The cycle the app could not see.

A 38-year-old woman had been logging her period in three different apps for four years. Each app showed her a tidy 28-day pattern and marked her symptoms as “within normal range.” She knew the second half of her cycle was not normal — three days a month she could not work, the kind of pain that made her cancel meetings, plus a sleep collapse the apps simply did not have a field for. Two GP visits returned the same line: cycles vary, this is common, take ibuprofen.

She stopped opening the apps. For four cycles she wrote four lines a day in one note: cycle day, sleep, pain (1–10), one sentence about energy. At the end of the four months she pasted the whole thing into a free reasoning chat tool and asked it to find the pattern. The model did not assume a 28-day average. It read her own four cycles and surfaced the actual shape: a luteal-phase pain cluster that started day 18 and tracked perfectly with the sleep collapse two nights earlier — a pattern none of the apps had a category for because the apps were built around bleed days, not the two weeks before.

She brought one paragraph to her next GP visit: dates, the pain–sleep correlation across four cycles, one specific question about luteal-phase support. Within three months she had the first investigation that had ever been ordered for her. The diagnosis (a manageable hormonal pattern, not the worst-case she had been quietly fearing) was useful. What was more useful was that the conversation had finally moved past “cycles vary.”

Across four months she replaced three subscription apps with one note. Her pain days dropped from three a month to one after a six-week protocol her GP and a women's-health physio agreed on. She has since shared the four-line template with her sister and two friends; one of them has since had her own previously-dismissed symptom finally investigated.