Body by AI is built on 228 peer-reviewed citations. We're proud of that number. It represents decades of exercise science, nutrition research, and metabolic studies that inform every recommendation our coaching makes.
Here's what I need to be honest about: almost none of those 228 citations studied transgender individuals specifically.
This isn't a criticism of the research community — although more funding would help. It's a statement of fact about where the science is in 2026, and it matters because you deserve to know what's evidence-based, what's extrapolated, and where we're working with best-available reasoning rather than direct evidence.
What Studies Actually Exist
The body of research on transgender exercise physiology is small but growing. Here's what we have:
Body Composition Changes During HRT
This is the most studied area, primarily because it overlaps with endocrinology research:
- Klaver et al. (2018) — Tracked body composition changes in trans men and trans women over 1 year of HRT. Found significant increases in lean mass (trans men) and fat mass (trans women), with redistribution patterns matching expected hormonal effects
- Wiik et al. (2020) — One of the few studies to measure both muscle size and strength in trans individuals after 1 year of HRT. Key finding: while thigh muscle area increased in trans men and decreased in trans women, strength changes didn't perfectly track muscle size changes
- Harper et al. (2021) — Systematic review finding that testosterone suppression in trans women reduced hemoglobin, lean mass, and strength over 2+ years, but not to cisgender female reference ranges
Metabolic Changes
- Klaver et al. (2018) — Documented changes in insulin sensitivity, lipid profiles, and body fat distribution during HRT
- Fighera et al. (2019) — Tracked weight and metabolic parameters in trans women during the first year of feminizing HRT
Sports Performance
Most of the "trans exercise" research has focused on the competitive fairness debate, not on how to train trans individuals effectively:
- Harper (2015) — The earliest study of transgender athletic performance, focusing on race times of trans women runners
- Roberts et al. (2021) — Air Force study on push-ups, sit-ups, and run times in trans women over 2 years of HRT
- Jones et al. (2023) — Systematic review in Sports Medicine calling for more research on transgender exercise physiology, noting the literature is "insufficient to make specific exercise recommendations"
Almost Nothing
- Optimal training programming for individuals during HRT
- Protein timing and dosing during HRT-driven body composition changes
- Periodization models for trans individuals
- Recovery capacity changes during HRT
- Training responses to specific modalities (resistance training, HIIT, endurance) during transition
- Bone mineral density changes and their implications for load-bearing exercise
- Psychological effects of exercise during transition (a few small studies exist, but nothing systematic)
The Limitations of What We Do Have
Even the studies listed above come with significant caveats:
Small sample sizes. Most studies have 20-50 participants. Some have fewer than 15. Statistical power is limited, and results may not generalize.
Short timelines. Most studies track participants for 1-2 years. HRT-driven physiological changes continue beyond that window. We have very little data on 5-year or 10-year outcomes.
Self-selected populations. Study participants tend to be younger, healthier, and more engaged with healthcare than the general trans population. Results may not apply to individuals with comorbidities, older individuals, or those with limited healthcare access.
Cisnormative baselines. Most studies compare trans individuals to cisgender reference ranges. This tells us how trans physiology differs from cisgender norms but doesn't tell us how to optimize it.
Focus on body composition, not training.The question researchers have mostly asked is "how does HRT change the body?" The question trainers need answered is "how should training change in response?" These are related but distinct questions, and the second one has barely been studied.
Confounded by exercise status.Few studies control for or even document participants' exercise habits. Body composition changes in a sedentary individual on HRT are different from those in someone doing structured resistance training — but most studies don't distinguish between the two.
What Body by AI Does With Available Evidence
Given these gaps, our approach is:
1. Start With General Exercise Science, Then Adapt
The core principles of progressive overload, protein requirements for muscle protein synthesis (1.6-2.2 g/kg/day per Morton et al., 2018), and evidence-based programming don't change because someone is trans. Physics is physics. Muscle protein synthesis is muscle protein synthesis.
Where HRT-specific evidence exists — like the timeline of lean mass changes on testosterone therapy, or fat redistribution patterns on estrogen therapy — we incorporate it into the coaching context.
2. Track Individual Data, Not Demographic Averages
This is where our approach has the most impact. Body by AI tracks your body composition, your workout performance, your progression rates, and your recovery patterns. When the population-level evidence is sparse, individual-level data becomes more important, not less.
If your lean mass is increasing at a rate consistent with early testosterone HRT, your coach notices that trend in your data and adjusts programming accordingly — even without a large-scale study telling it exactly what to expect.
3. Use the Katch-McArdle Formula
Most BMR calculators use the Harris-Benedict or Mifflin-St Jeor equations, which require a binary sex input. Katch-McArdle uses lean body mass: BMR = 370 + (21.6 × LBM in kg). It doesn't care about your sex at birth or your gender identity. It cares about your lean mass. For individuals whose hormonal profile doesn't match traditional male/female metabolic assumptions, this is a more accurate foundation.
4. Never Use Gendered Benchmarks
When the evidence base can't tell us what "normal" performance looks like for a trans man at 8 months on testosterone, the answer isn't to use cisgender male norms or cisgender female norms as a proxy. The answer is to stop using norms altogether and compare you to yourself.
How the Research Flywheel Works
Body by AI includes a research contribution system. When users share exercise science articles, our system:
- Extracts the study details — authors, publication, methodology, sample size
- Assesses methodology quality and relevance
- Tags the study by relevant populations (including trans-specific research)
- When multiple users flag the same study, it's escalated for coaching integration
Transgender exercise physiology is one of the topics we actively track. When new research is published — and it is being published at an increasing rate — we want to know about it quickly.
An Invitation to Researchers
If you're studying transgender exercise physiology, we'd love to talk.
Body by AI has a unique dataset: real-world training data from users across a range of ages, HRT statuses, training levels, and goals. With appropriate anonymization and user consent, this data could help close some of the research gaps described in this article.
We're specifically interested in collaborations around:
- Optimal training volume and intensity during HRT transitions
- Protein requirements during HRT-driven body composition changes
- Recovery capacity changes across the HRT timeline
- Psychological effects of structured exercise during transition
- Bone mineral density and load-bearing exercise during HRT
If you're a researcher, a graduate student, or a clinician working in this space — reach out. The fitness industry won't build better tools for trans individuals until the evidence base grows, and the evidence base won't grow without data and collaboration.
Why Honesty About the Gaps Matters
I could have written a blog post that cherry-picks the available studies and presents them as a comprehensive evidence base. A lot of fitness companies do exactly that — cite three studies and imply that their entire product is "clinically proven."
That's not how we operate. The evidence for transgender exercise physiology is emerging. It's growing. It's promising. And it's incomplete.
Body by AI navigates this honestly:
- Where direct evidence exists, we use it
- Where it doesn't, we extrapolate from the best available general exercise science
- Where even extrapolation is uncertain, we rely on individual data and adapt in real time
- And we're transparent about all of this, because you deserve to know the basis for the advice you're getting
The research gap is real. Closing it matters — not just for better fitness apps, but for better health outcomes for millions of people. We're committed to being part of that process.