The n=1 experiment missing its dependent variable.
You measure glucose, HRV, sleep stages, and body temp. Therma adds the column everything else is trying to predict: how you actually felt. Sixty seconds a day, correlated weekly.
By Omar Rantisi, Founder of Therma2 min read
The problem
The modern stack measures everything except the outcome. Glucose, HRV, sleep architecture, readiness scores: all proxies, all pointed at a variable nobody instruments, which is how the day actually felt. So experiments run open-loop.
The supplement gets credit for a good Tuesday, the cold plunge gets blamed for nothing, and the confound (the argument, the deadline, the dread) never enters the model. Objective data without subjective ground truth is half a dataset.
“Your Oura knows the HRV dropped 14 points. It doesn't know why.”
How Therma helps
Therma closes the loop. One 60-second check-in a day supplies the dependent variable, and the weekly reveal correlates it against sleep and HRV from Apple Health and Oura.
Now the experiments mean something: whether the protocol moved the felt outcome or just the proxy, which changes carry across weeks, and how much of readiness was actually mood wearing a chest strap. Ground truth, at the cost of a minute.
What you get
- 1The 60-second check-in
One calibrated data point a day: the mood ring plus an adaptive question. Sixty seconds, consistent instrument, no drift.
- 2The weekly reveal
Weekly correlation of mood against sleep and HRV. The reveal is the analysis layer your spreadsheet was pretending to be.
- 3Chat
For the weeks the data and the feeling disagree. Chat sits with the part of the experiment that does not fit in a cell.
Common questions
Which integrations does it read?
Apple Health and Oura, which covers Apple Watch and ring data: sleep and HRV land automatically once connected. Your check-ins stay yours, encrypted and never used to train models.
How fast do correlations show up?
The reveal starts being interesting after about two weeks of check-ins, and gets sharper as the baseline builds. Single-variable experiments read best over a few weekly cycles.
Subjective data is noisy. Why bother?
Because it is the outcome. Noise averages out over daily samples, and a consistent 60-second instrument beats retrospective memory by miles. The alternative is optimizing proxies and hoping.
Omar Rantisi
Founder of Therma. UCLA Math + Sociology. Building tools for the space between silence and therapy. Not a therapist. Just someone who needed this to exist.
Therma · Emotional Wellness
A place to put what you’re carrying
Daily check-ins. Guided reflection. A companion that meets you where you are. Therma is built for the moments between therapy sessions, between good days and hard ones.