Endo is different in every body. pum studies yours.

you, compared with you

Most trackers compare you to a model of the average user. pum runs the only comparison that holds up for endo: you against yourself. It weighs your painful days against your own calm ones, and your sleep before flares against your sleep in easy weeks. After a treatment change, it compares the months after against the months before.

Clinical researchers use this same design when a condition varies too much between people for averages to mean anything. Endometriosis is exactly that condition.

pum knows what your cycle explains

A naive tracker sees Tuesday's pain and Monday's bad sleep and calls it a pattern. But if Tuesday was also day 26 of your cycle, the cycle may explain both. pum knows what phase every data point landed in, and asks first what the cycle accounts for. So a "sleep affects your pain" finding means sleep, not the week before your period.

confidence on everything

Every finding shows its evidence: how many cycles it rests on, and how sure pum is. Early signals are labelled early. When there isn't enough data for an honest answer, pum tells you that instead. Real patterns take months to establish, and pum won't dress a hunch up as a finding.

predictions that respect your treatment

pum learns your cycle's own rhythm and updates as new data arrives, instead of assuming 28 days and hoping. It also knows treatment rewrites the rules: an IUD, hormonal therapy, or surgery changes what a cycle even means. pum adjusts, instead of pretending nothing happened.

That honesty has an edge most apps won't touch: if your period is suppressed or gone, pum will not invent a countdown to it. It will not show a due date, or a "period late?" alert, for a bleed that isn't coming. Where your body still cycles hormonally, pum can read position from temperature: hedged, in plain language, with its confidence showing.

what pum is not

• Not a diagnosis. pum finds patterns in what you log. It cannot tell you whether you have endometriosis, and it will never claim to.
• Not medical advice. pum presents your data and stops. What you do with it, including taking it to a clinician, is yours.
• Not a score. You rate your own pain when you log it, on the same scale a pain clinic would hand you. pum never rates you back: no daily grade, no readiness score.

grounded in the research

Each thing pum asks you to log earns its place: sleep, cycle phase, medication, and the endo-specific symptoms generic apps skip. pum's method comes from published research, and is tested in simulation before it ships. A few of the findings behind it:

1. How far endo has spread is a poor guide to how much it hurts. In over 1,000 women, disease stage barely predicted pain severity. So pum reads your pain from your own data, not from an average patient.

   Vercellini P, et al. 2007. Human Reproduction.

2. Endo pain follows the cycle, but not the same way for everyone. In 473 women, about half had pain that moved with the cycle. About three quarters hurt most just before the period. This is why pum knows where in your cycle every symptom landed.

   Schliep KC, et al. 2015. Human Reproduction.

3. Poor sleep and chronic pain travel together. Across 208,190 adults, sleep problems nearly doubled the odds of chronic pain. So pum studies your sleep beside your next-day pain.

   Santos M, et al. 2023. Rheumatology.

4. The 28-day cycle is mostly a myth. Across 4.9 million tracked cycles, length varied widely between women, and within the same woman. This is why pum learns your own rhythm instead of assuming one.

   Li K, et al. 2020. npj Digital Medicine.

5. Wrist temperature is useful, and easily fooled. Ovulation signals from wrist temperature carried a high false-alarm rate. So pum hedges any temperature-based estimate and shows its confidence.

   Zhu TY, et al. 2021. Journal of Medical Internet Research.

6. The fairest test for a condition like endo compares each woman with herself. Clinical research uses this design when a condition varies too much between people. pum's analysis is built on it.

   Maclure M. 1991. American Journal of Epidemiology.