#273: Using a Brain-Computer Interface, a Quadriplegic Typed at the Same Speed That Most People Write, & More

A brain-computer interface (BCI) allows individuals to control computers with their minds. In a BCI experiment at Stanford, researchers doubled the speed at which a paralyzed individual typed, from 9 to 18 words per minute, the speed at which people write by hand. Previous BCI experiments instructed participants to imagine moving cursors and selecting letters on keyboards. The Stanford experiment asked a participant to imagine writing letters with his hand and then used neural networks to decode his brain waves. The system worked better because each handwritten letter is unique and “has a different pen trajectory,” allowing neural nets to decipher intended letters more easily. This experiment demonstrates that humans and neural networks can and should work together.

The experiment also could be pointing to the promise of Neuralink. Neuralink is increasing the number of implantable brain electrodes by an order of magnitude, adding dramatically to BCI possibilities.

Researchers published the first genome wide association study (GWAS) in 2005, just a few years after the completion of the Human Genome Project. Throughout the 2010s, improvements in next-generation sequencing (NGS) increased the scalability, breadth, and cost-effectiveness of GWAS studies. Based on thousands of sequences, these studies aim to discover genetic variants associated with traits and diseases. People with early-onset heart disease, for example, might have common genetic variants with statistically meaningful associations.

Despite recent advancements, GWAS findings have translated poorly into clinical practice. Biostatisticians use GWAS data to compute polygenic risk scores (PRS) measuring the ‘hot spots’ of genetic variation that yield overall ‘risk scores’ for conditions like cardiovascular disease and cancer. That said, moving PRS into the clinic has been challenging, a topic on which ARK previously has written. One reason is that PRS rely on the statistical associations between the genome and diseases: they do not consider the underlying mechanisms of disease that often are controlled by proteins.

We have noticed that the convergence between and among several technologies – mass spectrometry sample prep, next-generation protein sequencing (NGPS), and protein affinity assays – are taking the same path as NGS during the early 2010s. Once again, they are enhancing the scalability, breadth, and cost-effectiveness of analyzing all human proteins—the proteome.

Though only a ripple today, a wave of proteogenomic studies – population-scale analyses of both genome and proteome – is likely to enhance genomics research during the next few years. Instead of limiting research to statistical associations, these studies are likely to provide real insights into the functional mechanisms of disease, giving PRS the opportunity to impact biologic discovery and, ultimately, clinical practice.

The global stablecoin market has increased 10x in the last 12 months, crossing $100 billion in total market cap according to The Block.

Stablecoins, crypto assets which track the price of fiat currencies, offer users a familiar unit of account to facilitate transactions on open blockchains. As a result, their role in trading on and between crypto exchanges has increased dramatically, particularly outside the US, we believe for several reasons: limiting the need for access to US bank accounts, offering near instant settlement, and subjecting transactions to fewer regulatory requirements. In fact, recently stablecoins have outpaced the USD and bitcoin in pair trades on exchanges globally.

Often issued as tokens on the Ethereum blockchain, stablecoins also are enabling decentralized finance. They account for a majority of assets in Compound’s decentralized lending market which offers deposit yields much higher than those available in the traditional financial system.

As stablecoin market caps scale, competition among their issuers appears to be intensifying. Tether dominated the early stablecoin era but is losing share to newer entrants such as USDC which has been issued by companies such as Coinbase and promises a higher level of regulatory compliance and integration into the existing financial system.

Previously, ARK estimated that, at scale and maturity, autonomous ride-hail operators could price the service at just 25 cents per mile, covering both costs and a small profit. In the early days of service, operators probably would charge Uber-like prices close to $2 per mile, introducing lower price tiers after penetrating the premium paying consumers.

While ARK’s research thus far has broached the autonomous ride-hail market from the operator’s perspective, this week we examined the prices that consumers would willing to pay. At what price does the average consumer value his or her travel time? Based on reports from the Department of Transportation and a recent study sponsored by Lyft, ARK estimates that consumers value an hour of travel time at roughly $18 per hour, or approximately 72 cents per taxi mile. In other words, the addressable market for a service priced at 72+ cents per mile would be double our original estimates. Based on the approximate 2 trillion vehicle miles travelled (VMT) annually in urban US areas, the total available market would approach $1.6 trillion.

In other words, while autonomous ride-hail operators could charge as little as 25 cents per mile and make a profit, average US consumers might be willing to pay much higher prices based on the value they place on their time, potentially translating into more attractive profit margins for autonomous ride-hail platforms.