Top 10 Emerging Technologies Transforming Healthcare: AI, Gene Editing, Robotics, and Advanced Materials

The World Economic Forum in Davos released the 2018 Top 10 Emerging Technologies report: Augmented Reality, Personalized Medicine, AI-Driven Molecular Design, Digital Assistants, Implantable Cell-Based Therapies, Lab-Grown Meat, Electronic Therapies, Gene Drives, Plasma Materials, and Quantum Computing Algorithms.

VCBeat (WeChat: vcbeat) has found that most of these are related to healthcare or have significant application potential in the medical field.

VCBeat has translated the relevant report, which provides answers to a series of healthcare-related questions—such as how machine learning algorithms can assist in new drug development, how implantable cell-based therapies can transform the treatment of chronic conditions, how electronic therapies can reduce reliance on pharmaceuticals, and how light-controlled nanoparticles can be used in cancer treatment.

Virtual reality (VR) immerses users in a self-contained, fictional world. In contrast, augmented reality (AR) overlays computer-generated information onto the real world in real time. When you view or wear devices equipped with AR software and cameras—whether smartphones, tablets, headsets, or smart glasses—the corresponding applications analyze the incoming video stream, retrieve extensive data about the scene, and superimpose relevant data, images, or animations, typically within a three-dimensional space.

For example, displays that assist with safe reversing and the popular game Pokémon GO. Many consumer applications—such as apps that translate street signs for foreign tourists, tools that allow students to dissect virtual frogs, and features that enable shoppers to visualize how a chair would look in their living room before taking it home—also involve VR technology. In the future, this technology will allow museum visitors to view simulated holographic museum maps, enable surgeons to visualize patients’ subcutaneous tissues in three-dimensional scenes, facilitate novel collaboration between architects and designers, empower drone operators to control remote robots through enhanced imagery, and help novices rapidly learn tasks ranging from pharmaceutical R&D to factory maintenance.

In the coming years, software for designing applications should place greater emphasis on consumer products. However, as a key component of the Fourth Industrial Revolution, or Industry 4.0, augmented reality (AR) currently exerts significant influence in this domain: by integrating physical and digital systems, it enables systemic transformation in manufacturing, thereby improving quality and efficiency while reducing costs. For instance, many companies are conducting trials on their production lines. AR can provide the right information at the right time (such as guiding workers in part selection), thus lowering error rates and enhancing efficiency and productivity. It also visualizes issues within equipment and generates real-time images pinpointing the location of problems.

Market analysts from firms such as ABI Research, IDC, and Digi-Capital believe that augmented reality (AR) is on the cusp of going mainstream. They project that the value of the AR market, currently standing at approximately $1.5 billion, will grow to $100 billion by 2020. Major technology companies, including Apple, Google, and Microsoft, are committing substantial financial and human resources to the research and development of AR and virtual reality (VR) products, as well as related applications. In 2017, investment in the AR and VR sectors reached $3 billion, with half of that amount invested in the fourth quarter alone. The Harvard Business Review emphasizes that “augmented reality” is a revolutionary technology that will have a profound impact on all industries.

However, hardware and communication bandwidth limitations have posed obstacles to consumers’ daily use. For instance, many existing museum and travel applications require pre-downloading to enhance experiences through AR. Moreover, image quality may fail to meet user expectations. Nevertheless, with the advent of more affordable and faster AR mobile chips, the market entry of increasingly multifunctional smart glasses, and increased bandwidth, this field is poised for rapid development. Subsequently, through internet connectivity and real-time communication, AR will become an integral part of our daily lives.

In the 20th century, all women with breast cancer received the same treatment. Today, therapy has become more personalized: breast cancer is classified into distinct subtypes, each managed with corresponding treatments. For instance, in cases where tumors exhibit abnormal estrogen receptor expression, patients can take medications specifically targeting these receptors and undergo standard postoperative chemotherapy. In 2018, researchers took another step toward more personalized treatment. They discovered that a substantial proportion of patients have tumors with certain characteristics, indicating that they can safely forgo chemotherapy and avoid severe side effects.

Advances in diagnostic tools have accelerated the development of personalized or precision therapies for many diseases. These technologies can help physicians detect and quantify various biomarkers (signaling molecules that indicate the presence of disease), stratifying patients into distinct subgroups based on their susceptibility to disease and their potential response to specific treatments.

Early molecular diagnostic tools focused solely on individual molecules. For instance, diabetes management centered exclusively on glucose monitoring. However, over the past decade, “omics” technologies have continuously advanced, enabling researchers to rapidly and reliably perform whole-genome sequencing or quantify all proteins (proteome), metabolic byproducts (metabolome), or microorganisms (microbiome) in bodily fluids or tissue samples. The routine application of these technologies generates massive datasets, which artificial intelligence can mine to discover novel biomarkers with clinical utility. The integration of high-throughput omics technologies with artificial intelligence is ushering in a new era of advanced diagnostics, transforming the understanding and treatment of many diseases and empowering physicians to tailor therapies based on patients’ molecular profiles.

Some advanced diagnostic technologies have been applied to cancer. A technology called Oncotype DX can detect 21 genes, helping many women with breast cancer avoid chemotherapy. Another technology, known as “FoundationOne CDx,” can identify more than 300 gene mutations in solid tumors and pinpoint specific gene-targeted drugs that may be beneficial for patients.

Beyond cancer, certain technologies are also being applied to endometriosis, a common gynecological condition in women characterized by the presence of endometrial tissue outside its normal location, which typically requires surgical intervention for diagnosis. A non-invasive saliva test developed by DotLabs can identify endometriosis by measuring a panel of small molecules known as microRNAs. Furthermore, blood tests can help detect neurological disorders such as autism, Parkinson’s disease, and Alzheimer’s disease. Currently, the diagnosis of these conditions relies on clinicians’ subjective assessment of symptoms. Researchers are also exploring whether whole-genome sequencing, microbiome analysis, and measurement of hundreds of proteins and metabolites in healthy individuals can provide personalized guidance for disease prevention.

It is important to note that healthcare institutions and researchers using these diagnostic tools must strictly protect patient privacy. Furthermore, clear regulatory guidelines and standards are needed for biomarkers as diagnostic tools to facilitate the translation of novel biomarkers into clinical practice.

Nevertheless, advanced diagnostic technologies have begun to challenge the standard approaches to disease diagnosis and treatment. By guiding patients toward the most effective therapeutic interventions, healthcare institutions and professionals can reduce medical expenditures. In the future, we may establish cloud-based archives of biomarker data. Over time, these accumulating data will provide patients and physicians with real-time insights into treatment progress, accessible anytime and anywhere.

To design a new solar energy material, an anticancer drug, or a compound that prevents crops from viral infections, one must first address two challenges: identifying the correct chemical structure of the relevant substance, and determining which chemical reactions will link the appropriate atoms into the desired molecule or molecular assembly.

Generally, addressing the aforementioned issues relies on complex conjectures and serendipitous discoveries. However, this process is highly time-consuming and often involves numerous failed attempts. For instance, a comprehensive synthesis plan may comprise hundreds of individual steps, many of which can lead to undesirable side reactions or by-products, or simply fail to work. Nevertheless, leveraging artificial intelligence (AI) can enhance the efficiency of design and synthesis, making the entire process faster, easier, and more cost-effective, while simultaneously reducing chemical waste.

In artificial intelligence, machine learning algorithms can analyze all known experiments that have attempted to discover and synthesize relevant active compounds but ultimately failed. Based on the identified patterns, these algorithms can predict the structures of potential new molecules as well as their synthesis methods. However, while a single machine learning tool cannot accomplish all tasks, AI technologies are accelerating the design of drug molecules and materials.

For example, researchers at the University of Münster in Germany have developed an AI tool capable of repeatedly simulating 12.4 million known single-step chemical reactions to construct multi-step synthetic routes—30 times faster than human experts performing the same task.

In the pharmaceutical sector, AI-driven generative machine learning technologies have also witnessed rapid development. Most pharmaceutical companies screen millions of compounds to identify their potential as new drugs. However, even with robotics and laboratory automation tools, this screening process remains exceedingly slow and yields relatively limited results, covering only a small fraction of the estimated 10^30 theoretically possible molecules. By learning from datasets comprising the chemical structures and properties of known drugs (and drug candidates), machine learning tools can construct virtual databases of novel compounds that are structurally similar yet more practical and property-optimized, thereby facilitating the identification of drug leads.

Nearly 100 startups are leveraging artificial intelligence for drug development, including Insilico Medicine, Kebotix, and BenevolentAI. Among them, BenevolentAI has raised $115 million and plans to apply its AI technology to the development of therapies for motor neuron disease, Parkinson’s disease, and other conditions—spanning from novel molecule discovery to the design and analysis of clinical trials, with the aim of demonstrating drug safety and efficacy.

In the field of materials science, companies such as Citrine Informatics are adopting approaches similar to those used by pharmaceutical firms and collaborating with major corporations like BASF and Panasonic to accelerate innovation. The U.S. government is also supporting research in AI-driven design; since 2011, it has invested more than $250 million in the Materials Genome Initiative to build infrastructure that includes artificial intelligence and other computational methods, thereby accelerating the development of advanced materials.

Past experience has shown that new materials and chemicals may pose unforeseen risks to health and safety. Fortunately, artificial intelligence approaches can predict and mitigate these adverse outcomes. These technologies appear to significantly enhance the speed and efficiency of research and development for new molecules and materials, facilitating their market entry and contributing to improvements in healthcare and agriculture, strengthened resource conservation, and the production and storage of renewable energy.

Today, intelligent assistants such as Siri and Alexa leverage sophisticated speech recognition software to respond to user requests, generate natural-sounding speech, and provide relevant information tailored to specific queries. These systems must first undergo “training” by processing a vast array of potential human requests; researchers are required to design appropriate responses and organize them into highly structured data formats.

This work is highly time-consuming and can restrict digital assistants in task execution. These systems are capable of “learning”—their machine learning capabilities enable them to improve the matching between input queries and existing answers—but their scope is limited. Even so, this technology has still had a significant impact.

AI technology is continuously evolving, advancing toward higher levels of complexity. Next-generation systems can ingest and process unstructured data from diverse sources (such as raw text, videos, images, audio, and emails) and automatically generate sound recommendations on topics outside their training scope.

We have already seen this functionality in chatbots deployed on websites, which can answer questions posed in natural language by leveraging various datasets used during their training. These bots require relatively little, or even no, additional training for specific questions or requests. They combine pre-configured data with the ability to “read” relevant background materials. However, they do require some training in language and intent recognition before they can deliver highly accurate responses.

In June this year, IBM unveiled a more advanced technology: a system capable of engaging in real-time debates with human experts without prior preparation. Leveraging unstructured data—including content from Wikipedia, some of which was edited for accuracy—the system must determine the relevance and veracity of information, reorganize it into usable insights, and construct coherent arguments. It can also respond to the arguments presented by its human opponents. During the launch event, the system participated in two debates; in one of them, many audience members found its arguments more persuasive.

The development of this technology has spanned more than five years, and it remains in the research and development phase. It includes software capable not only of understanding natural language but also of detecting positive and negative emotions. Nevertheless, non-scripted artificial intelligence systems have defeated recognized human experts, laying the foundation for countless related applications. Such applications are likely to emerge continuously within the next three to five years, or even sooner. For instance, intelligent systems can help physicians rapidly identify research relevant to complex cases and then discuss the merits of a given treatment plan.

These intelligent systems will only be useful for learning existing knowledge, rather than creating knowledge like laboratory scientists or experts. Nevertheless, as machines become increasingly intelligent, they may lead to widespread unemployment. Addressing these issues requires human wisdom, and society has a responsibility to equip the next generation with the necessary skills.

Many patients with diabetes measure their blood glucose levels multiple times a day to determine the required insulin dosage. Transplanting insulin-producing pancreatic cells—known as islet cells—can simplify this cumbersome process. Similarly, cell implantation can improve the treatment of other diseases, including cancer, heart failure, hemophilia, glaucoma, and Parkinson’s disease. However, cell implantation has a major drawback: recipients must take immunosuppressants indefinitely to prevent immune rejection. These medications can cause serious side effects and increase the risk of infections or malignancies.

After decades of research, scientists have developed a method to encapsulate cells within semi-permeable protective membranes, shielding implanted cells from immune system attacks. These capsule-like structures still allow nutrients and other small molecules to flow in, while enabling essential hormones or other therapeutic proteins to flow out. However, merely protecting the implanted cells from damage is insufficient: if the immune system recognizes the protective material itself as foreign, it will trigger the growth of scar tissue around the “capsules.” This “fibrosis” blocks nutrient access to the cells, ultimately leading to cell death.

Currently, researchers are working to address the challenges posed by fibrosis. For example, in 2016, a research team at the Massachusetts Institute of Technology (MIT) published a method to make implants invisible to the immune system. After producing and screening hundreds of materials, the researchers identified a chemical gel called alginate, which is harmless to the human body. The researchers encapsulated islet cells within this gel and then implanted them into diabetic mice. These cells immediately began producing insulin to modulate blood glucose levels and maintained consistent glycemic control throughout the six-month study period, with no evidence of fibrosis. In another experiment, the research team found that blocking a specific molecule—colony-stimulating factor-1 (CSF-1)—on macrophages, which are key immune cells involved in fibrosis, could inhibit scar formation. The addition of this blocker further improved graft survival rates.

Currently, several companies are developing encapsulated cell therapies. Among them, Sigilon Therapeutics is advancing technology developed at the Massachusetts Institute of Technology (MIT) to design treatments for diabetes, hemophilia, and a class of metabolic disorders known as lysosomal storage diseases; pharmaceutical company Eli Lilly is collaborating with Sigilon on diabetes research; Semma Therapeutics also has relevant technologies targeting diabetes; Neurotech Pharmaceuticals has conducted implantation procedures in clinical trials for glaucoma and various ocular diseases characterized by retinal degeneration; and Living Cell Technologies is conducting clinical trials of grafts for Parkinson’s disease while developing treatments for other neurodegenerative disorders.

Currently, cells encapsulated in “capsules” are generally derived from animals, human cadavers, or human stem cells. In the future, implantable cell therapies may encompass a broader range of cell types, including those engineered through synthetic biology. Synthetic biology confers new functions upon cells by recombining their genes, such as enabling the on-demand release of specific drug molecules into tissues. Although these studies are still in their early stages and the safety and efficacy of encapsulated cell therapies have not yet been validated in large-scale clinical trials, existing findings indicate that this field holds substantial potential.

Startups such as Mosa Meat, Memphis Meats, SuperMeat, and Finless Foods are developing lab-grown beef, pork, poultry, and seafood. Investment in this sector has also been substantial. For instance, in 2017, Memphis Meats secured $17 million in funding from investors including Bill Gates and the agricultural company Cargill.

If this technology is widely adopted, lab-grown cultured meat can help avoid much of the cruelty and slaughter associated with conventional animal farming. It can also significantly reduce the environmental costs of meat production, as the process requires only the cultivation of cells rather than raising entire organisms.

Technicians first extract muscle samples from animals, isolate stem cells from the tissue, induce their extensive proliferation, and then differentiate them into primitive fibers to form muscle tissue. Mosa Meat states that a single tissue sample extracted from one cow can yield enough muscle tissue to produce 80,000 quarter-pound beef portions.

Some startups have stated that they expect related products to hit the market in the coming years. However, for this “cultured meat” to achieve commercialization, it must overcome certain challenges, such as cost and taste. In 2013, journalists reported that the patty of a hamburger made from laboratory-cultured meat cost over $300,000 to produce. Moreover, the meat contained too little fat, resulting in an excessively dry texture. Since then, costs have begun to decline; a report released by Memphis Meats this year indicated that the price of a quarter-pound of cultured ground beef was approximately $600. Given this trend, cultured meat could become a strong competitor to conventional meat in the next few years. Attention to meat texture and the appropriate addition of other ingredients can effectively address issues related to mouthfeel.

To gain market acceptance, cultured meat must be proven safe for consumption. Although there is currently no evidence that lab-grown meat poses health risks, the U.S. Food and Drug Administration (FDA) has begun considering how to regulate it. Meanwhile, producers of conventional meat have responded by arguing that lab-grown products are not truly meat and should not be labeled as such. Surveys indicate that public interest in consuming lab-cultivated meat remains low. Despite these challenges, companies producing “cultured meat” continue to devote efforts to product development. If they can successfully create affordable products with authentic taste, “cultured meat” could help align our daily dietary habits more closely with ethical standards and environmental sustainability goals.

Electrotherapy—treating diseases through electrical pulses—has a long history in medicine, exemplified by cardiac pacemakers, cochlear implants, and deep brain stimulators for Parkinson’s disease. This approach involves delivering signals to the vagus nerve, which is responsible for transmitting impulses between the brainstem and most organs.

Research by Kevin Tracey and colleagues at the Feinstein Institute for Medical Research has demonstrated that the vagus nerve can release chemicals that help regulate the immune system, suggesting potential new applications for vagus nerve stimulation (VNS). For instance, the release of a specific neurotransmitter in the spleen can deactivate immune cells associated with inflammation. These findings indicate that VNS may serve as an effective treatment for conditions characterized by electrical signaling dysregulation, such as autoimmune diseases and inflammatory disorders, particularly given that existing medications often lose efficacy or cause severe side effects. VNS is better tolerated because it targets a specific nerve, whereas pharmaceutical agents typically distribute systemically, potentially disrupting tissues beyond the intended therapeutic target.

To date, research on the application of anti-inflammatory therapies has yielded significant results. The vagus nerve stimulation (VNS) device developed by SetPoint Medical has been proven safe and effective in early human trials involving rheumatoid arthritis (joint inflammation) and Crohn’s disease (intestinal inflammation). Currently, SetPoint Medical is conducting additional trials for these two conditions. Electroceuticals are also being applied to other inflammation-related diseases, such as cardiovascular disease, metabolic disorders, dementia, and autoimmune conditions like systemic lupus erythematosus characterized by vagal hypoactivity. Furthermore, preventing immune rejection of transplanted tissues represents another potential application.

Most vagus nerve stimulators, including SetPoint’s devices and those used to treat epilepsy and depression, are classified as implants. Physicians typically implant these devices subcutaneously in the clavicular region. The leads from the implant are wrapped around a branch of the vagus nerve and deliver electrical pulses at preset intervals; frequency and other parameters are configured via an external electromagnetic programmer. Current implants generally have a diameter of approximately 1.5 inches, a size expected to decrease in future generations, which will also feature enhanced programmability.

Although the mechanism by which vagus nerve stimulation alleviates these symptoms remains unclear, the recent FDA approval of non-invasive handheld vagus nerve stimulators for the treatment of cluster headaches and migraines underscores regulatory endorsement of this technology. These handheld devices deliver mild electrical stimulation to the vagus nerve transcutaneously through the skin of the neck or via the ear.

Novel electronic therapies are not limited to targeting the vagus nerve. In late 2017, the FDA approved a non-invasive device that transmits signals through the skin behind the ear to branches of the cranial and occipital nerves, thereby alleviating opioid withdrawal syndrome. The device received FDA approval after demonstrating a reduction in symptom severity of more than 31% among 73 patients with opioid withdrawal syndrome.

The costs of implants and surgery may hinder the widespread adoption of VNS therapy, although this issue is expected to be mitigated as the technology becomes less invasive. However, cost is not the only challenge; researchers still need to gain a deeper understanding of several key aspects, including how vagus nerve stimulation exerts its effects in each specific case and how to determine the optimal stimulation pattern for each patient. Furthermore, electrical pulses delivered to the vagus nerve may also have adverse effects on surrounding nerves.

However, as more research and trials are conducted, electronic therapies such as VNS are expected to enable better management of most chronic diseases, reducing the medication needs of millions of patients.

Research into a genetic engineering technology is advancing rapidly, one that can permanently alter the traits of a population or even an entire species. This approach leverages gene drives, which are selfish genetic elements that spread rapidly through populations. Although gene drive processes occur naturally, they can also be engineered through genetic modification, offering numerous benefits to humanity. This technology can prevent insects from transmitting diseases, boost crop yields by modifying plant pests, enhance coral resilience to environmental stress, and stop invasive plants and animals from disrupting ecosystems. However, researchers recognize that altering or even eradicating a species could have profound consequences. Consequently, they are establishing regulatory frameworks to govern gene drives, spanning from laboratory research to clinical trials and broader applications.

For decades, researchers have been exploring how to leverage gene drives to combat diseases and other challenges. In recent years, the advent of CRISPR gene-editing technology has accelerated this research, making it easier to insert genetic material into specific locations on chromosomes. In 2015, several papers reported successful trials of CRISPR-based gene drives in yeast, fruit flies, and mosquitoes. One experiment successfully drove an anti-Plasmodium resistance gene through mosquito populations, which theoretically should limit the transmission of the parasite. Another study successfully altered the female reproductive capacity of a different mosquito species.

In 2018, researchers tested a CRISPR gene drive system in mice, aiming to manipulate their coat color. However, they found that the system was effective only in females. Even so, the findings supported the possibility that this technology could help eradicate or alter invasive mouse or other mammalian populations, which threaten crops and wildlife or spread diseases.

The U.S. Defense Advanced Research Projects Agency (DARPA) has invested $100 million in gene drive research aimed at combating mosquito-borne diseases and invasive rodents. The Bill & Melinda Gates Foundation has invested $75 million in an institution to research gene drives targeting malaria.

Although the current landscape in this field is promising, gene drives have raised significant concerns. Could they inadvertently harm or disrupt other wild species? What are the risks associated with eliminating selected species from ecosystems? Could malicious actors weaponize gene drives, thereby impacting agriculture?

To prevent these alarming scenarios from occurring, a research team has developed a “switch” that must be activated by administering a specific substance before the gene drive can take effect. Meanwhile, numerous scientists are working on frameworks to guide the progression of gene drive testing at each stage. For instance, in 2016, the U.S. National Academies of Sciences, Engineering, and Medicine reviewed this research and provided recommendations for its implementation. In 2018, a large international working group established a framework to manage the entire process from laboratory research to public release. The group specifically highlighted the application of gene drives for malaria control in Africa, stating that if this technology were implemented, people in African regions would benefit significantly.

In addition to mitigating the risks inherent to the technology itself, many researchers also seek to avoid incidents or errors that could trigger public or policy backlash. In a 2017 paper on using gene drives to eradicate invasive mammals, Kevin M. Esvelt of MIT and Neil J. Gemmell of the University of Otago in New Zealand stated that such backlash could set back research by a decade or more. For malaria alone, delays in research could result in millions of preventable deaths.

In 2007, Harry A. Atwater of the California Institute of Technology predicted in Scientific American that his so-called “plasmonics” technology could yield a range of applications, from highly sensitive biosensors to invisibility cloaks. A decade later, various plasmonic technologies have become commercial realities, while others are transitioning from the laboratory to the marketplace.

The underlying principle of these technologies primarily involves controlling the interaction between electromagnetic fields and free electrons in metals (typically gold or silver), which determine the electrical conductivity and optical properties of the metal. When illuminated by light, free electrons on the metal surface undergo collective oscillations, forming what are known as surface plasmons. In bulk metals, free electrons reflect incident light, causing the material to appear lustrous. However, when metal particles are reduced to the nanoscale, their free electrons are confined within an extremely small space. The specific oscillation frequency depends on the size of the metal nanoparticles, resulting in a limited range of resonant frequencies. During resonance, surface plasmons absorb only incident light that matches their oscillation frequency while reflecting the rest. This phenomenon, known as surface plasmon resonance, can be leveraged to fabricate devices such as nanoantennas and high-efficiency solar cells.

One of the optimal applications of plasmonic materials is in sensors for detecting chemical and biological agents. Researchers coat plasmonic nanomaterials with a specific substance capable of binding to target molecules, such as bacterial toxins. In the absence of toxins, light incident on the material reflects at a specific angle. However, if toxins are present, they alter the surface plasmon resonance frequency, thereby changing the angle of the reflected light. These changes can be measured with high precision, enabling the detection of even trace amounts of toxins.

Several startups are developing products based on this technology, including internal battery sensors that monitor battery performance to help improve power density and charging rates. Another device can differentiate between viral and bacterial infections. Plasmonics is also applied to magnetic storage on disks. For example, heat-assisted magnetic recording (HAMR) devices increase storage capacity by heating tiny spots on the disk at the moment of writing.

In the field of medicine, researchers are testing the ability of light-activated nanoparticles to treat cancer in clinical trials. The nanoparticles are injected into the bloodstream and subsequently accumulate within tumors. Irradiating the tumor with light at a frequency matching the surface plasmon resonance causes the particles to generate heat through resonance. This heat can kill cancer cells within the tumor without damaging surrounding healthy tissue.

An increasing number of new companies are turning their attention to plasmonic technology. They will need to ensure that their products are reasonably priced, reliable, durable, mass-producible, and integrable with other devices. Despite these ongoing challenges, the prospects for the field remain broad. The emergence of metamaterials—synthetic nanomaterials that enable plasmonics to produce unusual optical effects—has allowed plasmonics researchers to use materials beyond gold and silver, such as graphene and semiconductors. Research from Future Market Insights predicts that the North American market for plasmonic sensors will grow from nearly $250 million in 2017 to nearly $470 million by 2027.

In the coming years, quantum computers are expected to surpass classical computers, driven by advancements in related hardware and algorithm domains.

Quantum computers leverage quantum mechanics to perform computations. Their fundamental unit of computation—the qubit—is analogous to the classical bit (0 or 1). However, it exists in a quantum superposition of two computational states: it can be both 0 and 1 simultaneously. This property, along with their distinctive quantum entanglement, enables quantum computers to solve certain types of problems more efficiently than any classical computer.

Although this technology is exciting, it is highly susceptible to interference. For instance, decoherence can disrupt its functionality. Researchers have found that quantum computers with thousands of qubits can address the issue of decoherence through quantum error correction techniques. However, to date, the largest quantum computers—such as those developed by laboratories including IBM, Google, Rigetti Computing, and IonQ—contain only dozens of qubits. John Preskill of the California Institute of Technology has termed these devices “Noisy Intermediate-Scale Quantum (NISQ)” computers, which currently lack error-correction capabilities. Nevertheless, extensive research dedicated to developing algorithms for NISQ devices may enable them to perform certain computational tasks more efficiently than classical computers.

The growing access to NISQ machines by users worldwide has significantly accelerated the development of this technology, enabling an increasing number of researchers to develop and test small-scale programs on such devices. An ecosystem of startups focused on quantum software is gradually taking shape.

Researchers have identified two types of algorithms for NISQ devices: simulation algorithms and machine learning algorithms. In 1982, the renowned theoretical physicist Richard Feynman proposed that one of the most powerful applications of quantum computers is simulating natural processes involving atoms, molecules, and matter. Many researchers have developed algorithms to simulate molecules and matter on NISQ devices (as well as on fully error-corrected quantum computers). These algorithms can enhance the design of new materials and find applications in fields such as energy and health sciences.

Developers are still evaluating whether quantum computers are better suited for machine learning tasks, in which computers learn from large datasets. Algorithm tests on NISQ (Noisy Intermediate-Scale Quantum) devices have shown that quantum computers can indeed improve tasks such as machine learning and information classification, and generate new statistical samples. At least three research groups have highlighted the approach of Generative Adversarial Networks (GANs), which has had a significant impact on the field of machine learning in recent years.

Although many algorithms function properly on existing NISQ devices, no one has yet provided a formal proof demonstrating that they are more powerful than algorithms executed on classical computers. Such a proof is exceedingly difficult to establish and may take several years to complete.

In the coming years, researchers are likely to develop larger and more controllable NISQ devices, as well as error-corrected quantum computers with thousands of physical qubits. Researchers in related algorithms believe that NISQ algorithms are sufficiently effective and hold promise for surpassing state-of-the-art classical computer algorithms. Although error-corrected machines are still required, the prospects in this field are very broad.

【Report Link】

http://www3.weforum.org/docs/Top10_Emerging_Technologies_report_2018.pdf