Self-experimentation

DIY biohacking: what could help, what could hurt

A summary of everything identified so far, written for someone deciding whether to build one of these and use it on themselves. It is deliberately blunt about the unknowns.

Bottom line. The published effect is reversal of cellular senescence, meaning arrested cells start dividing again and their telomeres lengthen. That is biologically remarkable and, of all rejuvenation strategies, it is the category with the highest theoretical cancer concern. There is no human safety data beyond a suspended 10-person pilot and an ongoing 20-person trial that has not reported. Anyone doing this is running an experiment on themselves with an unknown long-term risk, and should say so plainly rather than call it a wellness routine.

Why this is an attractive DIY target anyway

Technically within reach

The stimulus is weak (roughly 0.5–4 mW/cm², about a thousand times below therapeutic physiotherapy ultrasound), and the components are the same family as ultrasonic cleaners.
No drug, no injection, no import of a controlled substance. Nothing to swallow, and the exposure stops when you switch it off.
The intervention is dose-controllable and measurable: unlike a supplement, you can verify exactly what you delivered, with a hydrophone.
Parameters are published well enough to aim at, and the gaps are listed in open questions.

Scientifically useful

Nobody has independently replicated this. A calibrated open device plus shared data is a real contribution, whatever the result.
The commercial path (patents, clinics, "longevity spa" pricing) will likely make access expensive. An open design keeps the knowledge public.
Cheap, well-characterised hardware also enables the safety experiments nobody is publishing: does LFU also wake up damaged or oncogene-arrested cells?
It fits the biohack.it model: structured protocol plus biomarkers, published openly.

Potential benefits reported so far

All of these come from cells and mice from one group, plus anecdotal human examples inside a patent application. None is established in humans.

Claim	Evidence level	What it would mean for a self-experimenter
Senescent cells resume proliferation; senescence markers normalise	in vitro, one lab	The core mechanism, but not directly measurable in yourself
SASP (inflammatory secretions) suppressed	in vitro	Potentially measurable: IL-6, TNF-α, hs-CRP, GDF-15 and similar in blood
Physical performance in old mice 7–10× sham	mice, small groups	Measurable proxies: grip strength, SPPB, TUG, 6-minute walk, VO₂
Senescence-marker burden in kidney and pancreas reduced	mice	Not measurable at home; blood proxies only
Insulin resistance in aged diabetic mice	tested and failed	The paper that tested this found no improvement in glucose control, body composition or physical function — only a muscle transcriptome shift toward less inflammation
Lifespan extension	not significant per group	Not testable by an individual
Osteoarthritis pain ↓ ~70%, diabetes symptoms ↓ ~75% (patent examples, single cases)	anecdote in a patent	Treat as marketing-grade evidence, not data
Warm-water immersion itself (the sham arm) improves sleep, mood, blood pressure, recovery	separately established	A real confounder: you may feel better from the bath, not the ultrasound

Potential risks

1 · Oncogenic risk the serious one

Senescence is, in part, a tumour-suppression mechanism: a cell with DNA damage or an activated oncogene is locked out of the cell cycle by p53–p21 and p16–RB. An intervention that releases that arrest, without repairing the underlying genome, removes a safety barrier. The published in-vitro phenotype includes cell-cycle re-entry, reduced γH2AX and nuclear p53, and longer telomeres — exactly the profile that raises the concern.

Arguments that it may be lower risk

The stimulus is weak, transient and mechanical, not genetic reprogramming.
The proposed pathway (Ca²⁺ → autophagy up, mTORC1 down) overlaps with well-known geroprotective mechanisms.
No tumours were found at autopsy in the treated mice.
Persistent senescent cells are themselves pro-tumourigenic via the SASP, so leaving them alone is not risk-free either.

Arguments that the risk is real and untested

The mouse groups were tiny (6–8) and the animals were already very old. Tumour latency in humans is years to decades.
Nobody has tested LFU on oncogene-induced or DNA-damage-induced senescent cells, the dangerous kinds.
Whether cells that resume dividing still carry their DNA damage is unknown, and whether their checkpoints still work is unknown.
The same group reports ultrasound killing tumour cells at the same frequency and only 12–18 dB higher pressure, and a separate 2025 study reports that mechanical PIEZO1 activation can prime cancer cells toward a more aggressive state. See the cancer branch.
Telomere lengthening is a two-edged sword: it is also a step most cancers need.

Our position: the in-vitro oncogenic-safety experiment is part of this project, not an afterthought, and nobody should treat whole-body immersion as routine before it exists. See the plan in biological feedback → safety monitoring and question E1 in open questions.

2 · Electrical risk immediate and lethal

You are putting a high-voltage-driven transducer into water that a human body is immersed in. Langevin transducers are typically driven at 100–300 V. A potting failure, a cracked ceramic, a chafed cable or a condensation path can put that into the bath.
A residual-current device will not save you here. In water the let-go threshold is around 5 mA, so a 30 mA RCD tolerates currents that can already incapacitate you; and no RCD type is characterised above 1 kHz, so 33 kHz leakage is outside what any of them is tested to detect. This is the mechanism behind electric-shock drowning, and fresh water is worse than salt water for it.
What actually protects you is galvanic isolation plus construction: a floating supply with two means of patient protection from mains, SELV along the cable with any step-up transformer sealed inside the potted transducer, insulation monitoring instead of residual-current tripping, IPX7/IPX8 potting, and bonding of every reachable conductive part.
Every installation standard that deals with a person in water converges on 12 V a.c. or 30 V d.c. maximum for anything they can touch. A Langevin transducer wants 100–275 V. That tension is the central engineering problem of the whole-body version, and it has an answer, but it has to be designed for rather than assumed.
Also: a second person present, a hard session timer, and a non-self-resetting thermal cut-out independent of the controller.
Bench work with cells or a foot bath is a much lower-risk starting point, and is where we start. Doing it in your own bathtub adds its own risks — bathroom wet zones, being alone in warm water, posture and hearing, other people using the same tub: see the home bathtub page.

3 · Hearing — the risk nobody mentions underrated

The only quantitative exposure limit that exists for a human coupled into a 20–100 kHz field in water is the ACGIH waterborne ceiling of 177 dB re 1 µPa, which is about 1.0 kPa peak. The target dose of 4–8 kPa is 12–18 dB above it.
That limit is about hearing, not tissue damage, and its evidence base is long-term tinnitus after waterborne ultrasound exposure. In a clinical trial of 36 kHz contact ultrasound on skin, tinnitus was the most frequent adverse effect.
In an immersion bath the skull and cochlea are directly liquid-coupled to the source — the exact geometry the limit was written for.
Mitigation: head out of the water, hearing protection, short sessions at first, and stop immediately at any ringing, fullness or hearing change. Get a baseline audiogram before starting, and repeat it.
Note also that Health Canada's guidance on body parts in a kilohertz ultrasonic liquid bath is, in effect, don't. There is no standard that authorises what we are doing.

4 · Acoustic and thermal risk

Inertial cavitation is unlikely at this level: the threshold in ordinary air-saturated water at 33 kHz is around 130 kPa, roughly 13–32× above the target dose. Good news, with a caveat: standing waves can locally double or triple the pressure, which eats into that margin.
Stable cavitation is a different story. For bubbles near acoustic resonance — about 80–115 µm at 33 kHz — the threshold for growth by rectified diffusion in air-saturated water is only a few kPa, i.e. right at our working level. Warm tap water, skin, body hair and thirty minutes is precisely how such bubbles form, grow and migrate to pressure antinodes. Microstreaming around them is both a plausible mechanism of the claimed effect and a plausible hazard. Degassing is what separates the two, and it has to be measured, not assumed.
Forget the mechanical index. It is defined for megahertz diagnostic ultrasound and is known to be unreliable below ~500 kHz. "MI is only 0.02" is not a safety argument at 33 kHz.
Hot spots. With a 4.5 cm wavelength, standing waves can give local pressures several times the nominal value. Without a field map you do not know your dose.
Hearing and subharmonics. 33 kHz is inaudible, but non-linearity and gating produce audible subharmonics; prolonged exposure to intense airborne ultrasound is linked to unpleasant effects. Keep the tank covered and the room quiet.
Heat and circulation. A 32–35 °C bath for 30–45 min is a cardiovascular load in itself (hypotension when standing up, fainting, dehydration).
Implants and gas pockets. Metal implants, joint prostheses, pacemakers, cochlear implants, stents, IUDs, recent surgery, and gas-containing spaces behave unpredictably in an acoustic field. Anyone with these should stay out.

5 · Biological uncertainties beyond cancer

Senescence has useful roles: wound healing, fibrosis resolution, tumour suppression, placental and embryonic processes. Suppressing it everywhere may have costs.
Effects on stem-cell pools, the immune system and fibrotic tissue are unknown in humans.
Pregnancy: ultrasound exposure of a fetus outside diagnostic settings is not acceptable. Absolute exclusion.
Interaction with drugs and conditions (immunosuppression, active infection, autoimmune disease) is unstudied.

6 · Measurement and self-deception

Warm baths feel good. Without a sham phase and objective markers, you will conclude that it works.
Single before/after blood values are noise: inflammatory markers swing with sleep, infection, exercise and time of day. Multiple baselines are mandatory. See study design.
Epigenetic clocks have substantial technical variability; a small "age reversal" may be measurement noise.

Who should not do this

Based on the UT trial criteria plus our own risk reading. This is a starting list to review with a physician, not a medical clearance.

Any history of cancer, current cancer, or known precancerous lesions or monoclonal conditions (e.g. MGUS).
Pregnancy or breastfeeding.
Implanted electronic devices (pacemaker, ICD, neurostimulator, cochlear implant, insulin pump), or metal/joint implants in the immersed region.
Open wounds, active skin infection, rashes, poor lower-limb wound healing; incontinence (hygiene of shared water).
Unstable cardiovascular disease, uncontrolled hypertension, recent myocardial infarction or stroke, orthostatic intolerance, epilepsy.
Immunosuppression, active infection, or colonisation with multidrug-resistant organisms.
Anyone who cannot get out of the bath unaided, or who would be alone during a session.

House rules for this project

Bench before body. Cells and calibration first, then the hand study with the other hand as control, then a limb bath, and whole-body immersion last, only with an electrical safety case.
No unverified dose, no session. A build that has not passed the acceptance test produces data that are not worth having.
Sham-controlled and pre-registered. Declare the protocol and the endpoints before starting; run a sham phase; publish the result whichever way it goes.
Baseline first, and enough of it. At least 2–3 baseline blood draws and functional tests before the first session.
Medical partner. Blood tests interpreted with a clinician; exclusion criteria reviewed; a stop rule written in advance.
Nothing sold, nothing claimed. No kits for sale, no treatments, no therapeutic claims, no charging anyone for a session. See IP rights.
Publish the negatives. Failed builds, bad calibrations and null results go on the site too.