This paper is available in PDF format here.
ABSTRACT
Recent advances in biotechnology and the life sciences have led to new and emerging paradigms for biological detection. For instance, technologies for analyzing brain activity are advancing rapidly and may soon find their way into a multitude of consumer electronics and medical devices. Other technologies are using biological or bio-inspired methods to analyze chemicals present in air, including those of biological origin, allowing the technologies to detect and sense compounds—such as disease biomarkers or industrial pollutants—with unprecedented speed and precision. What capabilities might these technologies unlock? What economic and societal drivers are influencing their development? What ethical, legal, and social issues do they raise? The National Academies of Sciences, Engineering, and Medicine hosted a virtual workshop on Cutting Edge Scientific Capabilities for Biological Detection on January 20, 21, and 28, 2022, to explore emerging technologies for biological detection and critical issues related to their development and use.
The workshop was organized by the National Academies’ Standing Committee on Biotechnology Capabilities and National Security Needs, which facilitates engagement between the national security community and biotechnology stakeholders. The Standing Committee’s work includes the identification of advanced biotechnologies with promising capabilities to meet national security needs and early-stage research that may lead to new or enhanced biotechnologies. Under this charge, the planning committee for the workshop identified two biological detection areas of interest with recent advancements and promising capabilities to highlight in the workshop: technologies for the detection of neural signatures, and technologies related to the digitization of olfaction and the detection and analysis of volatile organic compounds (VOCs).
Presenters and attendees from government, academia, and the biotechnology industry gathered for sessions of live discussions and pre-recorded talks examining cutting-edge research and developments related to these technologies. Sessions explored key advances and their potential applications in the near and long term; the innovation ecosystem that is driving the research, development, and application of these technologies; and critical societal implications of their adoption. Attendees also explored workforce needs, infrastructure, and policy and governance associated with the development and use of biological detection research and technologies.
EMERGING TECHNOLOGIES FOR DETECTING NEURAL SIGNATURES
Nita Farahany (Duke University) and Diane DiEuliis (National Defense University) moderated sessions examining technologies designed to detect and analyze human brain activity. Analysis of signals emanating from the brain can be interpreted to infer cognitive and affective states and form the basis for direct interfaces connecting brains with the technologies used. The speakers discussed recent developments in this field along with near- and long-term applications; barriers and constraints; and ethics, security, and privacy issues.
Technologies for Interpreting Brain Activity
Several speakers shared examples of technologies that measure the brain’s electrical activity. Allan Levey (Emory University) offered a brief overview of how electroencephalography (EEG) provides information about brain activity by measuring electrical activity across the brain’s surface. EEG technology has been widely used in research and medical applications for decades, where it has proved especially useful for detecting seizures and states of consciousness. However, EEG typically requires the subject to wear a cap with multiple sensors covering the entire scalp, limiting its use outside of laboratory or medical settings, and only allows access to the top surfaces of the brain, limiting insights into activity in deeper regions of the brain.
Levey and Jonathan Berent (NextSense, Inc.), Dan Furman (Arctop, Inc.), and Maria Ruiz-Blondet (Neurable) presented on recent advances in EEG and related technologies that are poised to overcome some of these limitations and open new opportunities for brain wave analysis in consumer electronics, research, and medicine. The speakers highlighted examples of how brain signals can be obtained from headwear that is more practical than the traditional EEG cap, such as glasses, earwear, or hats (Arctop, Inc.); headphones (Neurable); and earbuds (NextSense, Inc.). Arctop, Inc., for example, is developing a software package, NeuosTM SDK, for a variety of head-mounted devices capable of interpreting the brain’s electrical signals to support brain-based interaction with consumer electronics. For instance, Furman described how the technology has been used in demonstrations to authenticate a user’s identity and to allow users to direct a digital interface to flip and sort cards without any physical or voice controls. Neurable’s headphone-based EEG system, Ruiz-Blondet explained, uses electrodes to continuously measure brain activity for insights on the wearer’s attention and focus. The system is being developed as an aid to help users stay on task amid distractions, part of the company’s broader goal of developing an everyday brain–computer interface for a variety of use cases.
Levey and Berent described how NextSense, Inc.’s EEG earbuds provide real-time analysis of brain activity in the temporal lobe, the brain’s center for memory, language, and personality. The in-ear sensors can record 50 hours of brain activity and store the data in the earbud or export it to the cloud via Bluetooth. In proof-of-concept demonstrations, Berent discussed that the technology has been shown to be capable of tracking a wearer’s attention (e.g., determining which of two simultaneous audio recordings a person is paying attention to) and feelings toward stimuli (e.g., detecting if a person likes or dislikes a photo). The device also can be used as a platform to influence brain activity. Berent explained how experiments have shown evidence that stimulation delivered via the earbuds can affect sleep patterns and even trigger memories and induce a hybrid state of consciousness that enables the wearer to undertake some conscious activities while sleeping.
Todd Constable (Yale University) described another approach to gaining insights from the human brain that is based on the synchronous activity of anatomically distinct regions of the brain. The approach, known as functional connectivity mapping, uses functional magnetic resonance imaging (fMRI) to decipher brain activity based on small changes in blood flow. Constable’s team developed a method to map a person’s “functional connectome”—akin to a fingerprint of the brain—and used it to produce maps of up to 70,000 connections. These maps can provide insights on a person’s fluid intelligence, working memory, attention, personality traits, and more.
Applications of Brain Interfacing Technologies
Speakers discussed near- and long-term applications envisioned for brain interfacing technologies. One application that has already been well demonstrated is biometrics. Ruiz-Blondet and Furman described how brain signals can be used to discern the unique EEG signatures of individual people to verify a person’s identity, which could be useful for supporting authentication for restricted-access systems. As with any technology, Furman noted that there are tradeoffs between convenience and security. While brain-based biometrics may be less convenient than some other methods because it requires a device to be in contact with the head, this approach could be highly secure at a population level given the vast number of differences that can potentially be discerned between individuals’ “brainprints” as compared to fingerprints.
Another application area with both near- and long-term potential is medical monitoring and intervention. Berent said in-ear EEG could be used for monitoring seizures and guiding drug titration in people with epilepsy given the strong signals that the technology has been shown capable of acquiring. Constable said functional connectome approaches could hold significant potential for diagnosing cognitive or neurologic disorders and monitoring patients’ response to treatment by providing an objective, detailed view of a person’s brain activity and how it changes over time. He also noted that functional brain analysis holds promise as a tool for researching brain development, the effects of preterm birth, and the processes that influence intelligence.
Furman and Ruiz-Blondet discussed how brain wave analysis could support more intuitive, frictionless interaction with technology in the next decade. For example, Ruiz-Blondet said devices could interpret brain signals to discern when a person is focused on their task and when would be an optimal time for a notification or signal, providing a more seamless and less intrusive interaction that adapts to the user’s state of mind. Furman added that having feedback from the user’s brain can help make technology more empathetic and personable, enabling the technology to understand and respond appropriately to the user’s mental and physiological state.
Finally, speakers explored how brain analysis technology could be used as a basis for personalized education and learning. Constable said functional connectome approaches are already being pursued by international businesses as a potential tool for assessing and predicting cognitive skills in children and informing remedial interventions, though he said it will still be a while before the technique will be able to reliably predict cognitive performance. Furman, Ruiz-Blondet, Levey, and Berent described how wearable devices could be used to track focus and even assess how information is being encoded in the brain while a person is learning, enabling systems to provide feedback to individualize the educational experience and help the learner stay on task.
Ethics, Security, and Privacy Issues
Farahany noted that the information collected by brain scanning technology is sensitive by nature. In addition to concerns about protecting information that can reveal a person’s identity and health status, brain signal analysis raises special concerns related to mental privacy. By accessing cognitive and affective states, these technologies tap into what people consider the most sensitive and essential parts of their identity and what it means to be human, Farahany noted, raising significant ethical, security, and privacy issues.
How brain signal data are collected, handled, and used has important implications for the security of these data and the privacy of the individuals whose data are being accessed. For example, are the data stored and computed on the device, or sent to and stored on the cloud? Is all of the raw data kept, or is it filtered and processed before storage? Is the use of the data limited to what is currently technologically feasible, or could it be used in different ways in the future? Furman said these remain open questions in the field. While his company takes a user privacy-centered approach, different companies may take different approaches. Furman and Berent said their companies are focusing more on edge computing solutions that largely keep the data local to the device, while Ruiz-Blondet said Neurable takes a cloud-based approach. Furman underscored the need for regulation and transparency on the part of companies collecting, analyzing, and storing this type of data.
Several speakers stressed that it is vital to ensure those who use the technology understand and consent to the way their data will be collected, handled, and used. Because obtaining functional connectome data requires an fMRI scan, this approach involves some level of consent on the part of the person whose brain is being assessed, Constable noted, although it is still important to protect individuals’ privacy in terms of how the data are used and shared. Raw functional connectome data can potentially be used to derive a wide range of insights beyond the original purpose for which the data are collected. For example, he said it is possible to derive an image of a person’s facial phenotype from brain data alone, meaning patient privacy could include manipulating or “defacing” the data to remove information about potentially recognizable physical features before sharing. For EEG data, Berent and Ruiz-Blondet noted that it can be valuable to offer the opportunity to “donate” their brain data to science or to help companies improve their algorithms, but that this should be done transparently and with attention to protecting individual privacy. Given that numerous companies are actively working to bring the next generation of cutting-edge brain analysis devices into clinical and consumer use in the near term, Berent said there is an urgent need for solutions to ensure people who contribute brain data can retain ownership and control over how that data are used. “I think that this is [going to be] here before we know it,” Berent said. “It’s coming, and so we have to be really thoughtful and make sure that we have safeguards around the data and that the users are in control of what they do with it.”
Brain analysis technologies have unique capabilities that raise unique issues. Even when they are not being used in the context of health assessment, for example, devices could pick up health-relevant information, Furman noted. What happens if a brain analysis device detects a health problem in a user? Who should be notified, and how? He advocated for prioritizing the rights of users over business models, with careful attention to how companies use the data. “It’s your own data, you should own it, you should have a place you can control it … and get the maximum value out of these really exciting and powerful and fun, invigorating technologies,” said Furman. “We need to be very deliberate and intentional about how we use this, how we potentially regulate it.”
Future Outlook
The speakers identified future opportunities, barriers, and research priorities for emerging technologies that analyze brain signals. Constable and Ruiz-Blondet pointed to a need for larger and more diverse data sets to better understand how brain signals work (for EEG devices) and relate individual behavior to functional brain organization (for functional connectome applications). The speakers also said that additional research is needed to understand the full scope of the information that is accessible via the diverse spectrum of brain signals, identify which areas are most useful for applications, and transition to commercial deployment.
Ruiz-Blondet said headphone-based EEG can detect signals from as far back as the visual cortex, which allows those devices to capture information relevant to attention, muscle activation, and other activities that can be useful for voiceless interaction with technology. She explained that while there is still work to be done to determine which signals are most useful for what types of information, having a clear signal is key. She noted that advances in materials would be useful to ensure sensors maintain optimal connection with the skin. She also mentioned that Neurable is exploring ways to give users feedback to ensure the headphones are placed properly for a good signal and not occluded by hair, for example.
Berent speculated that in-ear EEG would ultimately be able to access about 85% of the brain; his company is focusing on determining which of the accessible areas hold the greatest potential for medical and other applications. Levey added that some research suggests it is possible to use in-ear brain stimulation to block or enhance a memory, but Berent said the noisiness of EEG signals make it unlikely for this to be feasible in the near term. Other potential long-term medical applications for in-ear devices they discussed, which further research and development could enable, include addressing sleep problems; applications that involve stimulating the vagus nerve; and delivery of high-frequency gamma waves to guide neuroimmune interventions for the treatment of certain brain diseases. Berent cautioned, however, that any neuromodulation requires great care to avoid unintended effects.
In the context of EEG-based brain biometrics, participants discussed the potential limitation of whether early-stage results and current findings for authentication and identification applications are scalable when applied across an entire population. Some participants expressed optimism as current results show robust reliability for authentication; however, they noted that additional research for identification applications might be needed as the technology scales. Regarding scaling, Furman said that the components and supply chains to integrate “brain ID” capabilities into consumer electronics are already in place. He suggested that big tech companies will likely play the largest role in determining when these capabilities hit the market, though he speculated that it could be within the next year. “There’s really nothing holding back this brain ID from being rolled out to a commercial product,” he said. Beyond biometrics, Furman posited that the future of the field will rely on greater appreciation for the technological capabilities of brain analysis and a workforce that is equipped to think creatively about potential uses. “I think one of the big gaps in adoption is simply the awareness and understanding of what this technology can do,” he said. Ultimately, Arctop, Inc., predicts that brain analysis capabilities will be integrated into about 1 billion devices in the next 10 years.
FINAL THOUGHTS
Throughout the workshop’s presentations and discussions, participants examined a variety of technologies in the field of biological detection, summarized in Table 1. Technologies that detect neural signatures and detect chemical markers in surroundings likely hold great potential for applications aimed at
improving human health, national and personal security, and much more. Some of these technologies are already on the brink of commercial deployment, where they may soon find their way into clinics and consumer electronics. Others are still decades away from practical use but could unlock new ways of understanding bodies and the world. Across all of these use cases and development trajectories, many participants stressed that it will be important to consider how structures, norms, and processes can help to support ethical practices—including those that enhance transparency, privacy, and security—in the development and adoption of biological detection technologies.
Table 1. Overview of example technologies discussed at workshop.