🦟 Mosquitoes Are a Choice, and Singapore Chose Less Dengue

Somewhere tonight, a mosquito is deciding whether you are dinner. Turns out that in Singapore, at least, the humans got a vote too. This week: a city-scale trial that made dengue much rarer, an AI that taught mRNA vaccines to handle heat, a weekly insulin, jellyfish with hats and job prospects, and a Labcorp bake-off of needle-free blood gadgets. Grab a coffee. Swat responsibly.

RESEARCH & NEWS

Every few years the US rediscovers that mosquitoes ignore borders. In a new Works in Progress essay, writer Maya Rosen argues that mosquito-borne disease is now a choice in the United States, a country where CDC counted a record 3,798 dengue cases in 2024, 359% above the 2010-2023 annual average. Her villain is not the insect. It is fifteen years of regulatory ping-pong.

Her best exhibit is Singapore. The National Environment Agency split 15 clusters into 8 that received male Aedes aegypti carrying Wolbachia and 7 that served as controls. The strain was wAlbB. When those males mate with wild females, the eggs never hatch. The males were also irradiated to sterilize any stray females that slipped through sex-sorting, so Wolbachia isn't meant to settle in permanently.

Think of it as flooding the dating pool with charming, entirely infertile suitors. In release zones home to 393,236 people, 6% of dengue tests came back positive after at least six months of exposure versus 21% in control zones, a protective efficacy of 72% (95% CI 58 to 82). Female mosquito trap counts fell from 0.18 to 0.041 per trap, while control areas climbed from 0.19 to 0.277.

To be fair, the trial was unblinded and had no placebo, and the authors warn that Singapore's intense background vector control means results may not generalize. Singapore is not waiting: NEA aimed to reach 800,000 households, half of all households in Singapore, by 2026, and its ministry now says the project will cover more than 50% of households by end-2026. Meanwhile, Rosen notes, the first full nationwide US registration for a live mosquito biopesticide only landed in April 2024.

Mosquitoes are a choice. Singapore picked. America is still on hold with customer service.

RESEARCH

The link in our inbox promised freeze-drying any protein. The paper behind it does something narrower and arguably more useful: it keeps mRNA vaccines alive without a freezer. Today's RNA vaccines need storage at -20 to -80 °C, which is a logistics nightmare anywhere the power grid has opinions.

The MIT team, with lead authors Jinbi Tian and Khanh Tran and senior authors Ana Jaklenec and Robert Langer, built an AI framework called AGENT that pairs high-throughput experiments with Bayesian optimization. Per the Nature Biotechnology paper, it optimized formulations in six iterations within one month, yielding water-free Moderna-like and Pfizer-BioNTech-like lipid nanoparticles that retained 100% bioactivity after more than two months at 37 °C.

The trick is picking excipients (the sugars, salts, and polymers padding the nanoparticle) without testing every combination. The team screened nearly 50 FDA-approved candidates, kept five, and let the algorithm propose ratios of two formulations at a time. Less genius chemist, more sommelier who only needs a few sips. "The real beauty of this algorithm is that we can use it with small data sets," Jaklenec told MIT News. Before the AI, Tran said, "nothing would get us to 100 percent stability."

The fine print: the particles were vacuum-dried, not freeze-dried, and everything is preclinical. In rodents and non-human primates, stored vaccines matched fresh intramuscular shots, and microneedle patches worked too. Coverage of MIT's release adds that the vaccines could remain effective at room temperature for up to a year, a figure not in the paper's abstract.

Proteins in general remain on their own. Your freezer keeps its job, for now.

NEWS

Daily basal insulin means about 365 injections a year. Lilly's pitch for its new weekly shot is cutting that to 52, which is a lot fewer awkward moments with a pen.

On September 24, the FDA approved Onswik (insulin efsitora alfa-gobe) for adults with type 2 diabetes, injected once weekly on the same day, calling it the second once-weekly insulin approved for that group. The first was Novo Nordisk's Awiqli, approved in March 2026 after a July 2024 complete response letter over manufacturing questions and its proposed type 1 diabetes use. Lilly calls this Onswik's fourth global approval, after regulatory actions in Europe, Mexico, and Japan.

The data come from Lilly's QWINT program: more than 3,400 adults across four phase 3 trials, each meeting non-inferior A1C reduction versus daily insulin glargine U-100 or degludec U-100, or 3,439 if you add up the four trials. Participants ranged from insulin-naive adults in QWINT-2 to 730 people already on basal-plus-mealtime regimens in QWINT-4. Think of it as trading a daily espresso for a slow-drip cold brew that lasts all week.

The catch is the usual insulin catch. Hypoglycemia is the most common side effect; the label warns about medication errors and accidental overdose, and the FDA says Onswik is not recommended in type 1 diabetes. That tracks with QWINT-5, where level 2 and level 3 hypoglycemia ran higher with efsitora than with degludec in type 1. The trials were open-label, and we found no head-to-head against Awiqli.

Two weekly insulins in one year. The daily shot is not dead, but it is updating its LinkedIn.

RESEARCH

Last month it was cyborg cockroaches. This week the biohybrid beat goes underwater, with an honest label: the core paper is from 2024. Caltech engineers Simon Anuszczyk and John Dabiri showed a moon jellyfish with an electronic swim controller and a 3D-printed "forebody" hat swimming 4.5 times faster than natural jellyfish while carrying a payload up to 105% of its own body volume, published in Bioinspiration & Biomimetics in February 2024.

Why jellyfish? They live from the tropics to the poles and from the surface to below 10,800 meters, and they are among the most energy-efficient swimmers known. The implant zaps the bell muscle to set the pace. The forebody works like an arrow tip, cutting drag while giving sensors a seat. Testing happened in a purpose-built six-meter, 13,600-liter vertical tank, with just three jellyfish in the long-distance runs.

The pitch is cost. Dabiri puts it at about $20 per jellyfish, a rounding error next to research vessels that can cost more than $50,000 a day. Moon jellies lack a central nervous system and pain receptors, and the team worked with bioethicists, so this is less Frankenstein and more a very cheap submarine with a pulse.

The current hook: a December 2025 Biomimetics paper ran field trials in the ocean, finding that removing the jellyfish's own pacemaker organs (rhopalia) made them far more obedient, and that faster species like Cassiopea and the box jellyfish Alatina alata sped up vertical profiles. Grad student Noa Yoder has described a deep-sea version using pressurized glass spheres. A March 2026 conference talk described a full-depth-rated attachment whose depth readings checked out in field tests to 1,700 meters, though the jellyfish itself only swam shallow profiles. A peer-reviewed deep-ocean dataset is still on the to-do list.

The jellyfish, notably, did not apply.

RESEARCH

We were promised that blood draws are outdated. The paper behind that headline is more modest and more useful: a head-to-head of the gadgets trying to replace the phlebotomist. Labcorp researchers led by Bradley Collier compared four upper-arm capillary devices, a fingerstick, and a standard venous draw in 41 healthy volunteers, published in Bioanalysis in October 2025.

The contenders were RedDrop One, Tasso+, YourBio Health's TAP MicroSelect, and Labcorp's own Comfort Draw. Each lances the upper arm, a spot with a low density of pain receptors, then uses gentle vacuum to coax blood into a small tube. Think of a tiny, polite leech, three of which carry FDA 510(k) clearance.

Volunteers rated the arm devices 0.5 to 1.8 out of 10 for pain, versus 2.5 for fingerstick and 1.5 for venipuncture, and picked an arm device first 71.8% of the time. Nobody's favorite was the fingerstick. Apparently, RedDrop One scored slightly worse than a vein. Lab results lined up with venous blood similarly across devices, though every capillary method except Tasso+ produced more hemolysis than a vein, and TAP collected the least blood.

Now the methods section. All 41 volunteers were Labcorp employees, trained professionals took every sample (no home self-collection), and the authors work for the company behind one of the devices. They found no single winner. A separate review notes potassium and platelets remain tricky for arm-based tools. And 28.2% still ranked the plain old vein first.

Our sweat ring and tear lens stories tracked a molecule or two. This is about a full lab panel without a vein. Progress, not a funeral.

Every story this week is really about the last mile. Wolbachia already works; the holdup is a regulator's inbox. mRNA vaccines already work; the holdup is the freezer. Insulin already works; the holdup is 365 pokes a year. Ocean sensors already work; the holdup is getting them thousands of meters down cheaply. Blood tests already work; the holdup is the vein. Turns out a lot of modern biotech is not inventing the thing, it is getting the thing to where it is needed.

And all of them carry their own fine print: an unblinded trial, animals not people, open-label comparisons, three jellyfish, 41 employees. The last mile is hard. So is the evidence.

Which holdup would you fund first? Hit reply, we read everything. And if a friend forwarded this, return the favor.

Keep questioning everything (especially anything buzzing near your ear at 3 a.m.),

Prateek & Jere

P.S. Singapore needed two years of twice-weekly releases and a cluster-randomized trial to end up with 85% fewer mosquitoes than control areas. Here in Finland we just wait for October.