The Average Was Fine. The Pattern Wasn’t.

What my memory, my scale, and my written log each said about a month of protein — and which one was telling the truth

For twenty-five years, I have had some version of the same conversation in my office.

“How is the eating going?”

“Perfect. I have not cheated once.”

Then I look at the scale, the meter, or the blood work, and the numbers tell a less flattering story. Usually the problem is not dishonesty. Usually the problem is memory.

Memory is not a court stenographer. It is a novelist. It keeps the dramatic scenes and trims the boring ones. It puts the heroic moments under bright light and shoves the ordinary mistakes into the basement.

That is not just my clinical impression. In 1992, researchers brought in people who said they could not lose weight despite strict dieting. When the researchers measured actual intake and actual energy use, instead of relying on recall, they found something brutal. On average, the participants reported eating far less than they really ate and doing far more exercise than they really did.[1] Their metabolism was not broken. Their memory was generous.

This month I learned that memory can lie in the opposite direction too.

I became convinced that I had been starving my muscle. My bathroom scale appeared to agree. So did the AI assistant helping organize my notes. Then I opened the one witness that had kept a same-day record, and my handwritten log told a different story.

This piece is about that collision: what I thought happened, what the written record said actually happened, and the harder question underneath both. Did I damage muscle at all?

To answer that honestly, we have to talk about fasting, protein timing, muscle rebuilding, insulin, ketones, and the limits of bathroom body-composition scales. I am going to separate what is established from what is merely likely, what is speculative, and what simply has not been studied in someone living where I am living metabolically.

That last part matters. Most of the published protein research was not done in a 60-year-old physician deliberately trying to keep his blood beta-hydroxybutyrate, or BHB, above 4.0 for months. The science gives us important pieces. It does not give us every answer.

Glossary first, so nobody has to guess

Protein floor. The minimum amount of protein I set for myself on an ordinary day. Think of it as the minimum lumber delivery for a building that is always under repair.

Ideal weight. A reference weight based on a healthy body-mass index, or BMI, which compares weight to height. I do not automatically calculate protein from every pound a person currently carries.

Muscle protein synthesis. The process of building new muscle protein. Muscle is not a statue. It is more like a house under constant renovation.

Amino acids. The smaller building blocks that make up protein.

Gluconeogenesis. The liver making new glucose from raw materials such as glycerol from fat, recycled lactate, and some amino acids. Think of it as an emergency generator in the basement that turns on when fuel is not arriving through the front door.

Insulin. A hormone that helps move and store fuel. In this story it matters because insulin is one of the strongest brakes on ketone production.

BHB. Beta-hydroxybutyrate. This is the main ketone I measure in a drop of blood. In this experiment, my personal floor is 4.0.

Body-composition scale. A bathroom scale that sends a tiny electrical current through the body and uses an equation to estimate fat and lean mass. The scientific name is bioelectrical impedance analysis, or BIA. It does not look directly at muscle.

Overreaching. Training harder than the body has recovered from. It is not an injury, but it is often the step before one.

Time-restricted eating. Eating inside a set daily window. An early window closes earlier in the day rather than drifting late into the night.

Telogen effluvium. A temporary wave of hair shedding that can show up months after a major stress on the body, including crash dieting, illness, surgery, or inadequate nutrition.

Now we can talk like normal people.

Start with what I actually saw

In mid-September, I was frustrated. My ketone readings had stayed below my personal floor for most of two weeks, and I did what patients sometimes do when a number refuses to cooperate. I tried to force it. I skipped meals. I trained hard. I hiked, ran, and did pull-ups while sleeping too little.

On September 19, Day 95 of this experiment, I hiked six and a half hilly miles after eating almost nothing that day. I am sixty years old, and my knees have put me on a cane twice in the last two years. So that hike felt like a win. It was the best test my knees had passed all year. Looking back, it was also the high point of a week my body had not actually agreed to.

Then four clues showed up almost at once.

Clue one: the scale

I compared two body-composition readings taken thirteen days apart. The scale’s equations estimated that my fat had fallen, and that my muscle had fallen too. It was an estimate, not a look at the muscle itself, and I will come back to why that matters. But in the moment, it scared me.

Clue two: the glucose

On the night of the hike, my blood glucose was 76. By the following afternoon it was 98, even though I had not eaten in between. When food is not coming in, the liver makes its own glucose through gluconeogenesis, and some of the raw material can come from amino acids.[2] So the rise told me the liver had been busy. It did not tell me where the raw material came from.

Clue three: the watch

A few days earlier, my watch had labeled my training status as overreaching. In plain English, my training load had outrun my recovery. That matters, because a body that is underfed, underslept, and overworked is not in a calm rebuilding state.

Clue four: my wife

She sees me every day. She said I looked gaunt. That is not a laboratory measurement, but it is still data. Sometimes the person living with you notices the truth before your spreadsheet does.

Put those four clues together and the story felt obvious. On September 20 I stood in front of the camera and told you I had been starving my muscle of protein for two weeks. I believed it.

My AI assistant believed it too. For three days it repeated the same conclusion. On days when I had not mentioned eating in my summary, it treated silence as nothing. So now the witnesses all seemed lined up: my memory, my scale, and the machine summarizing my notes.

Then I checked the witness with the best records.

What the notebook said

Since June, I have written down every blood reading and every time I ate. I went back to that log and reconstructed the two weeks properly.

My average protein intake was above my floor. Not below it. Above it.

That sentence overturned the emotional headline in my head. The nearly empty days were real, but they were not the whole story. Around them were days when I ate close to twice my floor, often late and often in one large sitting.

Memory had done what memory does. It saved the uncomfortable days because they were dramatic. It dropped the quieter days that made up for them, because those were less emotionally bright.

My patients sometimes remember themselves as better than the log. This time I remembered myself as worse than the log. Same error, different direction.

That is worth saying plainly. Memory is not biased only toward flattering stories. It is biased toward vivid stories. A hungry day is vivid. A catch-up meal on a later night is not.

Three patterns that appeared once the record was on the table

Once the written record was laid out, three patterns became visible. I am naming them because names help us hold onto the shape of things. These are self-observations from one person over one month. They are not trial results.

Pattern one: the memory gap

The story in my head and the story in my notebook disagreed. The notebook had dates, times, and entries written the same day. My memory had a mood. The mood was real. The conclusion built from it was wrong.

That may be the most useful lesson in this whole piece. Not the ketones. Not the scale. The simple fact that a written record can save you from a persuasive but false storyline.

Pattern two: feast and famine

The average was acceptable. The rhythm was not. Two days were close to empty. Several other days were close to twice my floor. I spent less time in the steady middle than I would have guessed.

This is the part averages hide. An average is a steamroller. It flattens hills and holes into one smooth-looking road. But the body does not live on the average road. It lives through the actual hills and holes, in order.

Average a feast with a famine and you can create a perfectly respectable-looking day on paper. The body never experienced that paper day. It experienced the famine first, then the feast. Those are not the same thing.

Pattern three: the late plate

Most of my largest meals came late at night. My official daily ketone number is the first reading of the following day, usually in the early afternoon. When I stopped eating by roughly eight in the evening, those next-day readings averaged 2.9. When I ate at nine or later, they averaged 2.4.

That difference got my attention. It is also weak evidence. The sample was small, the groups were uneven, and the later meals clustered during a stretch when sleep and training were also worse.

So I am not calling this a finding. I am calling it a signal worth testing. A finding is something you can start teaching. A signal is something you are still interrogating.

The readings behind pattern three (blog only)

Here is every day from September 4 to 22 that fits one group or the other. The number is the first blood ketone reading of the day, which is my official number. The time is when I last ate the night before. I left out the two nearly empty days and the day-long fast, because they belong to neither group.

Look at the early group and you can see the problem with my own comparison. Three of its five days come from the first ten days of September, when everything was going better. The other two, September 18 and 19, read 2.1 and 1.7. They came during the worst stretch of sleep and training all month, even with an early last meal. So the timing of the meal and the state of everything else are tangled together, and fifteen days cannot untangle them.

One more data point, added after the letter was written. On September 22 I planned to start eating early. I ate on plan in the afternoon. Then at ten at night someone offered me dinner, and I said yes. The next day’s first reading, on September 23, was 1.5. That makes eleven late days, averaging 2.3. It fits the pattern. It still does not prove it.

Now we get to the mechanism

This is the part where pieces like this often become unreadable, so let me slow it down. We are really asking three questions. How is muscle rebuilt? How does fasting change fuel use? And can a feast later in the day fully erase a famine earlier in the day?

Link one: muscle is rebuilt every day

This part is established. Muscle is always being broken down and rebuilt. One side is called muscle protein breakdown. The other is muscle protein synthesis. If synthesis keeps up, muscle is maintained. If breakdown outruns synthesis for long enough, lean tissue can decline.

That is why a reassuring weekly average may not tell the whole story. The renovation crew tears out old material every day. If the lumber stops arriving for a day while the tearing out continues, a good weekly total may not undo that day.

Link two: fasting changes where fuel comes from

This part is also established, but it needs careful wording. When food stops arriving, the body shifts its fuel strategy. The liver makes more glucose through gluconeogenesis. The raw materials include glycerol from fat, lactate recycled from working tissues, and some amino acids from protein turnover.[2]

Early in fasting, protein contributes more to the system. As ketones rise, the brain begins using more of them. That reduces the demand for glucose, and that helps conserve protein over time.[2]

But that adaptation is not instant. A near-empty day on top of hard training and short sleep is not the same as a calm fast in a well-rested person lying on a couch. So September 19 created poor conditions for recovery. It does not prove measurable muscle loss.

Link three: can one feast erase one famine?

This is where the argument gets interesting. On September 20, on camera, I gave you a confident answer. I said muscle can only use a limited amount of protein at one time, so one huge meal cannot make up for hours spent near empty. That is clean. It is also too clean.

Some short human studies do support a more even pattern. In trained young men, four evenly spaced servings after exercise produced a stronger muscle-building signal across twelve hours than either two larger servings or eight smaller ones.[3] Another small study found that a more even distribution produced about 25 percent more muscle protein synthesis across twenty-four hours than a dinner-heavy pattern.[4]

If we stopped there, the verdict would sound settled. But we cannot stop there.

A 2023 tracer study challenged the idea that muscle has a strict ceiling for using protein from one meal. The larger protein serving produced a larger muscle-building response (what scientists call the anabolic response), and it lasted more than twelve hours.[5] That does not mean timing never matters. It does mean the old statement, “anything above a small amount is wasted,” is too crude to survive modern data.

Longer trials complicate the story further. When researchers looked not just at short-term signaling but at actual lean mass over weeks, even and dinner-heavy patterns often produced similar outcomes when total intake and training were similar.[6][7]

So the narrower statement that survives is this. More even protein intake may create more frequent rebuilding opportunities, but a large serving is not automatically wasted. That is a much less dramatic sentence. It is also a more honest one.

Link four: late meals and the ketone brake

The brake is established. The timing effect in me is not.

Protein can stimulate insulin even when it barely moves blood glucose in healthy people.[8] And insulin is one of the strongest hormonal brakes on ketone production.[2] So a large late meal could plausibly keep that brake pressed for part of the night, with some effect still visible in the first ketone reading the next day.

That mechanism makes sense. My small comparison does not prove it happened. Same label again: plausible, not established.

Coffee break

If you got lost in the machinery, come back here.

I thought I had starved my muscle for two straight weeks. My memory said so. My scale seemed to agree. The AI summarizing my notes repeated the same story. Then my written log showed that my average protein intake was actually above my floor.

But the average hid a rough pattern: two nearly empty days, several large late meals, hard training, and too little sleep. Those are not the same problem. The nearly empty days may have created poor conditions for recovery. The late meals may have made the next ketone reading harder to raise.

And the scale, which scared me most, is the witness I trust least. That is the short version. Now for the pushback.

Where the research pushes back hardest

I promised to show you the evidence that argues against my first interpretation. Some of it lands hard.

Pushback one: the scale cannot prove muscle loss

A body-composition scale does not see muscle directly. It measures electrical flow through the body and runs that measurement through an equation. Water-rich tissues conduct differently from fat, so body water is part of the estimate.[9]

The European Society for Clinical Nutrition and Metabolism has said that BIA works reasonably well in healthy people, but that it should not be relied on routinely when hydration is abnormal.[9] Mine was anything but steady. I was fasting, hiking for hours, and changing how much I drank from day to day.

So the scale’s reading could reflect water, glycogen (the sugar stored in muscle and liver), actual tissue change, measurement conditions, or some mix of all four. It is not proof of muscle loss. It is a noisy estimate that handed me a question.

Pushback two: there may not be a hard per-meal ceiling

This one directly weakens what I said on camera. The 2023 tracer study did not support the idea that muscle simply stops being able to use a larger protein serving after exercise.[5] The larger serving produced a larger and longer response.

That does not erase the possible value of spreading protein out. It does kill the simplistic line that one big protein meal is mostly wasted. I am retiring that line.

Pushback three: distribution may matter less over weeks than over hours

Short studies often measure signaling. Longer studies can measure what happened to the person. That distinction is not trivial. A short study may tell you what the workers on the construction site were doing from 8 a.m. to 8 p.m. A longer trial tells you whether the building was actually finished.

In a sixteen-week trial combining weight loss with resistance training, even and dinner-heavy protein distribution produced similar lean-mass outcomes.[6] A smaller eight-week trial in older adults also found no meaningful difference in lean mass, strength, function, or the anabolic response.[7] When the question is actual muscle retention, I give more weight to the longer outcome.

Pushback four: late protein is not automatically bad for muscle

Late eating may be inconvenient for next-day ketones. That does not make it bad for muscle. In one study, young men who exercised in the evening rebuilt more muscle protein overnight when they had protein before sleep than when they had a non-protein control.[10]

Those men were young, they were studied under controlled conditions, and they were not trying to keep BHB above 4.0. Still, I cannot honestly call late protein harmful to muscle merely because it was late. The more plausible cost in me would be the next ketone reading.

Pushback five: extra protein does not simply become sugar

I have said on camera that extra protein gets burned or turned into glucose. That sentence is too blunt. A tracer study found that less than one-tenth of the new glucose made over eight hours came from the labeled protein in the meal.[11]

That does not answer my exact question, because those subjects were not living in deep therapeutic ketosis. It does answer enough to retire the cartoon version of the claim. The liver makes glucose according to need. It does not automatically convert every extra amino acid into sugar.

Pushback six: my glucose rise has more than one possible explanation

My glucose rose from 76 after the hike to 98 the next afternoon, with no food in between. That fits increased glucose production by the liver. It does not identify the raw material. The liver can use glycerol, lactate, and amino acids.[2] Morning hormones can also raise glucose without food.

And my 76 was taken right after a long hike, before I had any water, with a question mark beside it in my own log. So the rise is compatible with protein breakdown. It does not prove that muscle supplied the glucose.

Pushback seven: the late-plate pattern is still fragile

My comparison used only fifteen days: five earlier nights and ten later ones. The later meals clustered during a more chaotic stretch, when sleep was worse and training load was higher. And my best ketone reading of the month followed a day-long fast, which tells me something about fasting, not about dinner timing.

So the late-plate pattern is interesting. It does not yet deserve the word finding.

How sure am I?

Let me label this the way I wish more health writing would.

Established

Recall is unreliable for reconstructing what you ate and how much you moved.[1]

Muscle protein is continually broken down and rebuilt.

Fasting increases the liver’s production of glucose, using glycerol, lactate, and amino acids as raw materials. As fasting continues and ketones rise, the body spares more protein.[2]

Protein can stimulate insulin,[8] and insulin suppresses ketone production.[2]

Body-composition scales are sensitive to hydration and should not be treated as direct measurements of short-term muscle change.[9]

Likely

My two nearly empty days, layered on top of hard training and short sleep, created poor conditions for muscle recovery. They may have produced a window in which muscle breakdown exceeded rebuilding. Neither my scale nor my glucose reading can tell us whether that became measurable muscle loss.

Contested

Spreading protein more evenly may improve the muscle-building signal over a single day.[3][4] Longer trials suggest distribution may not materially change lean mass when total intake and training are similar.[6][7] And the idea of a strict upper limit per meal has been seriously challenged.[5]

Speculative

My late meals lowered my first ketone reading the next day. The mechanism makes sense, and the pattern appears in my log. But the comparison is too small and too confounded to establish cause.

Not studied directly

Most of this protein research was not performed in someone deliberately living with BHB above 4.0 for months. How protein amount, timing, training load, muscle retention, and next-day BHB interact in deep therapeutic ketosis remains largely untested. My data show what happened in me. They do not yet tell us what will happen in everyone.

Why this matters beyond my own month

There are two lessons here. One applies to almost everyone. The other matters most when the experiment is demanding.

The log beats the story

My patients sometimes believe they ate better than they did. I believed I ate worse than I did. The correction is the same in both directions: record what happened while the day is still fresh.

In a four-center study of 1,685 adults, the number of food records people kept each week was one of the strongest predictors of weight loss during the program.[12] That does not prove the act of writing caused the outcome. But it does tell you something important.

A log does not care whether the day felt virtuous, hungry, chaotic, or saintly. It records. That alone makes it more honest than memory. A notebook may be the cheapest diagnostic instrument in medicine.

A floor only works if it is a floor

Most protein recommendations are tied to body weight. That creates a problem when a large share of body weight is fat, because fat tissue and muscle do not create the same protein needs. In my own practice, I set a protein floor using the body size the patient is moving toward, rather than blindly multiplying by current weight.

Critical-care guidelines use ideal rather than actual body weight when calculating protein for patients with obesity.[13] Those are intensive-care patients, not people pursuing ketosis at home. So I am not importing their targets. I am borrowing the narrower principle: body composition matters, and every pound is not metabolically identical.

Supervision matters too. With selected patients, I may deliberately go below a usual floor while watching what happens in the blood and how they feel. That is not the same as drifting into near-empty days because the meter made me frustrated.

One complication I would rather avoid is telogen effluvium, the delayed hair shedding that can follow a major physical or nutritional stress.[14] That lag matters. By the time the hair tells the story, the day that caused it may be long forgotten.

What I changed

On Tuesday, September 22, Day 98, I changed three things.

First, I stopped treating the weekly average as permission for an empty day. My protein floor has to function as an actual floor, not as a number I can rescue later with a feast.

Second, I moved my eating earlier, to give more time between my last meal and the first ketone reading of the next day. This is a self-experiment, not a universal prescription. One controlled study supports the broader question: in men with prediabetes, eating all their food early in the day improved insulin sensitivity and blood pressure even when body weight did not change.[15] That study did not test deep ketosis or my ketone target. It gives me a reasonable question, not a final answer. (As the table above shows, day one of that change lasted until ten at night. Day one starts again.)

Third, I write the record on the same day. I do not let the version of me who remembers at the end of the week overrule the version who was there.

Early in October, I will repeat the scale reading under the same conditions and report whatever it estimates. In the weeks after that, I will report whether eating earlier changes my first ketone reading. The result may support my theory. It may embarrass it. Either result is useful if it is recorded honestly.

The takeaway

If you remember one line from this piece, remember this one: the log beats memory in both directions.

If you are certain you have been perfect, write it down while it happens. If you are certain you have been terrible, do the same. Either way, the record may surprise you.

And do not let the average end the investigation. Two hungry days and several feasts can add up to a perfect-looking week. Your muscle and your meter never lived through that imaginary average day. They lived through each real day, in order.

I am a physician with a meter, a cuff, a watch, and a scale. I still needed a notebook, and my wife, to show me that the story in my head was not the whole story.

— Barry

I post my blood ketone number every day — including the bad ones — in a 90-second video: youtube.com/@neuraliftmd. Day one is there. So is every failure since.

If you want to do something like this with a physician watching, I’m taking on a small founding cohort. It starts with a conversation: calendly.com/hello-neuralift/30min

Educational only, not medical advice. Talk to your doctor before changing your diet, medications, or exercise.

Where all this comes from

1. Lichtman SW, Pisarska K, Berman ER, et al. Discrepancy between self-reported and actual caloric intake and exercise in obese subjects. New England Journal of Medicine 1992;327:1893–1898. doi:10.1056/NEJM199212313272701. — Ten “diet-resistant” adults studied for 14 days with measured intake and energy expenditure. They under-reported what they ate by an average of 47% and over-reported their activity by 51%. Their metabolism was normal. Small, but the classic demonstration that recall misrepresents intake.
2. Cahill GF Jr. Fuel metabolism in starvation. Annual Review of Nutrition 2006;26:1–22. doi:10.1146/annurev.nutr.26.061505.111258. — The standard account of how the body fuels itself without food: early fasting draws on amino acids and glycerol to make glucose; rising ketones spare protein over days; insulin is the main regulator of ketone production.
3. Areta JL, Burke LM, Ross ML, et al. Timing and distribution of protein ingestion during prolonged recovery from resistance exercise alters myofibrillar protein synthesis. Journal of Physiology 2013;591:2319–2331. doi:10.1113/jphysiol.2012.244897. — 24 trained young men, 12 hours after one resistance workout, the same total protein split three ways. Four even servings beat two large or eight small. Measures the rebuilding signal, not muscle size.
4. Mamerow MM, Mettler JA, English KL, et al. Dietary protein distribution positively influences 24-h muscle protein synthesis in healthy adults. Journal of Nutrition 2014;144:876–880. doi:10.3945/jn.113.185280. — 8 healthy adults, 7-day crossover. The even pattern gave 25% higher 24-hour muscle protein synthesis than the dinner-heavy pattern. Short and small.
5. Trommelen J, van Lieshout GAA, Nyakayiru J, et al. The anabolic response to protein ingestion during recovery from exercise has no upper limit in magnitude and duration in vivo in humans. Cell Reports Medicine 2023;4:101324. doi:10.1016/j.xcrm.2023.101324. — Healthy young men after exercise, four isotope tracers at once. A serving four times larger produced a bigger response that lasted more than 12 hours. The strongest challenge to the per-meal ceiling.
6. Hudson JL, Kim JE, Paddon-Jones D, Campbell WW. Within-day protein distribution does not influence body composition responses during weight loss in resistance-training adults who are overweight. American Journal of Clinical Nutrition 2017;106:1190–1196. doi:10.3945/ajcn.117.158246. — 41 adults with obesity, 16 weeks of calorie restriction plus resistance training three days a week. Even versus dinner-heavy protein made no difference to lean mass.
7. Kim I-Y, Schutzler S, Schrader AM, et al. Protein intake distribution pattern does not affect anabolic response, lean body mass, muscle strength or function over 8 weeks in older adults: a randomized-controlled trial. Clinical Nutrition 2018;37:488–493. doi:10.1016/j.clnu.2017.02.020. — 14 older adults, 8 weeks, mixed meals, even versus dinner-heavy. No effect on muscle, strength or function. Small.
8. Nuttall FQ, Gannon MC. Plasma glucose and insulin response to macronutrients in nondiabetic and NIDDM subjects. Diabetes Care 1991;14:824–838. doi:10.2337/diacare.14.9.824. — Review. Protein has little effect on blood glucose in people without diabetes, but it stimulates insulin secretion.
9. Kyle UG, Bosaeus I, De Lorenzo AD, et al.; ESPEN Working Group. Bioelectrical impedance analysis — part II: utilization in clinical practice. Clinical Nutrition 2004;23:1430–1453. doi:10.1016/j.clnu.2004.09.012. — The European clinical nutrition society’s guidance. BIA works in healthy people with a validated equation, but routine use is not recommended at extremes of BMI or with abnormal hydration.
10. Res PT, Groen B, Pennings B, et al. Protein ingestion before sleep improves postexercise overnight recovery. Medicine & Science in Sports & Exercise 2012;44:1560–1569. doi:10.1249/MSS.0b013e31824cc363. — 16 young men who trained in the evening. Protein 30 minutes before sleep raised overnight muscle protein synthesis by about 22% and turned overnight protein balance positive.
11. Fromentin C, Tomé D, Nau F, et al. Dietary proteins contribute little to glucose production, even under optimal gluconeogenic conditions in healthy humans. Diabetes 2013;62:1435–1442. doi:10.2337/db12-1208. — 8 healthy adults after an overnight fast, protein labeled with stable isotopes. Less than a tenth of the glucose produced over the next 8 hours came from the meal’s protein. The subjects were not in ketosis.
12. Hollis JF, Gullion CM, Stevens VJ, et al. Weight loss during the intensive intervention phase of the Weight-Loss Maintenance Trial. American Journal of Preventive Medicine 2008;35:118–126. doi:10.1016/j.amepre.2008.04.013. — 1,685 adults over 6 months. More food records kept per week was a significant predictor of weight loss. Observational within a trial, so it shows association, not cause.
13. McClave SA, Taylor BE, Martindale RG, et al. Guidelines for the provision and assessment of nutrition support therapy in the adult critically ill patient: SCCM and A.S.P.E.N. Journal of Parenteral and Enteral Nutrition 2016;40:159–211. doi:10.1177/0148607115621863. — Intensive-care guidance. For patients with obesity, protein is calculated on ideal body weight. Cited for the principle only; the targets are for hospital patients.
14. Guo EL, Katta R. Diet and hair loss: effects of nutrient deficiency and supplement use. Dermatology Practical & Conceptual 2017;7:1–10. doi:10.5826/dpc.0701a01. — Review. Low protein intake and crash dieting are recognized triggers of telogen effluvium, with shedding appearing months after the trigger.
15. Sutton EF, Beyl R, Early KS, Cefalu WT, Ravussin E, Peterson CM. Early time-restricted feeding improves insulin sensitivity, blood pressure, and oxidative stress even without weight loss in men with prediabetes. Cell Metabolism 2018;27:1212–1221. doi:10.1016/j.cmet.2018.04.010. — A supervised crossover feeding trial: a six-hour eating window ending before 3 PM versus a twelve-hour window, 5 weeks each, with weight held stable. Small, but carefully controlled.
16. My own record. Written ketone, glucose and eating log, Days 80–99 (September 4–23, 2026); body-composition scale readings September 5 and September 18; glucose 76 at 11:02 PM on September 19 (marked with a question mark in the log) and 98 at 1:45 PM on September 20; watch “overreaching” flag, week of September 14. Direct self-observation, not a published study.