Hastrup Review of Science

← All episodes

Episode 1210 min

Diet, Genes and Grip: What the Trials Actually Show

Six new papers, honestly explained: Diet, exercise and muscle · Heavier than your genes · Grip, gait and diabetes · Semaglutide, alcohol and smoking · Two heavy black-hole mergers · Distilled models for electrolytes.

MP3

Chapters

The papers

Transcript

The script the two AI voices read from. The times are worked out from the script, not heard, so a line can start a few seconds early or late.

Read the whole episode

Sponsor

Host 1Support for the Hastrup Review of Science comes from RefurbMiner.

Host 1One intelligent platform for your whole business. Invoicing, banking, and compliance in a single system, with every decision fully traceable. Privacy ready, and hosted in Europe.

Host 2Start your free fourteen day trial today. The link is in the show notes.

Cold open

Host 1Welcome to the Hastrup Review of Science — where artificial intelligence reads real science, so you don't have to pretend you understood the paper.

Host 1This week's lead: adding exercise to a diet adds some weight loss, but the trials do not show that it protects your muscle.

Host 2That assumption sits under a great deal of fat-loss advice, so it is worth seeing how thin the evidence for it is.

Diet, exercise and muscle

Host 1Start there. What did they do?

Host 2A preregistered meta-analysis of randomised trials in adults with overweight or obesity, twenty-three reports from twenty-one trials. The key design choice is that each comparator stayed separate, so diet plus exercise against diet alone is never pooled with diet plus exercise against exercise alone.

Host 1And against diet alone?

Host 2Across sixteen reports, combined programmes lost about four and a half kilograms more, with a confidence interval from roughly minus eight to minus one kilograms. Heterogeneity was ninety percent, and the prediction interval runs from about minus seventeen to plus eight kilograms, so a new trial could plausibly find no benefit.

Host 1What about fat-free mass?

Host 2No comparison showed reliable preservation; the pooled estimates were slightly negative, under one kilogram, and compatible with zero. Certainty was rated very low for every outcome, and the review cannot say which diet or which kind of exercise works best.

Host 1So the honest reading?

Host 2Exercise probably adds modest weight loss on top of dieting, but muscle sparing is not established by these trials. That is a gap in the evidence, not proof that exercise fails to protect muscle.

Host 1The full story is in this paper's deep dive, out this week in the feed.

Heavier than your genes

Host 1Next, weight and genes. A Finnish team asked what it means to weigh more or less than your genes predict.

Host 2They used about twenty-five hundred adults from the Finnish Twin Cohort and seventeen hundred from the Young Finns Study, with blood metabolites, plasma proteins and genome-wide methylation. For each person they computed a residual: actual body mass index minus the value predicted from a polygenic score, age and sex.

Host 1What did that residual track?

Host 2Above genetic prediction, a higher residual went with steeper rises in inflammatory and cardiovascular markers, such as C-reactive protein and blood pressure. Below prediction, a different set of metabolites and proteins showed the stronger associations, and epigenetic ageing rose with the residual in either direction.

Host 1Did the twins agree?

Host 2Twin pairs who differed in weight showed similar patterns, which controls for shared genes and upbringing. The caveats: it is a preprint, the polygenic score explained only five to ten percent of body mass index variance, and the clinical measures covered only about four hundred people.

Host 1Why does it matter?

Host 2It is a clean way to ask whether weight gained through environment carries a different biology from the weight you were built for.

Host 1The full story is in this paper's deep dive, out this week in the feed.

Grip, gait and diabetes

Host 1From genes to function. Can changes in strength and walking speed flag diabetes risk early?

Host 2This study used four national ageing cohorts, from the United States, England, continental Europe and China, all older adults with no difficulty in daily activities at baseline. It modelled each person's joint trajectory of grip strength and walking speed, then followed them for a median of six to twelve years.

Host 1The result?

Host 2There were about twelve hundred new diabetes cases. Compared with people who held high function, those on declining trajectories had between two point six and four point two times the hazard of new diabetes, depending on the cohort, after weighting for confounders.

Host 1Replication across four cohorts is the strength. The weakness?

Host 2It is observational, so declining strength could be an early sign of metabolic disease rather than a cause. Diabetes was physician-diagnosed or self-reported without glucose tests, which misses undiagnosed cases, and people with complete measurements may differ from those who dropped out.

Host 1The takeaway for someone who trains?

Host 2Grip and walking speed are cheap markers worth tracking, and this fits the case for keeping both. It does not show that training them prevents diabetes.

Host 1The full story is in this paper's deep dive, out this week in the feed.

Sponsor break

Host 1Do you have problems with unwanted hair? Then you are in luck.

Host 1An epilator removes hair at the root, so your legs stay smooth for weeks instead of days. Modern models are cordless, waterproof, and very reasonably priced.

Host 2I use mine every Sunday evening. A few minutes, and you come out of it smooth all over.

Host 1Many come with a cooling glove and interchangeable heads for different areas. Epilators from several brands are in stock right now.

Host 2Mine has two speed settings. The faster one really helps you get it all off.

Host 1Our referral link is in the show notes.

Semaglutide, alcohol and smoking

Host 1Now to addiction. Do the new weight-loss drugs reduce drinking and smoking in randomised trials?

Host 2A Bristol team pooled fourteen placebo-controlled trials of the glucagon-like peptide one drug class, eleven hundred and forty-nine participants, including trials run for diabetes or obesity that also recorded smoking or drinking. Follow-up ran from six to sixty weeks.

Host 1Alcohol first.

Host 2In the semaglutide trials, alcohol intake in grams per day fell by about half a standard deviation and craving by about zero point six, each from only two or three trials. Across all drugs in the class, days without alcohol rose by about a third of a standard deviation.

Host 1And smoking?

Host 2All four pooled smoking estimates leaned toward benefit, but every confidence interval included no effect; the odds ratio for quitting was one point four, anywhere from zero point eight to two point five. Semaglutide trials showed a small drop in cigarettes per day, a tenth of a standard deviation.

Host 1Why does it matter?

Host 2Existing alcohol medicines have modest effects and poor uptake, so a consistent signal from a drug many people already take deserves proper trials.

Host 1How firm is it?

Host 2Not very. It is a preprint, only two of forty-one results were at low risk of bias, certainty was moderate for one outcome and low or very low for the rest, and industry funding or author conflicts were common in the underlying trials.

Host 1The full story is in this paper's deep dive, out this week in the feed.

Two heavy black-hole mergers

Host 1To physics. Two of the heaviest black-hole mergers ever detected, one from twenty nineteen and one from twenty twenty-three.

Host 2A Rochester Institute of Technology group fitted both signals against more than two thousand full numerical-relativity simulations rather than approximate waveform models. The question was whether the odd signal shapes come from tilted, wobbling spins, called precession, or from an elongated, eccentric orbit.

Host 1Those two effects can look alike in a short signal.

Host 2That is the problem. The twenty nineteen event fit best as both eccentric and precessing, with an eccentricity of about zero point two at twenty hertz; the twenty twenty-three event showed similar precession but eccentricity close to zero.

Host 1Why does eccentricity matter?

Host 2Isolated binaries circularise long before merging, so measurable eccentricity points to a dynamical capture in a dense star cluster. The best-matching simulations also imply the final black holes were kicked at roughly four thousand and two thousand kilometres per second.

Host 1Caveats?

Host 2The eccentricity range for the twenty nineteen event is wide, from about zero point zero six to zero point four, and the simulation grid is finite. Earlier work found the twenty twenty-three result depends strongly on the waveform model, and this paper is a preprint.

Host 1The full story is in this paper's deep dive, out this week in the feed.

Distilled models for electrolytes

Host 1Finally, batteries. Simulating how ions move through a liquid electrolyte with near-quantum accuracy is slow.

Host 2A Berkeley-led team took a large pretrained machine-learning model of atomic forces and trained small, system-specific student models on its predictions, with no new quantum chemistry calculations. The students run about forty times faster, so a twenty nanosecond simulation takes days instead of months.

Host 1Do the students hold up against experiment?

Host 2Across sixteen sodium electrolytes and a set of lithium ones, computed conductivity tracked experiment over more than an order of magnitude and beat a widely used classical force field. The students inherit the teacher's errors, though: density runs about ten percent high, viscosity is over‑predicted by roughly three‑ to four‑fold in the most viscous solvent, and one conductivity was off by a factor of five.

Host 1Why does it matter?

Host 2It makes transport screening feasible on practical timescales, and the method is elegant because all the accuracy comes from the teacher. It is a preprint, and the students are only as good as that teacher.

Host 1The full story is in this paper's deep dive, out this week in the feed.

Sign-off

Host 1That is the week. Thank you for listening to the Hastrup Review of Science.

Host 2All six deep dives are out this week in the feed.

Deep dives from this episode

Get the next one

A roundup of new papers every week and a deep dive into one paper most nights. Free, no account needed: paste this address into any podcast app that can add a show by its address, or find “Hastrup Review of Science” by name in an app that searches Podcast Index, like Fountain or Podverse.

https://reviewofscience.org/podcast.xml
New here? Start hereAll episodes