The Global Grand Challenge of Our Time
Science and Technology Breakthroughs in 2021
Though the COVID-19 pandemic still weighed heavily upon us, science and technology forged ahead in 2021. Over the past year we witnessed a spectacular array of discoveries, and breakthroughs. This letter examines the year’s progress in science and technology through the lens of some of the toughest challenges before us. We begin with a focus on what many consider the greatest challenge of all time, understanding the human brain.?We then reflect on some of the progress made in other areas critical for human advancement, including space, energy, quantum computing, and gene editing. The letter concludes with a presentation of the innovation of the year.
Unlocking The Mystery of the Human Brain
How do we imagine…, abstract…, assess…, reason…., create…, love…, and hate? How do we feel without touch? How do we hear without sound? How do we see without sight? How does intuition come about? How are we conscious and self-aware??Unlocking the mystery of the human brain is not only the challenge of our time, but perhaps the greatest challenge of all time.
In 1889 the Spanish pathologist Santiago Cajal unveiled the first look at the architecture of the brain. He discovered that the brain is not a network of continuous filaments as previously thought, but rather a complex network of billions of brain cells with an interconnection network between them. The figure below is Cajal’s original sketch of the Purkinje neuron in the cerebellum [Cajal Institute].
130+ years later, we are still trying to understand how the brain truly works. Substantial advancements have been made, but at its core the human brain remains a mystery.
So how do we get there? If we were to decipher an enigma-type machine, we would begin by learning all of the parts and how they are connected (internal organization). But that alone would not solve the puzzle. We would still need to somehow decode the language it uses – try different input combinations, observe the output patterns, and then try to classify them. Current R&D approaches for deciphering the neural code are not much different. Scientists are working along three dimensions:
?Let’s take a deeper dive into each of these dimensions vis-à-vis progress in 2021.
?The quest is on to map the human brain: It is believed that the human brain has about a thousand different types of cells. To truly understand how the brain works, we need to understand each of these types of cells, how they connect and how they communicate with each other. In 2017, the NIH’s Brain Initiative launched a major effort, Brain Initiative Cell Census Network (BICCN), to discover and develop a complete catalog of the brain. In 2021, BICCN reported its most significant milestone since its inception. BICCN published 27 research papers in Nature depicting a comprehensive catalog of the motor cortical cell types in mouse, monkey, and human brains. With contributions from 258 authors, the flagship paper provides important biological insights into the brain and is expected to pave the way towards a deeper understanding of the relationships between brain structure and function. Key takeaways from the paper include:
Our traditional concept of the brain’s functional organization may be wrong: For decades, brain science orthodoxy has held that brain function is associated with distinct regions separated by well-established boundaries. For example, language and memory happen in the temporal lobes while perception is processed in the somatosensory cortex. Much like a map of countries separated by borders, function is assigned to physical location. In fact, our very definition of brain function is based on the way we conceptualize things such as vision, memory, feeling, and motor control. This may have biased the way we analyze the brain. Recent studies have shown that such division of labor across anatomical areas is just too simplistic. Brain function is more of a system of neural activities across classically defined regions. For example, Stanford University scientists have shown that neural activity associated with memory recall doesn’t directly map exclusively to the memory region but rather across several regions of the brain. The research suggests that rather than map function to regions, perhaps the better approach is to map psychological tasks to patterns of neural activity. The neural activity data can then be classified with machine learning. In a 2021 paper published in Nature, a team from Stanford University describe a novel framework for mapping domains of neurobiology based on such a data driven computational approach. This is a paradigm shift in the way brain function is considered and could open the door to the development of drugs that are customized to each patient’s needs.
Next generation Brain Computer Interfaces (BCIs) will change the world: Also called Brain Machine Interfaces, BCIs have been around for decades, helping the differently abled communicate and control prosthetic devices. UCLA’s Jacques Vidal coined the term BCI in 1973 and by 1997 the FDA approved deep brain stimulation for the treatment of tremor and Parkinson’s disease. Innovation and VC funding in BCI technology is accelerating. Below are some of the most disruptive next generation BCIs.
?Brain on Bluetooth - move things with your mind: Just imagine the possibilities! Everything from controlling all of your digital devices to having robots do what you want them to do by just thinking about it. We are not quite fully there yet, but well on our way. In December 2021, Philip O’Keefe became the first patient to post a message on Twitter using only his thoughts. Mr. O’Keefe is an amyotrophic lateral sclerosis (ALS) patient and uses a BCI developed by Synchron to control digital devices.?Synchron’s Neurointerventional Electrophysiology platform (Neuro EP) is a unique technology based on the premise that one can both listen to and stimulate the brain through the blood vessels.?This is no less than a paradigm shift in BCI technology. There are 400 miles of blood vessels navigating the brain. An electronic stent incorporated into the walls of the blood vessels can record and stimulate every part of the brain. Brain signals are detected and then converted into digital information that can drive Bluetooth devices. A special software layer utilizes translated brain signals to control apps on digital devices. Synchron’s novel approach allows unprecedented access to billions of previously inaccessible neurons. In 2021, the FDA cleared Synchron’s Neuro EP for human trials.
Neurograins for your brain: BCIs which are surgically implanted are typically monolithic with input/output pins like a bed of needles. Neurograins constitute a completely different approach for interfacing with the brain. A network of coordinated smart sensors, each about the size of a grain of salt, records neural electrical activity and reports to a central hub. The smart sensor network can also be used to stimulate the neurons, thus providing two-way communication capability. The Neurograin technology was developed by a team of researchers from Brown University, Baylor University, UC-San Diego and Qualcomm. It can support up to 770 Neurograins distributed across the cerebral cortex providing insight into brain activity in unprecedented detail. Power to the Neurograins is supplied wirelessly. The technology was successfully tested on a rodent in 2021 and the results published in Nature Electronics.
High throughput access to neurons: Neuralink and Paradromics are leading the way with implantable monolithic BCIs with large arrays of channels capable of reading from or stimulating the brain with over a thousand tiny wires (with plans to scale to even larger arrays). In 2021, Neuralink wired up monkeys and recorded their neural activity. As the monkeys played with a joystick that moves a cursor in a video game, their neural signals were captured and decoded in realtime, mapping neural signals to hand movements. The joystick was then disconnected but the monkeys were still able to control the cursor through their thoughts. Neuralink is planning human trials for 2022. According to Elon Musk, the startup is initially targeting the treatment of spinal cord injuries and neurological disorders.
In 2017, DARPA launched an R&D program with the goal of developing neural implants that record brain signals with very high fidelity. The program has a lofty goal for BCIs that can read from one million neurons, write to 100,000 neurons and interact with 1000 neurons in full duplex (read and write). Under contract with DARPA, the startup Paradromics developed a neural implant chip in 2021 that achieves the necessary signal fidelity. It employs miniaturized packaging for seamless integration as a cortical implant. The chip has an array of 1600 channels.
Noninvasive Wearables: The above BCIs all fall into the invasive BCI category because they require surgery for implantation. Though they can provide more direct access to neural cell behavior, they need to be designed to seamlessly integrate as cortical implants while minimizing the possibility of infection. Thus, the application space may be more limited to helping those individuals with more critical disabilities. DARPA’s Next-Generation Nonsurgical Neurotechnology (N3) program was launched in 2018 to develop more accessible brain-machine interfaces that don’t require surgery. Neural activity is recorded using optical, acoustic, and/or electromagnetic technologies. For example, a team at Rice University is using light scattering in neural tissue to record neural activity, and focused ultrasound to write to neurons.
Wearable BCIs have not yet been widely adopted. The key issues include form factor and user experience. Neurable, a startup out of the University of Michigan, is creating an every-day BCI that can be widely accessible. Neurable launched a headphone in 2021 that analyzes electroencephalogram (EEG) signals to help people focus. It uses soft electrodes made of cloth. Machine learning algorithms are employed to recognize if the brain is being distracted. Future applications include simple mind control of devices such as advancing slides in a presentation. If widely adopted, the greatest benefit will be the data that is gathered and analyzed from all the users and its potential for continually advancing the science of neural pattern analysis and interpretation.
BCIs will undoubtedly make a tremendous difference for people that are differently abled, and they will do so at scale. For the world at large, just imagine the possibilities of seamless continuous monitoring and analysis of the brain, helping us learn better, play better, and even live better. A great application area for BCIs to watch for over the next five years is the Metaverse. The Metaverse with all its dimensions from gaming to education to communication will be a great field of discovery and a large data source for BCI R&D.
While there has been tremendous progress in 2021, unlocking the mystery of the human brain may be many decades away. However, breakthroughs in science and technology sometimes happen when least expected, and 2021 witnessed a number of important achievements that will impact our understanding of the universe and how we live for years to come…Breakthroughs in technologies such as quantum computing, general artificial intelligence, as well as breakthroughs in the biosciences will one day come together to solve the ultimate puzzle of the human brain.
领英推荐
Majestic Achievements and Innovation of the year
2021 was yet another amazing year for science and technology with some of the year’s greatest achievements occurring in outer space. The year started with a bang and ended in triumph.
The sounds of Mars, touching the Sun and the origins of the stars: In February, probes from three nations, the UAE, China and the USA reached Mars, with the UAE probe, Hope, scanning the red planet’s upper atmosphere. NASA’s Perseverance rover recorded the sounds of Mars, and a helicopter hovered in its skies sending spectacular pictures through Perseverance back to Earth.
On December 14, for the first time in history a space probe touched the Sun and sampled particles and magnetic fields in the Sun’s upper atmosphere, where temperatures reach millions of degrees Celsius. So why did it not melt? Because temperature is not the same as heat. Temperature is related to how fast particles are moving and heat is related to how much energy is transferred. If there are particles moving fast, meaning high temperature, but only a few of them, they would not transfer much energy, meaning not much heat. The Parker Solar Probe travels through the Sun’s corona but faces heat up to about 2500 degrees Celsius. It employs a carbon composite material which serves as its thermal shield.
?The year came to a close with NASA launching the majestic Webb telescope to see the first bright objects that were formed after the Big Bang, by analyzing light emanating from stars 13.6 billion years ago. Thirty years in the making, the telescope is no less than a gigantic engineering achievement.
?A big-time milestone for fusion energy: High temperature nuclear fusion, the reaction that powers the sun, will one day, in our lifetime, be the solution to one of the world’s biggest challenges – clean sustainable energy, which is critical for addressing climate change. Fusion occurs at 100 million degrees Celsius inside the reactor. This creates a tremendous challenge along several dimensions, including how to reach such incredibly high temperatures, and how to contain the reaction without destroying the reactor. Due to these challenges, most fusion experiments last under one second. In May, China’s Experimental Advanced Superconducting Tokamak (EAST) broke the record by reaching plasma temperatures of 120 million Celsius for over 100 seconds and even reaching 160 million Celsius for 20 seconds. On the last day of the year, EAST achieved a sustained plasma of 70 million degrees Celsius for 1056 seconds!
?These high temperatures are reached by means of a special plasma. To contain the heat, large superconducting magnets are used to isolate the plasma from the walls of the reactor. In September, MIT and Commonwealth Fusion Systems (CFS) announced the development of the world’s strongest fusion magnet, weighing 10 tons and ramping up to a field strength of 20 Tesla. To be practical, the fusion reactor must generate more energy than it receives – this is referred to as “ignition”. In August, the National Ignition Facility at Lawrence Livermore National Laboratory announced that its laser fusion achieved a significant 1.3 megajoules of fusion energy. This is excellent progress but still short of the 1.9 megajoules required for ignition.
In parallel with this substantial progress, funding of high temperature nuclear fusion has dramatically accelerated with over $2.3 billion raised in Q4, including $1.8 billion to CFS’s Series B financing. Money well spent indeed!
Quantum Computing – who will win the race: Four technologies are making strides towards the crown. Superconducting qubits, trapped-ions; cold atoms; and photon-based qubits. Each has pros and cons which I will not go into in this letter. Rather, let’s focus on how not to compare. Qubits are the fundamental computing elements of a quantum computer (QC). So, it is natural to think that company X with a high number of qubits is better than company Y with a lower number. It is natural but wrong. Qubits are error prone. So, if company Y’s QC has a lower error rate, it is likely that its QC is actually better. A key measure is to look at the number of logical qubits (error corrected qubits). A 1000 physical qubit QC may only be equivalent to a few logical qubit QC. In this context, the quality (fidelity) of the physical qubits as well as error correction capability of the QC come into play. There are other factors involved in a comparison, such as number and quality of quantum gates and combinations thereof.
In October, University of Mayland researchers along with IonQ published a paper in Nature demonstrating that a logical qubit can be achieved with as few as 13 physical qubits (nine to encode a single logical bit and four to detect/correct errors). This is a significant development that might pave the way for practical quantum computers at scale.?In 2021, IonQ became the first full stack QC company to publicly list its securities when it began trading on the New York Stock Exchange.[1]
?Another measure of a QC is quantum volume (QV) – a term coined by IBM. QV is a single number measuring how well a QC can run quantum circuits of increasing complexity. The higher the better. In June 2020, Honeywell announced its first QC achieved a QV of 64 and by September of 2020 it doubled to 128. In July 2021, Honeywell together with Cambridge Quantum Computing achieved a world record QV of 1024.
?It was the year of the chip: 2021 marked the year that the world remembered to appreciate the semiconductor sector as a key driver of the global economy. Fueled by the impact of COVID-19 pandemic on work, education, and lifestyle, as well as crypto mining, demand for chips skyrocketed in 2021. PC demand alone increased by 13%.?Foundries around the world could not keep up with demand. Though chip shortages caused major pain points across various sectors of the economy, semiconductor technology continued to advance. In May, IBM revealed the world’s first 2nm chip using its nanosheet technology. This technology allows for the production of up to 50 billion transistors on a chip the size of a fingernail, with each transistor having the height of roughly the thickness of two strands of DNA. The technology can be used to either improve chip performance by 45% or reduce power consumption by 75% depending on how the chip is optimized. Consequently, smart phone battery single charge life can be quadrupled, i.e. you would only need to recharge your phone once every four days. The semiconductor industry is thriving again and will continue to do so over the coming years.
?CRISPR treats the untreatable: CRISPR-Cas9, the 2020 Nobel Prize-winning gene editing technology, is now being deployed directly in the human body to treat rare disease. Transthyretin amyloidosis is a deadly disease characterized by a buildup of a rogue version of a protein in organs and tissues. ?In a landmark study published in the New England Journal of Medicine in 2021, a CRISPR-Cas9-based drug targeting the rogue protein was infused into the bloodstream of 6 patients. The level of the targeted protein was reduced in all patients and by as much as 96% in one of the patients. In another major contribution, a paper by researchers at UCSF published in Cell, outlined a method to extend CRISPR’s architecture from the genome to the epigenome which controls the on/off switching of genes. The scientists were able to switch off a gene without having to edit the genetic code. The gene then remains inert for generations of the cell’s descendants.
Innovation of the year: Based on overall impact in advancing science/technology in service of humankind, the innovation of the year for 2021 is a tie between RoseTTAFold and AlphaFold2. Last year we highlighted AlphaFold2 for making significant progress towards solving one of biology’s most significant challenges: predicting the 3D shape of proteins from their amino acid sequences.?Proteins are the foundational worker bees of our body. They fight invaders (antibodies are proteins), enable metabolic processes to convert food into energy, repair tissue, maintain the structural integrity of our cells, synthesize hormones, maintain proper pH balance, act as enzymes for chemical reactions, and much more.
They say in biology that structure determines function. Proteins fold into 3D shapes and these shapes determine their function and how they interact with other proteins. Proteins that misfold can cause disease. For example, a buildup of misfolded Tau proteins in the brain has been associated with Alzheimer’s. Drugs are designed to modify the activity of proteins. Understanding the 3D structure of proteins and how they react with other proteins is essential for drug development.
On July 15, the University of Washington announced RoseTTAFold, a deep learning breakthrough tool that predicts protein structure from limited information in as little as 10 minutes on a gaming computer. RoseTTAFold was developed at the Baker Lab by Minkyung Baek, a postdoctoral scholar at the Lab. The tool generated structures for many proteins that were previously not well understood, as well as proteins directly associated with disease and disorders. ?RoseTTAFold is openly available to researchers through Github. In a single month alone, over 4,500 proteins were submitted. In August, Deepmind unveiled the structure of nearly 45% of all known human proteins, and in October identified 4,433 protein-protein complexes and how they interact.?The next step for both RoseTTAFold and AlphaFold2 is to model how proteins alter their shape while doing their work. RoseTTAFold and AlphaFold2 are truly transformative breakthroughs that will change medicine and create better more effective drugs for humanity.
Conclusions: The human brain, the organ behind every discovery and innovation, and the essence of civilization, is the most complex entity known. Unlocking the mystery of the human brain may lead to advancements in science and technology that we cannot even imagine. Perhaps one of the greatest attributes of the human brain is its resilience and adaptability. In spite of the COVID-19 pandemic, our society proved more resilient than ever in 2021, accelerating innovations and breakthroughs in science and technology. We witnessed significant progress in critical areas such as high temperature fusion, quantum computing and gene editing. This progress will continue to accelerate in 2022. The year is already off to a great start with the world’s first successful xenogeneic heart transplant - a historic achievement.
2022 will undoubtedly witness many more advancements in science and technology. Will this be the year that we finally see a breakthrough in battery R&D that takes electrification to the next level? Will this be the year that we take a significant step towards general AI? Will this be the year that we reach a significant milestone towards sustaining ignition in high temperature fusion??We shall see…
Acknowledgements
Many thanks to Natalia Moreno, Zeena Ojjeh, Howard Caro, Victor Zhirnov and Ibrahim Ajami for their help and suggestions.
References
[1] (NYSE: IONQ) In the interest of full disclosure, IonQ is a Mubadala portfolio company, and is also the first pure-play quantum computing company to publicly list its stock.
Ok to lack 3D vision. Need better eco-friendly vehicles around. When I was walking up/down hills, the scene of the busy road going down ???? is so flat right there. Taken a while to get used to the train and sitting nicely after such jogging. Now I drive a small manual car without an issue
Chief Risk Flaneur @ Private Market Fund
2 年Raphael Douady , Adil Abdulali , Philippe Chambadal , Franz Junior Bergman