scene Simulation” (D2PuLs) was the development of a highly adaptive and scalable digitally supported simulation system, enabling emergency services and hospital units to train for major incidents with mass casualties. The project is currently in the second phase of funding by the Federal Ministry for Research, Technology, and Space (BMFTR) under the program “Research for Civil Security,” funding measure “Innovations in Application—Practical Beacons of Civil Security” (Grant numbers: 13N15503 and 13N17018).

D2PuLs combines previously separate approaches to pre-hospital and clinical scene simulation with medical patient simulation to create a system that supports various simulation techniques through its modular design. This system can be aligned with both operational and command/tactical training objectives. When all components are utilized, it is particularly effective for the interprofessional and cross-sectoral training essential during an MCI.

Based on its modular architecture (building-block principle), the training environment provides a high degree of adaptability and scalability. This allows the system to facilitate diverse use cases-ranging from simple individual training on a smartphone at the station (e.g., patient simulation without situational display for training triage algorithms) to complex, cross-sectoral large-group training in Virtual Reality, featuring situational dynamics, command structures, and a patient flow extending from the incident site to the trauma room.

D2PuLs consists of several, partially optional components (Figure 1). The primary focus lies on the patient and (patho-)physiology model simulation, in which participants can observe, examine, and treat virtual patients. The entire system is built upon this training element, including situation and equipment management (prior to the actual training) and the evaluation component (post-training). Patient data can be linked by the system to third-party scene simulators, as D2PuLs itself provides only a rudimentary 2D scene simulation. This enables the parallel use of third-party scene simulation and the associated D2PuLs patient simulation. During development, the scene simulators “On Scene” by the company XVR Simulation B.V. and “FwESI” by Cybertrain Solutions GmbH were integrated for scene visualization.

Schematic overview of the D2PuLs system architecture. Interaction between the Core System, Patient Simulation, and Scene Simulation. Screenshots of the software “D2PuLs Patient Simulation” reprinted with permission of nVista technologies GmbH, screenshot of the software “XVR On Scene” reprinted with permission of XVR Simulation B.V., and screenshot of “FwESI” reprinted with permission of Cybertrain Solutions GmbH.

Diagram illustrating a core system connected to management and debriefing, patient simulation, and scene simulation. Management includes user interfaces with data and statistics. Patient simulation shows virtual scenarios of patients. Scene simulation depicts virtual environments with emergency scenes. Didactic concept and implementation

The implementation and design of the D2PuLs simulation platform is centered on the concept of experiential learning, as described by Kolb (14), among others. Simulations facilitate “learning by doing,” thereby anchoring knowledge more sustainably. Experiential learning conveys knowledge through the act of experiencing. The opportunity for “learning by experience” is a central element of practice-oriented competence development. The fundamental assumption of this didactic model is that effective learning occurs primarily when learners engage directly and practically with the subject matter. The moment new knowledge is linked to a personal experience, it is stored based on that experience and can be recalled and retrieved significantly more easily.

Accordingly, teaching and learning concepts based on experiential learning rely on the goal-oriented design of situations into which learners fully “immerse” themselves. These are situations that learners experience as real (presence), in which real emotions are triggered, and in which realistic action is possible. In the healthcare sector, it is now generally recognized that simulation training enables safe, adaptive, and competence-oriented preparation for the demands of daily patient care (15).

The D2PuLs simulation environment focuses on promoting the pre-hospital and clinical operational competencies required to manage major incidents involving an MCI. The simulation environment is designed so that the depicted training scenarios can be adapted at any time to the learners' level of competence and map individual learning objectives (12). The training scenarios incorporate decision-making processes, team interactions, and communicative workflows under conditions that mimic real operations as closely as possible.

The integration of structured reflection phases supports the sustainable retention of learned material and fosters metacognitive learning processes. Furthermore, the digital logging of all measures carried out in chronological order in the background also allows for detailed follow-up, enabling learners to critically reflect on their actions once again.

The training system focuses on fostering operational competencies that extend beyond mere knowledge transfer. Training scenarios are developed to depict decision-making processes, team interactions, and communicative workflows under operational conditions. For the reflection phase, a deliberate decision was made against an automatic assessment concept; instead, a trainer is always required for the final evaluation of actions. This ensures the necessary flexibility to accommodate various learning objectives as well as the cross-sectoral applicability of the system in different teaching and learning contexts. The developed digital format also opens up new possibilities for flexible, location-independent training concepts. A central aspect here is strengthening interdisciplinary cooperation between emergency medical services, fire departments, and clinical partners in order to optimize interfaces and establish common operational protocols.

 The physiology model

To ensure simulation participants are confronted with an operational situation that is as realistic as possible, it is insufficient to merely simulate a large number of patients. Patients must also possess a dynamically changing health status. Statically simulated injuries lacking temporal evolution are unrealistic and fail to account for the dynamics resulting specifically from the instability of individual emergency patients. Depending on the learning objective, experience level, and training stage, a portion of the patients should present with severe or life-threatening injuries. This is necessary to appropriately challenge the trainees and, for instance, to conduct goal-oriented training on the allocation of scarce resources.

Current recommendations from Triage Consensus Conferences, based on the analysis of real major incidents, provide concrete planning foundations for this purpose (16). Recent data from evaluations of major incidents indicate a distribution ratio of 20% life-threateningly injured patients (Immediate Treatment/SK I), 30% severely injured (Delayed Treatment/SK II), and 50% minorly injured (Non-urgent Treatment/SK III).

To address the resulting challenges, as well as the requirements for interprofessional and cross-sectoral simulation and training, it is essential to provide users with a treatment course that is as realistic as possible, along with extensive therapeutic options within the patient simulation. The objective here is not to train the correct manual execution of medical measures, but rather the correct selection and sequencing of interventions. The complexity of patient care arises from the multitude of available medications and devices, the sequence of application and treatment, and the variability of injuries.

Classical approaches utilizing static vital signs and predefined (“scripted”) parameter trends (e.g., for heart rate, blood pressure, oxygen saturation, etc.) are therefore unsuitable for adequately representing the required degrees of freedom. To address this complexity, a proprietary digital physiology model was developed within the project. This constitutes a mathematical simulation primarily based on preliminary scientific work—particularly at the University of the Bundeswehr Munich (17)—which was further developed within the scope of the project. The physiology model autonomously represents the functions and processes of the human organism required for patient simulation, as well as the effects of injuries and medical interventions, largely independent of predefined trends and treatment pathways.

The physiology model simulates the body's real-time reactions to injuries and illnesses, as well as its response to treatments performed by the user. It accounts for measurable parameters—such as blood pressure, respiratory rate, and pulse—as well as subjective factors like pain levels and states of consciousness. To automatically calculate all required values, it is necessary to incorporate additional parameters that influence the overall system. Only through the dynamic interactions between individual parameters can a change in respiratory rate, for example, influence oxygen saturation, which in turn impacts pulse and level of consciousness. Conversely, changes in blood pressure, medication administration, or potential traumatic brain injuries can also alter the state of consciousness.

For each patient, it is possible to create arbitrary combinations of currently over 20 different types of injuries and illnesses with varying degrees of severity. Abstracting and reducing the complexity of human physiology into a mathematical model—while capturing the dynamic interactions resulting from injury and illness based on pathophysiological changes and verifying their plausibility—required constant adjustment of the possible parameter combinations. This adjustment is achieved through extensive system testing and trials with actors from pre-hospital and clinical emergency medicine. Crucially, it involves detailed and rigorous validation of individual injury/illness combinations at an expert level within the project consortium.

Within the overall context of the patient simulation, the physiology model represents just one component running in the background. Data exchange occurs via appropriate interfaces both within the simulation itself and with external tools. The entire system was designed from the outset to facilitate connection to third-party scene simulators. A further element currently under development is an interface to a chat component utilizing Large Language Models (LLMs, similar to ChatGPT). This is intended to enable participants to practice communication with patients and further reduce exercise artificiality through natural interaction. Figure 2 illustrates the schematic architecture of the application's various components: ranging from the actual patient visualization with which the user interacts, through the physiology model and the experimental patient communication interface, to the third-party scene simulation tool.

Schematic architecture of the patient simulation. The system connects the 3D patient avatar with the underlying digital physiological model and an AI-based communication interface (LLM, work in progress), while integrating external scene simulation. Screenshot of the software “D2PuLs Patient Simulation” reprinted with permission of nVista technologies GmbH and screenshot of the software “XVR On Scene” with permission of XVR Simulation B.V.

Line graph depicting various physiological parameters over time in seconds. Multiple colored lines represent different data points such as heart rate, oxygen saturation, and temperature. The graph shows a noticeable peak and subsequent decline in some metrics. The y-axis ranges from zero to one hundred fifty, while the x-axis spans from zero to four thousand. The legend on the right lists parameters in German. Model validation tools

Due to the vast number of different physiological parameter combinations and their interactions, a complex web of dependencies emerges, the interrelationships of which are difficult to trace. To counteract this, a suite of tools was developed in parallel with the core model to visualize and trace these dependencies. Figure 3 shows a section of the development environment created specifically for this purpose. Visible here is the internal vascular system, featuring arterial vessels (red boxes), venous vessels (blue boxes), and the intervening microcirculation areas within the tissue (red ovals). The connection of these components forms the circulatory system through which blood flows within the body. The surrounding numerical values and menus illustrate the multitude of calculated parameters and interaction possibilities with the model. Within this environment, it is also possible to execute all interactions, external influences, and data collections that trainees can perform in the actual simulation environment. This makes it possible to perform a wide variety of actions during the validation phase and trace the physiology model's reactions, thereby testing and adjusting the system to ensure that the reactions are medically accurate and realistic.

Development and validation environment of the D2PuLs physiology model. This tool visualizes the complex physiological interdependencies within the model. It enables the detailed analysis of parameters and the verification of physiological reactions to simulated trauma. The control panel on the right allows developers to trigger specific injury patterns and pathologies as well as interventions to validate the model's dynamic response. Reprinted with permission of nVista technologies GmbH, © nVista technologies GmbH.

Diagram illustrating virtual patient simulation. On the left, an animated figure labeled “Virtual Patient” lies on the ground, seemingly injured. In the center, a schematic of a human body with labeled systems, titled “Digital Physiology Simulation.” Arrows indicate interactions like “Virtual Vital Parameters” and “Virtual Interventions.” On the right, an image of a “Virtual Patients in Mass Casualty Incident Scene” depicting a busy, festive outdoor setting.

The development environment shown in Figure 3 represents just one of the developed tools. However, since this tool can only display the current state of the model rather than its temporal progression, an additional visualization tool was developed (Figure 4). This graphical interface allows for the observation of parameter changes at any specific point during the simulation. This enables the assessment of rates of change and a more effective verification of temporal dynamics—for instance, determining whether a hemorrhage results in realistic blood loss or an appropriate progression through shock stages.

Developer Time-series analysis interface of the physiology model. This view plots multiple physiological parameters simultaneously to validate stability and reaction times. It enables developers to trace the specific impact of trauma and interventions on the patient's vital signs over time. Event markers on the time axis (bottom) indicate critical state changes, such as the onset of different shock classes or physiological reactions like vomiting. The interface is shown in the original German language used during the Development. Reprinted with permission of nVista technologies GmbH, © nVista technologies GmbH.

Flowchart depicting a training process with three columns: “Configuration,” “Group Training,” and “Debriefing & Evaluation.” Configuration includes scenario, patient, materials, role, and permission setup. Group Training covers role assignment, scenario moderation, learner guidance, and observation and logging. Debriefing & Evaluation involves providing feedback, assessing interventions, and decision-making. Reusable templates are indicated by diamond icons.

To facilitate discussion of physiological progressions, the system provides the option to save and export all physiology model values at any given timestamp. Using this data, the simulation can be restarted multiple times from that specific point to test system stability or to observe varying outcomes resulting from the application of different interventions. Furthermore, all external inputs to the model (injuries, environmental properties, and performed examinations/treatments) can be exported following a simulation run. This data allows for the future automated replication of the scenario at any given time. This functionality is critical for observing and verifying changes in model behavior; the corresponding tool re-simulates the original scenario conditions following codebase modifications and clearly highlights any (temporal) deviations. This ensures that modifications do not lead to unforeseen consequences in the model's behavior.

The described tools are utilized to generate possible physiological progressions for virtual patients, to analyze anomalies, and to discuss findings with subject matter experts. Quality control is conducted through the simulation of various injury patterns and subsequent treatments within the aforementioned environments. This process involves a detailed examination of individual parameters, both those accessible to the user (e.g., respiratory rate, peripheral perfusion, etCO2) and internal parameters used solely to sustain the simulation (e.g., oxygen supply-to-demand ratios in specific tissue sections, blood flow in individual vessels, pulmonary gas composition). Simulation results are evaluated by subject matter experts (Emergency Paramedics, Emergency Physicians) and benchmarked against both literature values and practical experience from pre-hospital and clinical settings. This ensures the generation of realistically aligned progressions across all available injury and treatment combinations.

 Injury patterns

The calculation of pathological changes in vital signs resulting from injuries is always based on a model of a healthy patient. This model represents the normal physiological state of an average adult male, defined according to medical textbooks and the practical experience of the participating medical experts. Upon the start of the simulation, a copy of this base model is generated for each patient, to which specific injury parameters are subsequently applied. The external influences introduced in this manner, such as blood loss and its sequelae caused by severe hemorrhage, alter various parameters, which in turn mutually influence one another through dynamic interactions.

Although the system attempts automatic physiological compensation for certain parameters, it is designed to reach its limits of compensation over time, depending on the severity of the injury. For example, blood loss can only be adequately compensated for a limited duration before the patient progresses through the four stages of shock. In the event of insufficient or absent intervention, the patient enters the decompensation phase, ultimately reaching the irreversible phase resulting in death.

Within the patient simulation, participants can perform a diverse range of diagnostic and therapeutic measures to stabilize the patients' condition. The system includes the most frequent procedures and diagnostic techniques employed in pre-hospital settings and trauma rooms. These include manually assessable parameters and treatments, oxygen administration, various medications, fluid resuscitation, blood gas analyses, diverse imaging modalities, a multitude of hemostatic measures, and many others.

Furthermore, all physiology model parameters can be transmitted to external simulation manikins and monitors via corresponding interfaces. This makes it possible to utilize the physiology model in the background while conducting parameter assessment and, to some extent, patient treatment through external hardware or applications.

 Programming and system architecture of D2PuLs

The patient simulation constitutes the primary application with which the participants interact. The physiology model is integrated into the system and simulates the behavior of each individual casualty in the background. Consequently, direct user interaction with the model is minimal and occurs primarily indirectly during the training evaluation.

The patient simulation is designed to run on mobile devices such as tablets as well as on laptops. This design aims specifically to lower entry barriers, ensuring that no major hardware investment is required before deployment, depending on the specific use case. This ensures that smaller departments or volunteer organizations can easily afford to acquire the system.

Training modes

Training can be conducted in single-player or multiplayer modes. In group training, multiple individuals working on their own devices can train on a scenario jointly. Similar to managing real operations, this allows for the definition of different areas of responsibility (e.g., triage and treatment) and enables action based on a division of labor. Group training does not need to take place at a single location; it can be distributed across various sites and organizations. For example, the pre-hospital part of the incident management can take place at an ambulance station, while the clinical part occurs in the training rooms or simulation centers of participating hospitals.

Group training offers several options for interaction between participants. In one setting, all participants are located in the same area, can see one another, and support each other in treatment. A second option represents a hybrid of group and single-player modes. Here, all participants receive the same patient simultaneously but treat them completely independently. This makes it possible, for instance in a classroom setting, to conduct individual patient treatments in a structured manner, followed by a joint discussion of the measures taken. This facilitates the rapid identification and discussion of differing approaches and potential errors in judgment.

 Hardware

Individual training has minimal hardware requirements beyond the specific device used for training. To view the graphics and text instructions clearly, a laptop or larger tablet is recommended. For group training, every participant and the operator (acting as the training coordinator) requires their own device. Furthermore, all participants require a stable internet connection. Since the operator's device may need to simulate a large number of physiology models (up to one model per patient for each participant, depending on the mode), a high-performance laptop (e.g., a gaming laptop) is recommended for this role.

If a third-party scene simulation tool is used, requirements increase significantly. In the “One scene simulation for all players” setting, a highly visible presentation medium such as a projector is suitable for sharing the scene visualization with participants. In this case, the operator's laptop can manage both the D2PuLs group session and the external scene simulation. If every participant is to play their own scene simulation, all participants require a suitable computer, an appropriate license for the simulator, and potentially a controller. These computers should meet the system requirements of the respective situation simulators.

In addition to digital devices, the training setting can be didactically enriched to a hybrid simulation setting using physical simulation manikins and other analog and visual aids. For instance, triage algorithms can be distributed as printed “cognitive aids” or visualized via projection in the room. This enables scalable support that can be flexibly adapted to the respective competence level of the learners. The use of physical simulation manikins does not further alter requirements, apart from the necessary compatibility of the manikin and its corresponding control software.

 Training capabilities

It is possible to train both the pre-hospital and clinical phases of incident management. This encompasses the situation at the incident site from the arrival of the first responders, through transport to the hospital, and finally triage and/or treatment in the trauma room. Depending on the scenario, different interventions and environments are available. In the clinical phase, facilities such as laboratory or blood gas analysis and a multitude of imaging modalities are available, which are absent in the pre-hospital setting. In the pre-hospital phase, patients are untreated and injuries are very fresh, representing the initial stages of the simulation. Conversely, in the hospital setting, patients arrive after a delay, as primary triage, initial care, and transport have already occurred. Consequently, injury presentations may have already evolved, and the physiology model has been running for several minutes. The focus here shifts to more precise diagnosis, further treatment, and potential surgery.

These two components can be played completely independently or as a single comprehensive training unit. When played jointly, the same patients treated in the pre-hospital phase—including the applied interventions—are handed over to the clinical phase, where other participants assume responsibility for secondary care, examinations, and triage. The simulated transport time to the hospital bridges these phases. If only one component is played, patients are transported to a virtual hospital and subsequently leave the simulation. In clinical-only scenarios, it is possible to inject pre-treatments performed by virtual first responders, allowing pre-treated patients to arrive at the hospital without requiring prior treatment by actual participants.

 Simulation training workflow

Within the D2PuLs system, it is possible to conduct all stages of simulation training: from scenario and incident creation, through the definition of injury patterns and patients (Configuration), to the actual exercise (Training) and the subsequent analysis (Debriefing and Evaluation). An overview of the required activities is provided in Figure 5, and these steps are detailed below.

Phases of a D2PuLs training session. The diagram illustrates the necessary steps for the operator/instructor, ranging from scenario configuration to the final evaluation. Reusable templates (marked with diamonds) support the configuration and role assignment process.

Flowchart depicting a medical simulation interface with detailed blood circulation data. The chart includes systolic and diastolic values, oxygen saturation, and CO2 levels. Multiple pathways and conditions, such as fractures and brain injuries, are highlighted. A sidebar lists categories like “Fractures” and “Pneumothorax,” indicating different medical conditions. Technical terms and numerical data are prevalent. Scenario creation

The core of every exercise is the simulated incident scenario. This can be defined within the scenario editor. Here, it is possible to define background information, select the map on which participants will move, and define the starting positions and injuries of the patients. The system is designed to allow the deactivation of the application's integrated 2D scene simulation in favor of connecting and utilizing third-party full-scale scene simulators. Thus, the scene simulators XVR and FwESI can be connected to the platform, replacing the movement on the map with these simulators. These simulators offer a fully featured 3D environment through which the participant can move using a controller. The internal scene simulation within D2PuLs is limited to an abstract 2D map (see Figure 6). When a user approaches a patient in the third-party scene simulation, the treatment view within the patient simulation opens.

Built-in 2D scene simulation (in a clinical setting). The interface displays the floor plan of a partner hospital to facilitate spatial orientation. The minimal scene simulation enables the management of patient flow to the colored operational areas, such as triage area (blue) and the shock room (yellow). Reprinted with permission of nVista technologies GmbH, © nVista technologies GmbH.

Floor plan interface with labeled rooms and icons. Rooms include “Triagebereich” and “Schockraum 1,” with user markers P0024, P0029, P0022, and P0021. Navigation controls are on the right. Timer reads 00:09:27. Patient configuration

The application allows for the creation of patients and their assignment to various scenarios. These patients are then displayed in the simulation within a 3D environment where they can be examined and treated. When creating a patient, the appearance of the 3D avatar can be determined (body proportions, skin tone, hair, clothing; randomized or manually), and injury patterns can be defined and assigned to affected body parts. Visible injuries such as hemorrhages or burns are displayed on the patient's skin and may generate dynamic blood pools and bleeding effects. Injuries are categorized by type (laceration, stab wound, burn, pneumothorax, organ injuries, etc.), each of which can be selected in varying degrees of severity (mostly mild, moderate, and severe, or classified according to specific standards such as the Organ Injury Scale). Furthermore, for clinical scenarios, it is possible to add pre-treatments performed by virtual first responders at the scene or during transport (for scenarios where only the clinical part is to be simulated). The resulting combinations of measures can be seen in Figure 7.

Visual representation of medical interventions. To provide realistic visual feedback during training, performed measures are dynamically rendered on the 3D avatar. The central figure illustrates a composite view of various possible treatment measures applied simultaneously. Arround it is a selection of available treatments, ranging from vascular access and monitoring to advanced trauma life support equipment. Reprinted with permission of nVista technologies GmbH, © nVista technologies GmbH.

A digital illustration of a human figure showcasing various medical support devices. The central figure is covered with multicolored sections and is equipped with a face mask connected to a ventilator bag, an IV drip, and a variety of braces and supports on limbs and neck. Surrounding the main figure are detailed close-ups of a breathing mask, arm brace, neck collar, abdominal support, and leg brace, demonstrating each device's design and application. The background is a plain gray.

The physiology model also simulates body position, skin characteristics, and other assessable properties of the patient and their effects. A deteriorating condition causes the patient to first sit down and eventually lie down. This is visually represented in the 3D avatar during the simulation, showing corresponding changes in posture and skin color (becoming pale or turning blue/cyanotic), which can be recognized by the participant.

 Materials configuration

In addition to limited personnel resources, the availability of materials and equipment represents one of the major challenges in an MCI situation. Consequently, the application includes an inventory system that can be stocked individually. This allows for the definition and stocking of various containers, cabinets, or backpacks for the pre-hospital phase. Users can create their own combinations of supplies tailored to the specific mission or deployment site from dozens of available items. During treatment, participants must identify and select the correct item from the appropriate backpack pocket or cabinet. This system can also be deactivated if desired.

 Role and permission configuration

Each participant must be assigned a role. The simulation can be conducted assuming roles such as Emergency Paramedic, Emergency Physician, Hospital Physician, etc. The qualification associated with the role influences both the equipment carried (different backpacks) and the authorization to perform specific medical interventions. It is also possible to configure the system to permit all actions for all participants.

 Training

The core application is the training session itself. Prior to training, a role is selected or assigned. In a group exercise, the instructor's task involves moderating the scenario, providing assistance, and observing the participants. Following the training, these observations should be discussed alongside the system recordings.

During training, participants can move across the 2D overview map (Figure 6) or within the third-party scene simulation tool to view the situation and the patients. Upon approaching a patient, they can select them. This opens the 3D view, where the actual examination and treatment takes place. The 3D environment differs depending on the scenario and patient location. For instance, the clinical setting displays a treatment room or hospital corridor (right side of Figure 8), whereas the pre-hospital setting places the user at the accident scene (left side of Figure 8). A virtual clipboard displays the initial impression of the patient and subsequently presents examination results throughout the course of treatment.

User interface of the patient simulation in different operational environments. The image compares the 3D visualization of a patient in a pre-hospital scenario (left) vs. an in-hospital emergency room scenario (right). Reprinted with permission of nVista technologies GmbH, © nVista technologies GmbH.

A group of medical professionals is gathered around a hospital bed with a realistic training mannequin equipped with wires and sensors. In the foreground, a person is holding a tablet displaying a monitoring interface.

Clicking on the patient's body parts opens relevant examination or treatment menus, categorized according to the (X)ABCDE scheme. By explicitly examining individual body parts, the participant can obtain information allowing for conclusions regarding perfusion or internal injuries. For example, a fracture limits the range of motion of a limb, internal abdominal bleeding leads to increasing muscular guarding, and an airway obstruction results in abnormal respiratory sounds.

The configured injury types and severities strongly influence which treatment options are viable. While the system does not directly forbid or prevent incorrect treatments, an incorrect intervention may result in low efficacy within the physiology model. The evaluation of the action is the responsibility of the instructor or the participants themselves and should be discussed during the debriefing phase. Hemorrhages, for instance, differ by severity and effective treatment methods (dependent on the severity and type of bleeding, visualized via bleeding effects and skin textures) and the affected body part. A minor laceration on the forearm bleeds significantly slower and less voluminously than a minor laceration on the thigh, due to lower perfusion in the affected vessel. Furthermore, a severe laceration cannot be effectively treated with a simple adhesive dressing; a moderate wound would only be slowed by such a dressing depending on the body part, whereas a minor laceration could be brought under control.

All actions in this view have associated treatment times that must elapse when a measure is applied (these can be deactivated depending on the scenario). Subsequently, the measure is visually represented on the avatar, and the effect is applied to the physiology model. If the measure consumes materials, these must be selected from the available containers or requested from adjacent rooms. In the clinical setting, laboratory values or imaging modalities can also be requested here. Upon completion of examination or treatment, a triage category can be assigned, and the patient can be cleared for transport or handed over to the appropriate ward/operating theater.

 Debriefing and evaluation

The training session is followed by debriefing and evaluation. The system itself does not provide any judgmental scores but exclusively presents objective facts, which must then be discussed and interpreted by participants and instructors. All actions taken by participants, their timestamps, and the concurrent physiological parameters of patients are logged and displayed. This aims at delivering a profound database to interpret, discuss and facilitate suggestions for improvement that are linked specifically to any given situation experienced in the simulation. Hence, all aspects of the simulation session could be discussed, such as the rationale and sequencing of measures performed, the scope of medical care, and the decisions made.

 Conducted field tests

The field tests conducted with the D2PuLs system so far were designed to get insights regarding the methodological feasibility of the simulation approach, the usability of the D2PuLs system according to its current Technology Readiness Level [TRL, see Ref. (18)] as well as the user experience of the simulation training in pre-hospital and clinical emergency care contexts. The primary objective was to examine the system's capacity to support different participant groups and instructional formats while collecting empirical data for iterative development.

Study design

Three one-day field trials were conducted in 2025, covering different stages of technological maturity [TRL 6–7, see Ref, (18)]. Two trials focused on pre-hospital emergency care, conducted in a vocational school for paramedics and a training ambulance station, while one trial was conducted in a clinical setting within the emergency department of a hospital.

A mixed-methods design was employed to all trials to capture both quantitative and qualitative data. Data collection included one structured online questionnaire, a guided debriefing discussion with metaplan card evaluation and the observation of participant behavior by members of the research team during simulation exercises. Based on these methods a triangulation of participant feedback and observational data was done with a focus on evaluating the usability of the D2PuLs system and the user experience of the simulation training.

Demographic and baseline information was collected prior to the exercises, including age, professional qualifications, prior exposure to simulation-based training, and self-assessed technology readiness using the short scale of Neyer et al. (19). This facilitated stratification of participants and helped to interpret participant feedback and observational data regarding the interaction with the system and the simulation training overall.

Each field test followed a standardized workflow consisting of five phases:

Introduction—Overview of the project, system objectives, and trial workflow presented via lecture and demonstration.

Baseline Assessment—Collection of participant characteristics and prior experience through online surveys.

System Familiarization—Guided introduction to the simulation system, including scenario overview, navigation, and hands-on interaction with the patient simulation modules.

Simulation Exercise—Participants engaged in one or two simulation runs, either individually or in groups, with systematic observation and real-time logging of actions, decision points, and physiological responses of virtual patients. In group settings, roles were assigned to reflect operational hierarchies and functional divisions such as triage, treatment, and coordination. Some trials incorporated “patient copies” to enable individual participants to treat identical patient profiles concurrently, facilitating comparative analysis of treatment strategies. Hybrid scenarios integrating digital simulation with physical manikins were implemented in clinical trials to provide haptic feedback and enhance situational realism.

Debriefing and Evaluation—Structured reflection facilitated through guided discussions and metaplan card exercises. Metaplan card queries were used to gather general insights from the participants' perspective, utilizing positive (green) and negative (red) cards. The survey content covered all system areas, such as the 3D avatar representation, UI and UX, avatar reactions to interventions, the visualization of performed measures, and integrated injury patterns. These insights were subsequently clustered and discussed in a structured debriefing.

Pre-hospital training scenarios emphasized situational assessment, primary triage, and initial patient care, whereas clinical scenarios concentrated on hospital triage, trauma room management, and integration of pre-treated or self-referred patients.

 Pre-hospital training configurations

Pre-hospital exercises were conducted in both individual and group modes, with and without third-party scene simulation. The group size for pre-hospital trial groups ranged from 15 to 25 participants. Individual training required participants to independently perform scene assessment, triage, and initial patient care. In the conducted group training sessions, participants were assigned different roles, and the situation was managed collaboratively based on a division of labor. Tasks included those of the first-arriving unit in the disaster zone, primary triage, and the establishment and operation of Casualty Collection Points. In the group mode with copies of patients, every participant treated a copy of the exact same patients within the same scenario. This setting was particularly effective for making differences in triage results and emergency care visible, facilitating subsequent discussion.

 Clinical training configurations

Regarding the training of clinical staff, the conducted trial involved 13 participants representing nursing and medical staff. Clinical training was also tested in both single-player and group modes. Group size of the clinical trial conducted in 2025 was 13 participants. The digital 3D situation simulation component was not utilized in the tested settings, as the training focus lay on the tasks and challenges of clinical triage and the handover to further medical care (e.g. trauma, operating, treatment rooms). However, to adequately account for the spatial conditions of an emergency department, the specific floor plan of the training hospital was integrated into the 2D situation simulation. This representation allows for the distribution and spatial assignment of incoming patients to various work zones, enabling the direct visualization and comprehension of limited spatial resources and overcrowding situations in specific areas.

Using the physiology model patient health states were configured corresponding to the moment of arrival at the hospital. The system allows the configuration of patients arriving pre-treated by EMS personnel as well as untreated (e.g., self-referrals/walk-ins). The physiology model reflects the patient's condition according to measures performed or omitted during prior phases.

By connecting the D2PuLs software platform to haptic components—such as Gaumard patient manikins—a hybrid in-situ simulation training setting was created (Figure 9).

Hybrid in-situ simulation training. The setup demonstrates the integration of the D2PuLs digital platform with a high-fidelity Gaumard manikin. The tablet in the foreground acts as the connection to the simulation engine, calculating the physiology model and visualizing the incident scene in real-time. It drives the patient's dynamic condition, while the interdisciplinary team performs physical interventions on the simulator, creating a seamless link between digital decision-making and haptic skill application. Reprinted with permission of Nuremberg Hospital, South Campus, © Nuremberg Hospital, South Campus.

A split image shows two scenarios. On the left, a person is lying on a road beside a monitor with highway signs and a speed limit visible. On the right, a patient lies on an operating table in a medical room, with surgical equipment around and a visible pool of blood on the floor. Each scene displays a user interface with icons representing medical processes, such as diagnostics and therapy. Results

The conducted field tests generated quantitative questionnaire data and qualitative feedback from structured debriefings in both pre-hospital and clinical training configurations as described in Section 2.5. The empirical results reported below are based on the aggregated questionnaire datasets from the pre-hospital field tests (EMS context) and the clinical trial (emergency department context).

Sample characteristics and baseline exposure

Across the pre-hospital field tests, 61 participants completed the questionnaire. The sample was predominantly young: 77.0% (47/61) were aged 21–30 years, while 9.8% (6/61) were ≤ 20 years and 8.2% (5/61) were 31–40 years. In terms of professional background, the pre-hospital sample consisted mainly of paramedic trainees (Auszubildende/r Notfallsanitäter/in) (77.0%, 47/61), complemented by EMT-level staff (Rettungssanitäter/in) (11.5%, 7/61), qualified paramedics (Notfallsanitäter/in) (6.6%, 4/61), and first responder roles (Einsatzsanitäter/in/Sanitätshelfer/in) (4.9%, 3/61). Reported EMS work experience was heterogeneous; 32.8% (20/61) indicated no prior experience, while 26.2% (16/61) reported 2–5 years and 19.7% (12/61) < 2 years.

In the clinical trial, 13 participants provided baseline questionnaire data. The age distribution shifted toward older groups: 46.2% (6/13) were 31–40 years, and 23.1% (3/13) each were 41–50 years and 51–60 years. The professional composition covered nursing and medical staff, with 46.2% (6/13) reporting nursing qualifications with specialist training in emergency/anesthesia/intensive care, 7.7% (1/13) general nursing, 23.1% (3/13) residents, and 23.1% (3/13) specialists/consultants. Clinical work experience clustered in higher ranges: 30.8% (4/13) reported 11–20 years and 30.8% (4/13) >20 years.

Prior exposure to relevant triage training differed between settings. In the pre-hospital sample, participation in (pre-)triage training was common: 50.8% (31/61) reported having attended such training once and 26.2% (16/61) multiple times; 23.0% (14/61) reported none. In the clinical sample, 46.2% (6/13) reported no prior clinical triage training, while 23.1% (3/13) had attended once and 30.8% (4/13) multiple times (Table 2).

Sample characteristics and baseline exposure (pre-hospital vs. clinical).

Variable Pre-hospital field tests Clinical trial
Questionnaire responses (n) 61 13
Age ≤ 20 years 6 (9.8%) 0 (0.0%)
Age 21–30 years 47 (77.0%) 1 (7.7%)
Age 31–40 years 5 (8.2%) 6 (46.2%)
Age 41–50 years 3 (4.9%) 3 (23.1%)
Age 51–60 years 0 (0.0%) 3 (23.1%)
Paramedic trainees 47 (77.0%) —
EMT 7 (11.5%) —
Paramedics 4 (6.6%)
First responders 3 (4.9%)
Nursing — 1 (7.7%)
Nursing with specialist training (ED/ICU/anesthesia) — 6 (46.2%)
Residents — 3 (23.1%)
Senior Physicians/Specialists — 3 (23.1%)
Prior triage training: none 14 (23.0%) 6 (46.2%)
Prior triage training: once 31 (50.8%) 3 (23.1%)
Prior triage training: multiple times 16 (26.2%) 4 (30.8%)
Overall technology readiness (M ± SD) 2.19 ± 1.08 1.87 ± 1.43

Technology readiness (short scale; lower values indicate higher readiness) was above the neutral midpoint in both settings. The aggregated technology readiness score was M = 2.19, SD = 1.08 (n = 61) in the pre-hospital sample and M = 1.87, SD = 1.43 (n = 13) in the clinical sample. Subscale means in the pre-hospital group were technology acceptance M = 2.21 (SD = 1.17), technology competence M = 1.78 (SD = 0.84), and technology control belief M = 2.57 (SD = 0.94). Corresponding clinical subscale means were acceptance M = 1.95 (SD = 1.41), competence M = 1.58 (SD = 1.20), and control belief M = 2.08 (SD = 0.97).

 Previous use of simulation techniques (baseline)

Participants in both settings reported that low-technology or analog formats were the most frequently used simulation techniques in their prior training history. In the pre-hospital sample, “human mime” and paper patient cards were rated as used “frequently” on average (means below 3 on a 1–5 frequency scale where lower values indicate more frequent use), whereas the use of smartphones, tablets, laptops, and VR headsets was rated close to “almost never.” High-fidelity manikins were also reported as rarely used. In the clinical sample, paper patient cards were rated as the most frequently used technique, while VR headsets were not used. Compared with the pre-hospital sample, tablets and high-fidelity manikins showed greater variance in the clinical dataset, reflecting mixed exposure across clinical roles.

 Training evaluation outcomes

Training outcomes were assessed using four global ratings covering overall appraisal, individual learning gain, technical implementation, and organizational process. In the pre-hospital dataset, ratings were consistently in the favorable range, with the lowest (most favorable) mean observed for organizational process (M = 2.03, SD = 0.66, n = 61) and the highest mean for self-assessed learning gain (M = 2.72, SD = 0.70, n = 61). In the clinical dataset, item-level response counts were n = 12, indicating one missing response for these evaluations. Clinical participants rated the training overall as M = 1.80, SD = 0.80 (n = 12) and the organizational process as M = 1.60, SD = 0.80 (n = 12). The technical implementation item showed a higher mean in the clinical dataset (M = 2.60, SD = 0.90) than in the pre-hospital dataset (M = 2.42, SD = 0.83), while the learning gain item was lower in the clinical dataset (M = 2.20, SD = 1.00) than in the pre-hospital dataset (M = 2.72, SD = 0.70). Across both settings, variability (SD) was moderate, suggesting inter-individual differences in perceived learning outcomes and in the evaluation of technical aspects (Table 3).

Training evaluation ratings (mean ± SD; lower values indicate more favorable ratings).

Outcome (5-point rating scale) Pre-hospital field tests Clinical trial
Overall rating of the training 2.20 ± 0.64 (n = 61) 1.80 ± 0.80 (n = 12)
Self-assessed learning gain 2.72 ± 0.70 (n = 61) 2.20 ± 1.00 (n = 12)
Technical implementation 2.42 ± 0.83 (n = 61) 2.60 ± 0.90 (n = 12)
Organizational process 2.03 ± 0.66 (n = 61) 1.60 ± 0.80 (n = 12)
Qualitative feedback themes from debriefings (metaplan clustering)

Qualitative feedback obtained during the structured debriefings converged on four recurring domains across settings: (i) realism and dynamic behavior of patients, (ii) user interface and usability, (iii) functional scope and clinical logic of available measures, and (iv) didactic guidance and onboarding needs. Participants highlighted the relevance of observing time-stamped actions and patient reactions for subsequent discussion during debriefing, consistent with the evaluation concept described in the training workflow. Across clusters, comments indicated that the breadth of available measures and the navigation logic created an initial operational load, while participants also reported that the system outputs supported structured reflection by making actions and temporal sequences explicit.

Overall, the empirical results of the field tests produced valuable usability- and training-related feedback in both pre-hospital and clinical contexts. The datasets also document differences between participant populations (age, professional background, and experience) and provide a baseline for subsequent iterative development and targeted onboarding measures aligned with the workflow of the field tests.

 Discussion

The D2PuLs system tries to unify pre-hospital and clinical simulation systems into an adaptable platform, allowing simulations of MCI scenarios to be conducted on a large number of end devices. Through the use of a physiology model, patient injury progressions are simulated in real time and communicated to participants via a 3D representation or connected manikins in hybrid simulation settings. Patients can be individually created from over 20 injury types so far, allowing for examination and treatment by participants. Equipment, scenarios, and participant roles can be configured. Simulations can take place in group or individual training sessions, and it is possible to replace the integrated minimal 2D scene simulation with external 3D scene simulators.

The field tests conducted so far primarily served to gather feedback regarding the usability and the maturity level of the system as well as the user experience of the simulation training itself. They showed that the component-based approach of D2PuLs is methodologically feasible for various teaching and learning settings enabling cross-sectoral and interprofessional training and different learning objectives. However, up to now testing was primarily conducted on small scale MCI scenes, where the number of participants and simulated patients was below 20 each. While these numbers seem sufficient for evaluating the system's usability during development, operational deployment may require managing significantly larger numbers. The system's behavior during larger major incidents, such as an MCI-100 (involving up to 100 casualties), has not yet been tested. Additionally, no cross-sectoral trials—specifically the “live handover” of pre-treated patients to a hospital—have been conducted to date.

Although the technical and functional development of the D2PuLs training system is well-advanced, there remain areas that are not fully elaborated. For instance, participant onboarding requires further development. Due to the large number of diagnostic and treatment measures available, there is a high degree of complexity in the menu navigation, which, according to previous observations, complicates immediate entry into the simulation training.

Consequently, concepts to improve operational understanding include demo videos and the implementation of a tutorial scenario, in which a training session can be played through with digital guidance. In this tutorial environment, pop-ups and color-coded cues serve to introduce the user to the respective operating concepts step by step. On the instructor's side as well, the system's functional scope creates significant complexity, necessitating corresponding training measures or concepts. Video tutorials or Train-the-Trainer courses seem to be conceivable solutions to address this.

Following the results of the conducted field tests so far, the modular architecture of the D2PuLs system and its associated configuration capabilities seems to be the right approach to establish the D2PuLs platform as a digital foundation that can be expanded to cope the needs of different settings. For instance, the simulation can be augmented with external components, such as third-party scene simulators, physical simulation manikins, or printed triage algorithms.

A challenge that could be well-addressed through the system's configuration capabilities is to cope with regionally different concepts for command and control guidelines, protocols, algorithms, treatments, competencies, authorizations, and equipment standards, all of which are crucial for realism. Mapping these regional differences fosters the trainees' identification with the simulation (“train as you fight”) but also generates complexity that must be managed. For example, a range of distinct triage algorithms are employed across different countries and federal states. While depositing these algorithms as visual aid panels for the user presents no challenge, the algorithms themselves require different medical measures, which must then be implemented within the patient simulation and the physiology model. Furthermore, these local algorithms often reference national or international standards, such as the German S3 Guideline on “Polytrauma/Severe Injury Treatment” (20) or the S2k Guideline “Disaster Medical Pre-hospital Treatment Guidelines” (21), which are revised at regular intervals.

The configuration of various inventory containers also remains a point of discussion. The field tests revealed that efficient material and equipment handling is an issue. On one hand, a new instructor or operator should not have to spend the first few hours filling backpacks and cabinets with the materials typically found in the specific organization or region. On the other hand, the exact composition of these containers is difficult for outsiders to ascertain, let alone keep up to date. The same applies to the permissions associated with different roles. A conceivable remedy to this could be a community-populated store, similar to the “Workshop” concept found on gaming platforms like Valve's Steam. This would allow inventory configurations created by participants to be shared via the workshop for others to use. Participants in the same region or with similar requirements could then download these loadouts and deploy them in their own scenarios without creating everything from scratch. These considerations are currently in a very early conceptual phase.

Furthermore, there is still a substantial backlog of improvements and optimizations on the software side. The physiology model is continuously optimized to reduce performance requirements and allow more models to run in parallel on the same device. Current research focuses on enabling the simulation of dozens of models in parallel on a single laptop. While the model enables realistic patient simulation with high degrees of freedom regarding treatment measures—making the consequences of these actions tangible through the dynamic representation of health status—it is highly complex and demanding regarding validation and further development. The 3D treatment environments are also currently under development, receiving continuous updates with new visualizations of interventions, improved animations, and reduced memory usage. Feedback regarding UI and usability gathered from regular field tests with project partners is continuously incorporated into the user interface.

Field tests have highlighted that communication skills represent a critical key competency for participants. To specifically foster this skill and further reduce exercise artificiality, future development focuses on two new systems. First, patient communication supported by Large Language Models (LLM) is intended to reduce reliance on pre-recorded inputs (injects) and provide a consistent communication component [similar to the study by Ref. (22)]. Instead of receiving information via button press, participants must verbally obtain the patient's history (e.g., using the SAMPLER scheme), to which the virtual patient responds dynamically and appropriately according to their physiological state. Second, an integrated audio system is being developed that utilizes simulated radio channels and spatial voice rendering (“Proximity Voice”) to realistically map complex team communication in stressful or decentralized situations, thereby significantly increasing immersion.

In summary, the conducted field tests and the achieved level of maturity of the D2PuLs system indicate that the approach possesses the potential to bridge the existing gap between theoretical education, paper-based simulation and resource-intensive full-scale field exercises. The combination of physiological simulation and scene visualization enables high-frequency digital training of Mass Casualty Incidents at a depth that was previously difficult to represent “in silicio.” Although technical advancement and the reduction of complexity remain challenges, the modular and cross-sectoral approach provides a forward-looking foundation for modern educational concepts. With the planned integration of AI technologies, the degree of realism will increase further, enabling the system to make a sustainable contribution to strengthening the operational confidence of responders and, ultimately, optimizing patient care in real major incidents.

 Data availability statement

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

 Ethics statement

Written informed consent was obtained from the individual(s) for the publication of any potentially identifiable images or data included in this article.

 Author contributions

RK: Conceptualization, Resources, Validation, Writing – original draft. TP: Conceptualization, Resources, Validation, Writing – original draft. LE: Conceptualization, Software, Visualization, Writing – original draft. AB: Software, Writing – review & editing. CBr: Software, Writing – review & editing. TR: Data curation, Resources, Validation, Writing – review & editing. SP: Data curation, Resources, Validation, Writing – review & editing. CE: Data curation, Resources, Validation, Writing – review & editing. LB: Software, Visualization, Writing – review & editing. MS: Funding acquisition, Project administration, Supervision, Writing – review & editing. CBa: Funding acquisition, Project administration, Supervision, Writing – review & editing, Writing – original draft.

 Conflict of interest

LE, AB, CB, MS, and CB were employed by nVista technologies GmbH.

The remaining author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

 Generative AI statement

The author(s) declared that generative AI was used in the creation of this manuscript. The author(s) declare that Gen AI was used in the creation of this manuscript. Gemini 3 Pro and ChatGPT was used for correction of grammar, spelling, sentence structure, and translation. No citations were introduced by Gemini or ChatGPT, and no diagrams or plots were created by Gemini or ChatGPT. The author(s) retain responsibility for all content.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

 Publisher's note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

 References Bauer C Loose T. Digital gestützte Simulationstrainings in der rettungsdienstlichen und notfallmedizinischen Aus-, Fort- und Weiterbildung. In:Prescher T Bauer C Dubb KH Hofmann T Koch S, editors. Rettungswissenschaft. Stuttgart: Kohlhammer (2023). p. 323–41. Deutsches Rotes Kreuz. Durchführung und Auswertung von MANV-Übungen. Hg. v. Deutsches Rotes Kreuz (2016). Available online at https://www.drk.de/forschung/schriftenreihe/schriften-der-forschung-band-iii/ (Accessed November 21, 2025). Kreisfeuerwehrverband Aschaffenburg Landkreis-Feuerwehren üben für Unfälle mit vielen Verletzten (2025). Available online at: https://www.kfv-ab.de/index.php/news/nachrichten/nachrichten/landkreis-feuerwehren-ueben-fuer-unfaelle-mit-vielen-verletzten (Accessed January 24, 2026). Deutsches Rotes Kreuz. DRK Lern Campus (2020). Available online at https://www.drk-lerncampus.de/news/manv/ (Accessed January 24, 2026). Deutsche Interdisziplinäre Vereinigung für Intensiv- und Notfallmedizin. MANV-Training (Dynamische Patientensimulation) (2020). Available online at https://akademie.divi.de/fort-weiterbildung/2295-trainer-dynamische-man-simulationen/ (Accessed January 24, 2026). Schütte M Hartart F. Fragebogengestützte Evaluation von Übungen mit Massenanfall von Verletzten (MANV). Notfall Rettungsmed. (2019) 22:522–7. doi: 10.1007/s10049-019-0584-2 Mentler T Jent S Herczeg M. Ein interaktives Trainingssystem zur Nutzung mobiler computerbasierter Werkzeuge bei rettungsdienstlichen Großeinsätzen. In:Grandt M Schmerwitz S, editors. Ausbildung und Training in der Fahrzeug- und Prozessführung. Bonn: DGLR-Bericht (2013). p. 103–18. Available online at https://www.dglr.de/publikationen/2013/53801010.pdf (Accessed November 20, 2025). García Ulerio J Al Khatib M Aammar B Ragazzoni L Barone-Adesi F Caviglia M. Simulation technology use in disaster medicine education and training: a scoping review. Front Disaster Emerg Med. (2025) 3:1636285. doi: 10.3389/femer.2025.1636285 Zhang Z Meybodi MM Ingale A Karimova L Vinnikov M. Extended reality technology for emergency medical service training: systematic review. Front Disaster Emerg Med. (2025) 3:1630167. doi: 10.3389/femer.2025.1630167 Ausbildungs- und Prüfungsverordnung für Notfallsanitäterinnen und Notfallsanitäter (NotSan-APrV). Bundesgesetzblatt Teil I. (2013) 2013:4280–304. Ausbildungs- und Prüfungsverordnung für Notfallsanitäterinnen und Notfallsanitäter (NotSan-APrV) BGBl. I S (2023). 4280. last amended by Article 12 of the Decree of June 7, 2023 (BGBl. 2023 I Nr. 148). Berlin: Bundesministerium der Justiz. Available online at: https://www.gesetze-im-internet.de/notsan-aprv/ (Accessed November 18, 2025). Bauer C Loose T. Simulationsbasierte Trainings für Großschadenslagen mithilfe der Digitalen Dynamischen Patienten- und Lagesimulation (D2PuLs). Intensiv Notfallbehandlung. (2023) 48:31–8. doi: 10.5414/IBX00600 Barte J Schubert P. Engagement im Ernstfall – Zahlen und Fakten zu freiwlligem Engagement und zivilgesell- schaftlichen Organisationen im Bevölkerungsschutz. Berlin: ZiviZ im Stifterverband (2025). Available online at https://www.bbk.bund.de/SharedDocs/Downloads/DE/Mediathek/Publikationen/Foerderung-Ehrenamt/sonderauswertung-ziviz_download.pdf?__blob=publicationFile&v=3 (Accessed January 24, 2026). Kolb DA. Experiential Learning: Experience as the Source of Learning and Development. Upper Saddle River, NJ: FT Press (2014). Rall M Jaki C Kolbe M Sandmeyer B Niederlich L Grosch S ., Simulationsbasiertes Lehren und Lernen im Gesundheitswesen, Positionspapier der Deutschen Gesellschaft zu Förderung der Simulation und der Medizin e. V. (DGSiM). Hildesheim: Deutsche Gesellschaft zur Förderung der Simulation in der Medizin e.V. (2025). Available online at: https://dgsim.de/services/positionspapier/ (Accessed November 21, 2025). Heller AR Hähn F Brüne F Wurmb T Franke A Kowalzik B. Ergebnisse der Sichtungs-Konsensus-Konferenzen und Glossar Sichtung. Notfall Rettungsmed. (2025). doi: 10.1007/s10049-025-01547-y Hofmann J. Ein Physiologiemodell für Tactical Combat Casualty Care Training in mobilen Serious Games. Wiesbaden: Springer Fachmedien (2020). doi: 10.1007/978-3-658-30202-3 Armstrong JR. Applying technical readiness levels to software: new thoughts and examples. INCOSE Int Symp. (2010) 20:838–45. doi: 10.1002/j.2334-5837.2010.tb01108.x Neyer FJ Felber J Gebhardt C. Entwicklung und Validierung einer Kurzskala zur Erfassung von Technikbereitschaft. Diagnostica. (2012) 58:87–99. doi: 10.1026/0012-1924/a000067 Deutsche Gesellschaft für Unfallchirurgie (DGU). S3-Leitlinie Polytrauma/Schwerverletzten-Behandlung [S3 Guideline on the Treatment of Patients with Severe and Multiple Injuries]. Version 4.0. Berlin: DGU; AWMF Register No. 187-023 (2022) (German). Available online at https://www.awmf.org/leitlinien/detail/ll/187-023.html (Accessed November 19, 2025). Deutsche Gesellschaft für Anästhesiologie und Intensivmedizin (DGAI). Katastrophenmedizinische prähospitale Behandlungsleitlinien [Disaster medicine prehospital treatment guidelines]. Version 1.0. Nuremberg: DGAI. AWMF-Register Nr. 001-043 (2023) (German). Available online at https://register.awmf.org/de/leitlinien/detail/001-043 (Accessed November 24, 2025). Cook DA Overgaard J Pankratz VS Del Fiol G Aakre CA. Virtual patients using large language models: scalable, contextualized simulation of clinician-patient dialogue with feedback. J Med Internet Res. (2025) 27:e68486. doi: 10.2196/6848639854611

Edited by: Andreas Follmann, University Hospital RWTH Aachen, Germany

Reviewed by: Federico Merlo, University of Eastern Piedmont, Italy

Tamorish Kole, University of South Wales, United Kingdom