Hello friends.

Some good news about Covid vaccine side effects, finally... Researchers at the University of Florida (yes, the land of vaccine scepticism and alligators) discovered that cancer patients who received an mRNA COVID vaccine around the start of immunotherapy lived nearly twice as long. Median survival jumped from 20 months to 37 months in lung cancer patients, which may not sound like much but in cancer terms, really is. 

The vaccine acts like a flare gun for your immune system, rallying cells away from tumours and into lymph nodes where they can mount a proper offensive. It's not targeting cancer specifically, just administering what amounts to a very aggressive pep talk. Surprisingly, the effect was strongest in patients whose tumours weren't expected to respond well to treatment. Lab tests in mice confirmed it: mRNA vaccines paired with immunotherapy turned resistant tumours into responsive ones.

🧐 What's in it for me? If confirmed in further trials, a mRNA booster could become standard alongside immunotherapy within five to ten years, hastened by the fact the infrastructure already exists.

💵 Out of the Lab: If mRNA vaccines prove to enhance immunotherapy at scale, the cancer treatment market could tilt heavily toward combination therapies, rewarding those positioned at both ends.

Generalised anxiety disorder affects one in twenty adults and conventional treatments are, to be generous, rubbish. SSRIs reduce symptoms by about 1.25 points on a 56-point scale which is roughly the therapeutic equivalent of putting a plaster on a broken leg. Enter MM120, pharmaceutical-grade LSD tested at UCSF, which reduced anxiety by 5-6 points beyond placebo. 

The drug promotes neuroplasticity, essentially giving your brain permission to stop running the same anxious thought loops like a broken record. Unlike SSRIs that require daily dosing and six weeks to maybe work, MM120 is a single supervised session with effects lasting up to 12 weeks. Side effects included hallucinations, which whilst fun inconvenient, are arguably better that the laundry list from SSRis. 

🧐 What's in it for me? If MM120 clears regulatory hurdles, it offers a genuine alternative for the millions who've concluded that SSRIs are little more than expensive placebos with side effects. The single-dose model is also appealing, though supervised administration means this won't be available via an app. Expect broader trials within two to three years.

💵 Out of the Lab: Psychedelic-assisted therapy is shifting from Burning Man to boardrooms. GAD represents a massive underserved market, and if MM120 succeeds, it could reshape mental health treatment toward periodic interventions rather than indefinite prescriptions.

Modern AI needs to process torrents of data in real time, but silicon chips are hitting their physical limits. You can't make transistors much smaller, you can't push electrons much faster, and adding more of them just generates more heat. The result is a bottleneck: AI models are getting hungrier whilst processors struggle to keep up. Optical computing sidesteps this entirely by replacing electrons with photons. Researchers at Tsinghua University built OFE2, a chip that uses beams of light instead of electrical signals to process numbers. The result? It performs calculations in 250 picoseconds (0.000000000001 seconds…), making it roughly 13,000 times faster than traditional processors.

Building on previously discussed photon technology, the breakthrough lies in data preparation. Previous systems struggled with phase stability when splitting light signals at high speeds, a problem roughly equivalent to conducting an orchestra where every musician is in a different time zone. The team solved this with an integrated on-chip system that keeps everything synchronised. Tests showed OFE2 could analyse CT scans and process live trading data faster than any electronic system, meaning traders could theoretically lose money at the speed of light.

🧐 What's in it for me? Optical processors could soon handle AI heavy lifting in healthcare imaging, financial trading, and real-time video analysis. For consumers, this means faster, more efficient AI without turning your phone into a pocket radiator. Commercial deployment likely within five to seven years if scaling challenges are addressed.

💵 Out of the Lab: If optical computing proves scalable, it can disrupt both AI hardware and high-frequency trading infrastructure. The winners will be those who can slot photonics into existing workflows without requiring data centres to be rebuilt from scratch.

Dry age-related macular degeneration affects 5 million people worldwide and has precisely zero approved treatments. It gradually destroys central vision, leaving patients unable to read, recognise faces, or do much beyond curse biology. Now, a European trial tested PRIMA, a 2mm microchip implanted beneath the retina that works with augmented-reality glasses to restore reading ability. After treatment 84% of participants could recognise letters and words, which in ophthalmology represents a red hot miracle.

The implant is thinner than a human hair and acts like a miniature solar panel. Patients wear glasses with a camera that captures scenes and beams them as infrared light onto the chip. AI algorithms convert this into electrical signals that travel through the optic nerve to the brain, which somehow makes sense of it all. After months of rehabilitation, patients learn to scan text and adjust zoom. One participant, Sheila, went from seeing only black discs to completing crosswords, which sounds modest until you consider she'd been functionally blind.

🧐 What's in it for me? If you or a loved one suffers from dry AMD, this could be life-changing within three to five years once regulatory approval broadens. The rehabilitation is significant, but for those facing blindness, learning to use prosthetic vision beats the alternative by quite some margin.

💵 Out of the Lab: Retinal implants open a new category in treating previously incurable blindness. If PRIMA gains broader regulatory approval, it creates a multi-billion market as ageing populations drive AMD prevalence higher.

Google's quantum processor, Willow, just solved a real-world physics problem 13,000 times faster than the world's fastest supercomputer, which is the computational equivalent of Usain Bolt racing your grandmother. Unlike their 2019 "quantum supremacy" claim involving an obscure task that classical algorithms eventually cracked anyway, this tackles genuine science: measuring how information spreads in highly entangled quantum systems.

The algorithm, Quantum Echoes, uses a time-reversal trick that sounds like science fiction but apparently works. It runs a quantum system forward, nudges it slightly, then reverses the process to create a "quantum echo" revealing how particles are linked. This measures an Out-of-Time-Order Correlator, which is fiendishly difficult for classical computers but straightforward for quantum systems. Entangled particles share states even when separated, making the system exponentially harder to model as more particles join the party. Classical supercomputers must track every possible state like a very anxious wedding planner; quantum computers process them all simultaneously like they've had several espressos.

🧐 What's in it for me? Practical quantum computing could revolutionise drug discovery, materials science, and climate modelling by simulating molecular interactions currently impossible to calculate. We're still years from consumer applications, but this suggests quantum computers might start solving genuinely useful problems within a decade rather than remaining expensive curiosities.

💵 Out of the Lab: If quantum computers begin consistently outperforming classical systems on practical tasks, the race to commercialise quantum hardware intensifies. Early movers in quantum software and cloud access could capture outsize value before hardware commoditises.

Three food studies flipped the wellness script this week.

However, before you start eating deep fried croissants, there’s a catch: 2 out of 3 of these studies include pretty massive conflicts of interest, and the last one is a lot less liberating than it sounds, so if you’d rather live in blissful ignorance, stop reading now. 

First, industrially processed fats appear harmless! [In a tiny study for a comically short period of time..]. The six-week trial tracking 47 participants found that interesterified fats in margarines and pastries show no negative effects on heart health. Arteries, they suggest, don’t care whether your fat was processed in a factory. And for the final nail in a delicious coffin, the research was supported by the Malaysian Palm Oil Board, which has roughly the same conflict of interest as a brewery funding research into whether beer is good for you.

Now onto bananas. They contain polyphenol oxidase, an enzyme that apparently destroys the heart-healthy flavanols in berries. Adding a single banana can supposedly reduce flavanol absorption by 84%, and as mentioned, this = bad. Whilst not as flimsy as the previous example, this study was partly funded by Mars which sells cocoa flavanol supplements, so make of that what you will. 

Lastly, some science that (as far as we can tell) isn't just an advert with lab goggles. It turns out that gluten sensitivity is mostly performance art. People blaming gluten are actually reacting to fermentable carbohydrates (FODMAPs) or other wheat proteins naturally found in bread. Under controlled conditions, reactions to gluten were identical to placebo.

The catch? FODMAPs and other wheat proteins are naturally present in most gluten-containing foods like bread and pasta, making them nearly impossible to separate in practice. However, some gluten products might be tolerable: pure vital wheat gluten and seitan. The real takeaway is that millions avoiding gluten might do better on a low-FODMAP diet instead, which could ironically allow some gluten-containing foods whilst restricting others based on FODMAP content rather than gluten itself.

Until next time, stay curious.

Like what you're reading? Share toast with a friend.