Accelerated innovation is crucial to revolutionize vaccine models, pathogen tracking and containment, and bioterror prevention.

Yll Bajraktari , 2021, “The Role of AI Technology in Pandemic Response and Preparedness: Recommended Investments and Initiatives,” https://www.nscai.gov/wp-content/uploads/2021/01/NSCAI_White-Paper_The-Role-of-AI-Technology-in-Pandemic-Response-and-Preparedness.pdf

Below we summarize several ways in which AI can be applied to advance healthcare, biomedicine, and biosecurity. These are broken down across four primary areas: 1) situational awareness and disease surveillance; 2) diagnostics; 3) production of vaccines, therapeutics, and medical equipment; and, 4) ongoing healthcare - protecting the healthy and treating the sick. AI is continually evolving. Investing in these areas will improve existing applications, lead to novel use cases, and offer even greater opportunities to keep our citizens safe. While it's unclear whether AI will significantly affect the course of COVID-19, these initial investigations suggest how AI can be used to enhance pandemic preparedness and response in the future. A. Situational Awareness and Disease Surveillance Understand the threat through disease modeling. AI is being employed to better understand how diseases, including COVID-19, spread and ways to fight them. AI models are predicting the structure of proteins associated with the virus and mapping the spread of infection, enabling researchers to determine the mechanisms the virus 10 employs to infect humans and how they can be inhibited. Natural Language 11 Processing helps researchers gain faster understanding of COVID-19 by combing through vast amounts of medical research literature to find answers and build on the work of other experts. In the future, the application of AI to virology could 12 dramatically expedite our ability to understand and combat emerging viral threats. Enable early outbreak detection and real-time disease surveillance. AI is currently being used to ingest and fuse near real-time data from varied sources––including news feeds, social media, public health data, clinical health data, and digital health tools––to monitor for outbreaks, gauge new hot spots, and measure the progress and severity of the virus. Canadian company BlueDot used an AI-driven tool to detect the COVID-19 13 outbreak in Wuhan, China, doing so before widespread recognition of the disease’s emergence. Boston-based enterprise AI provider DataRobot has builtAI-based models that predict COVID-19 spread down to the county level, creating a tool that can help inform local policy decisions. AI-driven platforms will be 15 increasingly capable of universal data ingestion of unstructured multi-domain data sources. These platforms will provide leaders and public health officials with tailored alerts and real-time situational awareness of emerging threats. Monitor and prevent zoonotic spillovers. AI-enabled predictive tools can help monitor for zoonotic spillovers, identify at-risk environmental areas and inform where, 16 when, and what data to collect; thereby contributing to both research and critical early warning systems. AI has been used to identify new species of virus, as well as animal 17 species that may carry pathogens that could jump to humans. Today’s AI capabilities 18 can already be used to sift through molecular, epidemiological, ecological, and climate data to identify vulnerabilities for such zoonotic spillovers. As more data is generated 19 through this work and as the technology progresses for merging multi-source data and models, zoonotic warning systems could combine with the outbreak detection platforms described above for earlier and more accurate disease detection. Advance bioforensics for threat analysis. When an outbreak occurs, bioforensic analysis plays a key role in determining the cause as natural, intentional, or accidental.20 COVID-19 has highlighted the inherent complexity in identifying the origin of a disease outbreak. Bioforensic capabilities seek to detect high-consequence biothreats and profile pathogen samples during an incident for unique signatures that can aid attribution. AI 21 is enabling advances in key sciences for next generation bioforensics—including genomics, proteomics, and metabolomics —that seek to answer both agent-specific 22 and threat-agnostic questions. AI will further progress bioforensics by enabling 23 inferential and predictive methods for interpreting data and identifying key indicators.24 Detect pathogens with advanced sensing. New intelligent sensor technologies could compile data on new pathogens (including instances of weaponized pathogens), trace exhaled droplets dispersed through the air and deposited on surfaces, detect early 25 onset of infection through wearable sensors, and encourage social distancing through 26 mobile proximity detection. Going forward, AI will likely play an increasing role in identifying novel sensing materials and ways to build new capabilities, as well as serve a functional role within certain sensors.B. Diagnostics Diagnose patients and triage care. AI is helping scientists develop more effective tests for COVID-19 by allowing them to more rapidly and comprehensively infer biological insights that contribute to test design, as well as model and design the 27 antibodies and enzymes that are the components of immunoassay diagnostic tests. 28 AI-enabled tools are also being directly used in diagnostics to assist and inform expert assessments. Tools that analyze chest scans have assisted radiologists in detecting diseases like breast cancer and tuberculosis. Some hospitals have applied similar 29 technology to detect COVID-19 and to assess disease severity and care needs. Others 30 have built novel AI-based tools that aid diagnosis based on clinical symptoms and routine laboratory tests. However, the success of these tools is highly dependent on 31 large amounts of disease-specific training data, making it difficult to rapidly transition AI tools designed for other diseases to help healthcare professionals detect and assess COVID-19. As more data and clinical training enhance diagnostic algorithms, AI will 32 be able to contribute in additional ways to diagnostic test development, and will more reliably support radiologists and emergency care teams in rapidly identifying disease indicators and informing patient care.33 C. Development of Vaccines, Therapeutics, and Medical Equipment Rapidly discover and design vaccines and therapeutics. Researchers are leveraging AI to create simulations of how the coronavirus might interact with different therapeutics and vaccines, vastly accelerating the discovery process by exploring a great number of possibilities and narrowing the field of viable candidates. Others are using 34 AI-driven simulations to assess feasibility of repurposing existing drugs as a first step toward treating COVID-19. Additionally, AI can help speed drug and vaccine 35 production by enabling continuous manufacturing processes and digital twins tosupport clinical trials. As the technology develops, AI has the potential to support 36 automated pipelines for building models that simulate possibilities, and synthesize and validate solutions, allowing researchers to respond to emerging viral threats more rapidly with efficiently designed therapeutics and vaccines.37 Manage supply chains. Today’s supply chains are complex, and in the private sector, AI is being used to create self-correcting, smart supply chains. AI can predict 38 disruptions and vulnerabilities, monitor and prioritize inventory, prompt on-demand manufacturing, order orchestration and fulfillment, and recommend corrective actions. We foresee a future in which AI enables intelligent workflows to manage complex and global supply chains for pandemic response, dynamically managing surges and failures.39 Drive advanced manufacturing. AI has enabled the manufacturing industry to speed production and enable innovative manufacturing techniques for medical equipment, including ventilator components, face shields, and personal protective equipment. AI advances manufacturing capabilities, productivity, and quality by helping manage supply chains for raw materials, predict the need for maintenance (thereby reducing production downtime), facilitate human-machine collaboration, and enable generative design. AI further optimizes 3D-printing, which proves useful in adapting to demands 40 for pandemic-related supplies. In the near future, we believe AI-powered advanced 41 manufacturing can provide the agility needed to manage surges in demand for critical items needed during pandemics. In this new era, AI could be used to optimize manufacturing, optimize logistics and supply chains, drive autonomous equipment and vehicles, empower intelligent robots, and direct next-generation 3D-printing and on-demand manufacturing.42 D. Ongoing Healthcare - Protecting the Healthy and Treating the Sick Manage patient risk and mortality. An existing tool in use in 20 hospitals around the United States predicts which patients will experience cardiac arrest and require intensive care, helping physicians determine which patients to admit to the hospital. AI-driven 43 models are being developed that can help predict a patient’s severity of illness and risk of mortality. Such a capability can inform crucial triage decisions when hospitals 44 become overcrowded, providing decision support tools to inform allocation of limited resources. The reliability of these tools will increase as AI design and engineering 45 principles improve, allowing independent testing with comparable performance metrics.   Reduce transmission rates through use of autonomous robots. AI is being used to limit human interaction by enabling new autonomous robotic capabilities and services. These include assisting clinicians in hospitals, monitoring re-opened public spaces, enabling telehealth services, and processing test samples. A Danish company has 46 developed a line of ultraviolet light emitting robots that can travel through hallways and rooms disinfecting areas with no human intervention. Hospitals in Boston have used 47 AI-enabled robots to diagnose COVID-19 patients remotely, enabling informed allocation of in-person care, minimization of exposure, and preservation of protective equipment. Today’s robots are employed in a limited capacity to carry out a small set 48 of specialized tasks such as relaying communications, delivering supplies and meals, monitoring patients, and disinfecting. In the future, a robot may be able to oversee and complete complex, multi-phased tasks such as laboratory testing pipelines that involve teams of robots conducting sample collection, preparation, test administration, error recovery, results interpretation, and results delivery. Remotely monitor patients. With appropriate sensor data, AI is being developed to help physicians remotely monitor patients, both severe cases in intensive care units and moderately ill patients from their homes, as well as conduct remote medical trials. A 49 hospital in Houston has created a “clinical command center” using high-resolution cameras and an analytics platform that uses machine learning algorithms to assess the state of critical patients based on real-time ingested data feeds from ventilators, electrocardiogram machines and oxygen pumps as well as a patient’s electronic medical record and vital signs. A clinical team in Boston is testing an AI-driven device 50 developed by Massachusetts Institute of Technology’s Computer Science and Artificial Intelligence Laboratory that remotely monitors a COVID-19 patient’s breathing, movement, and sleep patterns using wireless signals. Meanwhile, the 51 Department of Defense (DoD) is working with company PhysIQ to monitor COVID-19 patients and those at high risk using clinical-grade biosensors that share data with a Food and Drug Administration approved, AI-enabled platform. Remote 52 monitoring capabilities will likely become increasingly integrated into treatment programs and medical trials as the technology develops, reducing the risk of disease transmission and the burden on hospitals, as well as accelerating robust clinical testing of new medicines.