M-01 · Release
Introducing MUSCLE-1: A Real-Time Hypertrophy Prediction Model
// 01 — The problem
Nobody knows if their training is doing anything.
Most people have no idea if their resistance training is working.
They walk into the gym, follow a generic program, throw some weights around, then leave — without any quantitative way to know whether the session they just finished actually moved the needle on muscle growth.
This is still the case if you have a trainer or coach — the measurement layer for hypertrophy is broken.
You can be told to aim for 8–12 reps per set, train near failure, and progressively overload volume. However, without quantifying how each set you do actually contributes to muscle growth, you’re training in the dark.
// 02 — Why hardware fails
Hardware isn’t the answer.
If measurement is the problem, most people’s first instinct is to build new hardware. EMG, velocity trackers, force plates, wearable strain gauges — a new startup tries to ship a novel sensor every other year.
None of them work.
EMG amplitude doesn’t cleanly map to hypertrophic stimulus and is contaminated by skin impedance, electrode placement, and movement artifact.
Bar-velocity sensors capture intent on a single set but tell you nothing about how your muscles adapt to training week to week.
Force plates are impractical, and wearable strain metrics are nonsense for muscle growth.
None of these sensors answer the thing you actually need to know: did your training produce a growth-predictive stimulus on the muscle you just trained?
Hypertrophy isn’t directly observable in real time. It’s a downstream consequence of stimulus accumulated over weeks.
There is no sensor that watches a muscle grow. The only honest path forward is to measure the inputs that predict growth as precisely as possible and forecast the output. This is a modeling problem, not a hardware problem.
MUSCLE-1 is the software equivalent of what Whoop did for recovery. It takes an opaque physiological process and turns it into a daily, measurable, actionable per-muscle score.
The best part: no hardware required. Just track your sets like you do in any training app.
// 03 — The biology
What actually drives hypertrophy.
Hypertrophy is driven by two well-validated principles.
- 01
Mechanical tension drives hypertrophy.
Muscle grows when high mechanical loading recruits high-threshold motor units and holds them under tension. That tension activates mTORC1 signaling and the downstream protein-synthesis machinery that drives growth.
- 02
Mechanical tension is delivered by hard sets.
The quality of a set is determined by how close to failure it was taken. Current literature places a “hard set” at roughly 0–3 reps in reserve. Anything too far from that is junk volume — you lifted weights, but the work didn’t stimulate growth.
This collapses the practical problem of measuring hypertrophy into two questions:
- 01
How many of my sets were actually hard?
- 02
Am I doing more hard sets per muscle every week?
Population-level evidence shows a robust dose-response between weekly hard-set volume per muscle and hypertrophy, up to your recovery ceiling (how much load you can tolerate). If you’re doing more hard sets on a muscle week over week, that muscle is likely growing.
Consumer training apps don’t capture this because they can’t measure it. They count sets and reps but have no view into whether a set was hard. The “volume” they report — sets × reps × weight — is in units that aren’t biologically meaningful.
If you do 8 reps at 100 lb one rep from failure, then later do the same 8 reps at 100 lb seven reps from failure because you’ve gotten stronger — the first set drove hypertrophy and the second was effectively a warmup.
// 04 — The model
How MUSCLE-1 fixes this.
MUSCLE-1 has three jobs.
- 01
Computationally score proximity to failure.
Every set you log carries a quantitative quality score, weighted by the muscles actually loaded. Sets too far from failure earn little credit; sets close to failure earn a lot. We compute proximity to failure rather than ask you for it — published literature is clear that people routinely overestimate how close to failure they are. To our knowledge, MUSCLE-1 is the first consumer-grade system to automatically predict reps in reserve on every set.
- 02
Estimate stimulating volume.
“Volume” is a poor measure of hypertrophy, and “hard sets” alone undercount what a muscle actually absorbs across a week. MUSCLE-1 converts both into stimulating volume — total volume across all sets, normalized to how stimulating each individual set actually was.
- 03
Turn stimulating volume into a per-muscle weekly target.
Stimulating volume gets mapped to the primary and secondary muscle groups worked across every exercise, then aggregated into a weekly progress score per muscle. Targets escalate over time to drive overload.
If you do bench press one week, chest flyes another, and dumbbell bench press another, MUSCLE-1 maps chest stimulus across all three and adapts your training targets week over week to keep producing growth.
No app or piece of hardware has been able to predict hypertrophy on each individual muscle, week to week, from observable training data alone.
We’ve gamified training into something as simple as filling a progress bar.
MUSCLE-1 puts your training on autopilot.
// 05 — The product
Inside the product.
MUSCLE-1 predicts three metrics:
- 01
Weekly Stimulating Volume Target
How much stimulating volume you need to produce every week per muscle, across all exercises, to predict a growth stimulus.
- 02
Hypertrophy State
A daily per-muscle estimate of whether a target muscle is actively growing on a given day, based on how much stimulating volume you produced in each session.
- 03
Neural Fatigue
A daily per-muscle estimate of how neurally fatigued a target muscle is on a given day, based on how intense your training is.
The product experience is simple.
Log your training.
Scroll to play
Prana lets you log training sets in two ways:
- Type each set in manually, per exercise.
- Use AI to skip the manual busy-work.
All you have to do is feed your training set data into the system. You can use AI to make this easy, avoiding manual UI flows for tracking in case you hate it.
Each exercise you log is mapped to primary and secondary muscles trained. You can create an unlimited number of custom exercises — we estimate your strength curve on each individual exercise based on your training data week to week.
In practice: if the Triceps Pulldown machine in Gym A feels noticeably lighter than the machine in Gym B, or if a certain variation of Seated Dips hits your Triceps more than your Chest — you can create a custom exercise to log them separately. We re-calculate your strength curve on that specific exercise, then calibrate our estimate of stimulating volume based on it.
See your real-time target.
Scroll to play
While you log your sets, MUSCLE-1 calculates muscle-specific stimulus in the background.
Instead of guessing when you’re done in the gym, we give you a tiered progress bar to hit per muscle every week.
Once that progress bar is full, you’re done for that muscle for the week. We score how much stimulating volume produces a minimum effective stimulus for growth, how much produces a very high stimulus, and how much is likely overtraining.
If you hit your targets, we confidently predict growth on that muscle. We recommend splitting your week’s volume per muscle across two sessions instead of cramming everything into one.
Push harder.
Scroll to play
MUSCLE-1 learns how your body is adapting.
Every week, the model increases your hypertrophy target to make sure you’re overloading volume. These targets are informed by your strength curve over time, giving you a truly adaptive per-muscle intelligence layer to use for training.
In your dashboard, you’ll be able to see how your muscles are overloading stimulating volume over time, trends in training intensity, day-by-day hypertrophy state and neural fatigue, and exercise-specific metrics to track progress.
When you switch between bulks and cuts, start a new cycle, or are getting back into the gym after a few months off — simply start a new training phase in Prana. MUSCLE-1 refreshes all of its insights.
MUSCLE-1 gives you the complete intelligence layer you need to grow muscle, fast.
// 06 — Limitations
Where MUSCLE-1 stops.
MUSCLE-1 is purpose-built for hypertrophy and resistance training. The model works best for growth-focused training.
The model doesn’t work for strength-focused training, yet. If you’re looking to hit new PRs and pull more weight at your next powerlifting meet, MUSCLE-1 is not the right model for you. That’s because strength and hypertrophy are two different adaptive systems.
Additionally, MUSCLE-1 can only predict growth insofar as your body can support it. If you are in an extreme calorie deficit, MUSCLE-1 can only predict how much training you need to do to support lean mass retention.
Similarly, if your hormonal status is compromised and not supporting muscle growth, MUSCLE-1 can only go so far to predict growth.
This is why Prana does not look at training in isolation. We predict muscle growth, fat loss, and metabolic changes using all of your biomarkers, with models calibrated to you.
These models will be covered in a future announcement.
// Try it out
Run MUSCLE-1 on your training.Become superhuman.
Request an invite to the beta — and get an early taste of the body transformation system we’re building at Prana.
We’d love your feedback :)