The Use of Artificial Intelligence in Disaster Medicine

Einsatz von Künstlicher Intelligenz in der Katastrophenmedizin

Terence T. Huang^a, Ted C.-Y. Chang^b, Chung-Hsi Hung^b, Tzu-Hsiang Yang^b

^a Quanta Research Institute, Taoyuan City, Taiwan

^b Quanta Computer Inc., Taoyuan City, Taiwan

Summary

As the world faces an increasing number of natural and human-made disasters, the challenges of managing and responding to their impacts are becoming more pressing and complex than ever. At the same time, technological advancements – particularly in semiconductors, machine learning, and breakthroughs in artificial intelligence – have progressed at an unprecedented pace over the past decade. When these advanced technologies are applied to humanitarian purposes with benevolent intent, such challenges may, in fact, present opportunities to enhance the resilience, efficiency, and sustainability of essential public services. This article provides an illustrative perspective on how AI and complementary technologies can converge to offer practical applications in disaster medicine.

Keywords: disaster medicine; disaster management; AIoT; AI medicine; resilient medicine; resilient healthcare

Zusammenfassung

Da die Welt mit zunehmenden natur- und menschgemachten Katastrophen konfrontiert ist, werden die Herausforderungen bei der Bewältigung und Reaktion auf deren Folgen immer dringlicher und komplexer. Gleichzeitig haben technologische Fortschritte – insbesondere in Halbleitern, im maschinellen Lernen und bei Durchbrüchen in der künstlichen Intelligenz – in den letzten zehn Jahren ein beispielloses Tempo erreicht. Wenn diese fortschrittlichen Technologien für wohltätige Zwecke eingesetzt werden, können sich diese Herausforderungen sogar als Chancen erweisen, die Resilienz, Effizienz und Nachhaltigkeit wichtiger öffentlicher Dienste zu verbessern. Dieser Artikel bietet eine anschauliche Perspektive darauf, wie KI und ergänzende Technologien zusammenwirken können, um praktische Anwendungen im Katastrophenmanagement zu ermöglichen.

Schlüsselwörter: Katastrophenmedizin, Katastrophenmanagement, AiOT, Künstliche Intelligenz, KI-Medizin, resiliente Medizin, resiliente Gesundheitsversorgung

Introduction

As catastrophic incidents and geopolitical conflicts become increasingly prevalent, international communities have collectively recognized the growing need for mitigation, preparedness, response, and recovery measures for disasters [5]. The United Nations Office for Disaster Risk Reduction (UNDRR) has therefore established the Sendai Framework for Disaster Risk Reduction2015–2030, while the World Health Organization (WHO) has released the Health Emergency and Disaster Risk Management Framework, which emphasizes the continuum of care during turbulent times. Moreover, the COVID-19 pandemic – along with ensuing armed conflicts and escalating risks across multiple regions, posing tremendous threats to civilian lives and infrastructure – has further underscored the need for systematic resilience in healthcare systems and practices.

Disaster medicine is a vital discipline dedicated to maximizing life-saving outcomes and minimizing public health impacts while maintaining the provision of standard care for all. However, it is often confronted with formidable constraints, hazardous conditions, and humanitarian challenges in some of the most austere environments. It requires multidisciplinary collaboration and coordination of expertise across emergency medicine, prehospital medicine, public health, mental health, disaster management, and related fields. It encompasses new hospital building standards [4], contingency planning, logistics, transport, communication, and support for critical decision-making. Given the likelihood of a surge in overwhelming casualties and increased demands for medical care, disaster medicine must optimize the use of scarce human and medical resources [7] while anticipating potential disruptions to existing health systems and centralized services. As a result, it is imperative to evolve toward a more resilient mindset and decentralized approaches within the context of disaster medicine, ensuring that the transition from preparedness to response is as minimally disruptive as possible.

The advent of breakthroughs in artificial intelligence (AI), accelerated by rapid advancements in ultra-nanoscale semiconductors and novel machine learning (ML) algorithms, may illuminate this goal. However, it is essential to acknowledge that AI alone does not provide a solution; it requires complementary technologies and extensive research and development efforts, as well as collaboration among interdisciplinary partners and experts who share values and visions.

The following section presents the historical context of related research and collaborative developments that have contributed to advancing AI and supporting technologies for potential use in disaster medicine.

History

The world’s first AI Lab was established at the Massachusetts Institute of Technology (MIT) in 1970 by Prof. Marvin Minsky, one of the founding fathers of AI and a pioneer in neural network research. The AI Lab later merged with the Laboratory for Computer Science to form today’s Computer Science and Artificial Intelligence Laboratory (CSAIL). Beginning in 2005, with support from Quanta Computer (Quanta) – one of the world’s largest computing hardware companies – a series of research projects (“T Party” Project, Project Qmulus, AI Medicine, and Project AIR) were conducted at CSAIL in collaboration with Quanta Research Institute (QRI). Over the course of two decades, this collaboration has fostered advanced research in human-centric computing (e.g., speech interaction, natural language processing, and the Internet of Things for ubiquitous computing), cloud computing, machine learning, and, more recently, AI in medicine. The early visions of the research laid the foundations for digital transformation and world-changing paradigm shifts across consumer and enterprise domains that emerged a decade later.

In the 2010s, Quanta was among the first to partner with leading tech companies to develop high-performance data center hardware, enabling the proliferation of cloud services and big data. Around the same time, a momentous breakthrough in deep learning – achieving highly accurate predictive AI through deep neural networks (DNNs) – was realized by Prof. Geoffrey Hinton, recipient of the 2024 Nobel Prize, and his colleagues. It has consequently accelerated advancements in generative pre-trained transformer (GPT) innovations. With the first GPU-based AI supercomputer built by Quanta in partnership with NVIDIA, companies such as OpenAI began training large-scale GPT models on internet-scale data, ushering in a new era of innovative generative AI (GenAI) applications capable of human-like interaction, large-scale multimodal information processing, and high-quality creative content generation.

In the medical domain, computing made its debut in 1961 when IBM installed the first electronic health record (EHR) system at Akron Children’s Hospital in Ohio [6]. Since then, it has taken decades for digital innovations to have transformative effects in the medical field, given its highly regulated nature. Ironically, when the catalytic event of COVID-19 struck the world in late 2019, the adoption of digital health applications – such as telemedicine, telehealth [10], and AIoT (Artificial Intelligence of Things) [3] – finally began to accelerate.

Near the end of the pandemic, armed conflicts in Ukraine and the Middle East, along with major earthquakes, floods, and wildfires, impacted various parts of the world. The escalation of these threats has drawn the attention of governments and institutions toward more proactive approaches – planning and piloting innovative uses of 5G, IoT, and drone technologies in disaster medicine for scenarios such as managing mass-casualty incidents [9], enhancing disaster response [8], and enabling remote telemedicine and patient monitoring for casualties [1].

In Taiwan, an island situated along the Pacific Ring of Fire and a focal point of geopolitical tension, the Presidential Office also established special task force committees in 2024 to advance national resilience initiatives, including those focused on healthcare and emergency response. Quanta has been a primary contributor, providing a comprehensive framework of AI-ready smart healthcare solutions that have been adopted by leading medical centers in Taiwan and piloted in multiple field deployments for prehospital and hospital emergency response scenarios.

The objectives and foundations of this framework will be further examined for their practical applications in disaster medicine management.

Practical Management

Objectives and the Key Enabler

The overarching goal of utilizing AI and complementary technologies in disaster medicine is to enable a seamless continuum of care despite constraints in human and medical resources, while ensuring efficient, accurate, and high-quality care for both incoming casualties and existing patients. Secondly, it extends the coverage of care management to prehospital and unconventional settings, such as remote sites, field hospitals, and underground shelters. Thirdly, it enhances the resilience, efficiency, and sustainability of emergency care operations through comprehensive, data-driven decision support.

As established in the Smart and Inclusive Healthcare roadmap at the 2025 APEC meeting in Gyeongju, Korea – prepared by article co-author Ted C.-Y. Chang, Vice Chair of the Bio and Healthcare Working Group under the APEC Business Advisory Council – AI is recognized as a key enabler for reducing burden while elevating the quality of care. Using deep learning, predictive AI models can accurately analyze point-of-care data, continuously assess patient conditions, and predict risks for more effective treatment. Meanwhile, generative AI models can efficiently process and summarize insights from multiple modalities of information, providing user-friendly explanations and comprehensive overviews to support decision-making (Figure 1).

 

 Fig. 1: AI as a key enabler for continuum of care

While both types of AI applications have been extensively validated and employed in consumer and professional use cases, their requirements for domain-specific data training, optimization, and application-specific workflow integration still demand substantial effort and technical expertise. In addition, a well-balanced integration of different techniques along with foundational technologies and complementary platforms remains crucial to ensuring system interconnectivity, interoperability, security, privacy, and, most importantly, resilience. Together, these elements form the essential building blocks of practical solutions to support applications in disaster medicine.

The Foundational Building Blocks

Data is to AI what air is to life on Earth. The basic building blocks of human-centric AI applications start with AIoT devices, particularly medical-grade wireless vital sign wearables that can be easily interconnected with AI-enabled platforms to leverage real-time monitoring and AI-assisted assessment based on up-to-date multisensory patient data. For example, the foundational smart healthcare ecosystem developed by Quanta (known as QOCA) encompasses a range of wearable electrocardiogram (ECG) monitors, digital stethoscopes, pulse oximeters, thermometers, and other devices under development (Figure 2).

Fig. 2: The QOCA smart healthcare platforms ecosystem

Their wireless and user-friendly nature supports seamless integration with an interoperable telehealth platform for remote patient monitoring as well as a telemedicine platform for remote consultation and diagnosis, as deployed across medical centers in Taiwan. The use of 12-lead ECG with QOCA CAS (Cardio Analysis System) has also demonstrated effective outcomes, utilizing point-of-care AI analysis tools that have saved valuable lives. This includes cases of prehospital ST-elevation myocardial infarction (STEMI) assessment and ad-hoc field screening of critical cardiovascular anomalies that would otherwise have gone undetected (Figure 3 and Figure 4).

Fig. 3: QOCA 12-lead ECG for fast emergency application

Fig. 4: Example of AI-assisted cardio risk assessment

Even with vast amounts of data – such as ECG signals, genomic sequences, pathological and other types of medical images – predictive AI tools can efficiently identify previously unseen risks with human eyes and perform assessments anytime, anywhere, and for anyone. With AIoT integration and AI-assisted analysis, patient monitoring can be elevated to a new level, enabling a continuum of care for both acute and chronic patients across emergency and long-term care settings.

To expand the capabilities of predictive AI models for patient assessment, data collected with consent can be securely used for training on an on-premise platform such as QOCA aim (AI Medicine) in a user-friendly, no-code environment – without compromising data privacy. This enables internal teams to independently develop and validate their own AI models tailored to their specific fields of expertise. In the example below, a non-technical member of a medical team can use the graphical tool to train a predictive AI model using ultrasound images without requiring to code (Figure 5). Experienced experts in disaster medicine can also benefit from the flexibility it provides, allowing them to rapidly develop specialized AI models for validation before deploying them on portable Edge AI platforms such as QOCA aid (AI Diagnosis). This approach would open up new possibilities and foster faster innovations in AI-assisted triage and point-of-care applications, which could ultimately help save more lives.

Fig. 5: No-code training of an ultrasound AI model

Toward a Resilient Care Model for Disaster Medicine

Building on the fundamental technologies and elements of QOCA, a specialized configuration (Figure 6) – QOCA RAM (Resilient AI Medicine) – has been tailored for disaster medicine scenarios to meet the objectives of resilient care management and operations under emergency conditions. Given the inherent uncertainties and chaotic dynamics of such events, additional technologies and considerations are required. The system components must also be rapidly deployable at ad hoc locations, such as field hospitals and underground shelters, and be operable in standalone mode using batteries or portable power supplies. Additionally, they must be capable of communication through mesh networks.

Fig. 6:A resilient care model with QOCA RAM

A decentralized data infrastructure is implemented to enable the resilient medical backend with information on casualties, patients, staff, resources, medical capacities, and overall situational awareness. It is distributed across self-powered, resilient nodes to ensure redundancy and maintain minimum essential operations through a local mesh network under worst-case scenarios.

Each staff member carries an NFC e-pass tag for decentralized multi-factor authentication and secure identity verification. For every injured individual, a wearable NFC e-triage tag with a built-in light indicator is provided, serving as both a portable patient medical record device and a complement to the paper-based triage tag (Figure 7). It stores digitized triage documentation and continuously appends updates throughout the patient care journey. The data can only be securely accessed using a smartphone equipped with an authorized e-pass. A cross-platform mobile application allows guided triage and rapid tag updates with context-aware guidance. These updates are also stored locally on the phone and later synchronized with the resilient medical backend whenever connectivity becomes available – via mesh networking, 4G/5G connections, or other contingent networks.

Fig. 7:Portable patient medical records with e-triage tag

When individuals require continuous monitoring for prolonged field care (PFC), a portable patient monitoring kit equipped with AIoT wearables – including an ECG monitor, pulse oximeter, blood pressure monitor, and thermometer – and a durable tablet computer can be rapidly deployed to monitor and synchronize vital signs with a nearby patient management station for multi-patient monitoring with automated alerts. Intermittent vital signs can also be sampled and stored on the individual’s e-triage tag before relocation. The portable design and integration with AI-assisted monitoring enable the care of mass casualties or displaced patients with limited medical staff (Figure 8).

Fig. 8: Portable patient monitoring in underground shelter

Furthermore, in the case of patients requiring assistance from medical specialists who are not available at the current facility, a telemedicine kit is readily available, offering the option to use point-of-care diagnostic tools – such as digital stethoscopes, portable ultrasound and X-ray devices, and endoscopes – for remote or AI-assisted diagnosis to support more comprehensive assessments. It can also provide first-person view streaming for coordinated surgical procedures that require remote guidance from an experienced expert who is not on-site (Figure 9).

Fig. 9: Surgical teleconsultation with first-person View

Finally, the resilient medical backend provides a dashboard view of all managed casualties and patients, including their up-to-date conditions, flows and movements, pre-arrival notifications, and detailed records of their care journey from the prehospital stage through facility admission. Data including casualties, patients, and staff distributions, as well as facility and individual station resource capacities can be summarized and visualized for a comprehensive overview. The goal is to deliver data-driven decision support and high-level situational awareness before critical judgments involving major prioritization or broad operational impact are made (Figure 10).

Fig. 10: Decision support with visualized information

Overall, with the mechanisms of e-triage tag and the decentralized resilient medical backend, the aim is to preserve as much information as possible about casualties and patients from the point of triage onward, ensuring that sufficient data is available for AI-assisted individual care, effective treatment, efficient coordination, and overall decision support. With integrated patient monitoring and telemedicine capabilities, the goal is to further maximize patient care capacity despite limited human and medical resources.

Future Visions

While the lives of patients are of utmost importance, the well-being of responders and medical personnel under tremendous stress is equally critical. AI can also be leveraged to monitor their physical and mental health through AIoT wearables, providing early warnings when they, too, require care and support.

In a similar way, through AIoT and AI-assisted telehealth, non-critical patients can be discharged for remote monitoring and receive virtual medical assistance from “hospitals at home”, thereby freeing space and resources for more critical patients while being able to recover more gracefully with family and community support in a familiar setting.

Nonetheless, since the use of AI requires significant power and computing resources, it also limits the feasibility, scope, and performance of its applications in field deployments. Yet, with a recent shift toward more energy-efficient edge AI computing as of late 2025, it can be anticipated that powerful point-of-care AI innovations will become increasingly prominent and accessible – empowering medical staff with augmented capabilities, even among novice members. For example, with research already underway by Quanta collaborators at the Massachusetts Institute of Technology and other institutions, AI-assisted ultrasound operations and assessments may significantly enhance the identification of pneumothorax, pleural effusion, internal bleeding, and other critical traumatic conditions, leading to more efficient and accurate patient triage.

Additionally, as agentic AI and physical AI technologies advance toward maturity, practical virtual and robotic assistants may soon become a reality – capable of performing automated assessments and assistive procedures, along with drone-delivered supplies, for injured individuals in hazardous or otherwise inaccessible environments before evacuation becomes possible. Of course, these next-generation AI technologies would further augment staff and patient support in field-hospital and hospital-at-home settings.

Conclusion

The technological perspectives and examples presented in this article illustrate potential approaches to utilizing AI in disaster medicine to support the continuum of care, particularly in situations with limited resources and possible disruptions. However, they represent only one part of a broader goal: to build resilient systems and societies for the future. As stated in the 2025 APEC Leaders’ Gyeongju Declaration, “We reiterate our commitment to building resilient, sustainable, accessible, age-responsive, multisectoral and future-ready health and care systems across the region, while acknowledging the innovative potential of digital health and AI to enhance patient-centered health service delivery, early detection, diagnosis, treatment and overall health outcomes… Recognizing disaster risk management is a pivotal foundation for economic growth, we endeavor to secure a safe and resilient future” [2].

It is still very early in exploring what AI may achieve for humanity. Still, its most significant value will lie in serving the noble purposes of saving lives and strengthening societal resilience. Undoubtedly, envisioning the goals and applications of AI in disaster medicine will help guide its development toward the greater good of humanity.

References

1. Abualenain J. Use of Technology in Disaster Medicine. Eur J Emerg Med 2024;23(3):155-158. mehr lesen
2. Asia-Pacific Economic Cooperation (APEC). 2025 APEC Leaders’ Gyeongju Declaration [Internet].[last accessed November 2, 2025] Available from: https://www.apec.org/meeting-papers/leaders-declarations/2025/2025-apec-leaders--gyeongju-declaration. mehr lesen
3. Chang, TCY, Lin, YB. AIoT for Digital Transformation of Healthcare. In: CTSOC-NCT News on Consumer Technology. 2021;December: 9-14. mehr lesen
4. Groves H, Kushner AL, Guptaa S. Protecting health facilities: design options for armed conflict and climate change disasters. Emerg Crit Care Med. 2023;3(1):1-3. mehr lesen
5. Herstein JJ, Schwedhelm MM, Vasa A, Biddinger PD, Hewlett AL. Emergency preparedness: What is the future? Antimicrob Steward Health Epidemiol. 2021 Oct 13;1(1):e29. mehr lesen
6. IBM. Technology in healthcare [Internet].[last accessed November 2, 2025]. Available from: https://www.ibm.com/history/technology-in-healthcare. mehr lesen
7. Kocak H, Kinik K, Caliskan C, Aciksari K. The Science of Disaster Medicine: From Response to Risk Reduction. Medeni Med J. 2021 Dec 19;36(4):333-342. mehr lesen
8. Leone RM, Rainwater-Lovett K, Hanfling D. Adopting Technological Innovations to Enhance Disaster Event Response. Disaster Med Public Health Prep. 2025 Feb 14;19:e32. mehr lesen
9. Lindström V, Jepsen K, Heldring S, Kanfjäll T, Rådestad M. A real-time communication and information system for triage, positioning, and documentation (TriPoD) in mass-casualty incidents: a qualitative observational study. BMC Emerg Med. 2025 Jul 6;25(1):115. mehr lesen
10. Wosik J, Fudim M, Cameron B, et al. Telehealth transformation: COVID-19 and the rise of virtual care. J Am Med Inform Assoc. 2020 Jun 1;27(6):957-962. mehr lesen

Acknowledgement

The authors wish to express sincere appreciation to the colleagues at Quantafor their invaluable contributions to this work. Special thanks are extended to K.C. Chen, Sam Chang, T.Y. Hsieh, Dr. James C.H. Song, Dr. Eric Fang, Dr. Alex Huang, Jeffrey Chen, Petter Pan, Joe Yu, Happyer Huang, Dr. Yuman Chung, Vic Chui, Charles Shih, Yvonne Hsiao, Lewis Cheng, Chain Hsu, Max Chien, Elma Kuo, Harry Lin, Allen Chen, Wade Chen, and many more team members for their dedication and support.

Manuscript Data

Citation

Huang TT, Chang TCY, Hung CH, Yang TH. The Use of Artificial Intelligence in Disaster Medicine. WMM 2025;69(12):506-512.

DOI: https://doi.org/10.48701/opus4-787

Authors

Terence T. Huang
Sr. Director, Quanta Research Institute
Quanta Computer Inc.

No. 211, Wenhua 2nd Rd., Guishan Dist., Taoyuan City, Taiwan

E-Mail: Terence.Huang@quantatw.com

Dr. Ted C.-Y. Chang

Chief Technology Officer, Quanta Computer Inc.

Vice Chair of 2025 ABAC BHWG

ABAC: APEC Business Advisory Council

BHWG: Bio and Healthcare Working Group

Representative to ABAC for Taiwan since 2019

Chung-Hsi Hung

Sr. Director, Quanta Computer Inc.

Dr. Tzu-Hsiang Yang

Sr. Director, Quanta Computer Inc.

Manuskriptdaten

Zitierweise

Huang TT, Chang TCY, Hung CH, Yang TH. [Einsatz von Künstlicher Intelligenz in der Katastrophenmedizin]. WMM 2025;69(12):506-512.

DOI: https://doi.org/10.48701/opus4-787

Verfasser

Terence T. Huang

Sr. Director, Quanta Research Institute

Quanta Computer Inc.

No. 211, Wenhua 2nd Rd., Guishan Dist., Taoyuan City, Taiwan

E-Mail: Terence.Huang@quantatw.com

Dr. Ted C.-Y. Chang

Chief Technology Officer, Quanta Computer Inc.

Vice Chair of 2025 ABAC BHWG

ABAC: APEC Business Advisory Council

BHWG: Bio and Healthcare Working Group

Representative to ABAC for Taiwan since 2019

Chung-Hsi Hung

Sr. Director, Quanta Computer Inc.

Dr. Tzu-Hsiang Yang

Sr. Director, Quanta Computer Inc.