How it works

Five moving parts, and none of them are magic.

1. Everything lands in one table

However a number arrives, whether typed in, pasted off a lab report, read out of a spreadsheet, pulled out of an Apple Health export or averaged from a continuous glucose monitor, it becomes the same thing: a measure, a date, a value and a source. That is what makes a lab result and a step count comparable on the same timeline in the first place.

Around seventy measures are recognised by name, with their common abbreviations, so "HbA1c", "Hemoglobin A1c" and "A1c" all land in the same place. Anything unrecognised is reported back to you rather than quietly dropped.

2. Charts show the range, not just the line

A number on its own tells you very little. Every chart draws the usual adult reference range as a band behind the line, marks a one-day event as a dashed vertical and an event with an end date as a shaded stretch from its first day to its last, and labels the most recent reading. Hovering gives you the exact value, the date and any note you attached. Every chart also has a table underneath it, because sometimes you want the numbers.

3. Symptoms are the half nobody measures

Lab results say what was in your blood. They do not say that you had a headache on nine mornings in a month, or that the stomach upset lasted a fortnight and then stopped. You log a symptom with the day it started, how bad it was out of ten, and the day it ended, and it stays open until you close it off, so the day count keeps running on anything still going.

Names collect: a headache, a migraine and a head ache end up on the same row, and anything the catalog does not know is kept in your own words. What you get is one strip per symptom showing when it was there and how bad, and a table on the physician summary with the dates, the counts and the severities already worked out. A short list of symptoms, chest pain among them, comes back with a note saying to be seen rather than to keep tracking.

4. Events give the changes a cause to test against

Mark the day you started insulin, or walking, or a new job. Pick that event later and welldata averages every measure with at least two readings on each side of it, inside a window you choose, and shows you what moved and by how much.

This is a comparison of two averages, not a trial. It tells you where to look. A measure that shifted 20 percent after you changed something is worth a conversation; it is not proof that the change caused the shift.

5. The review runs two passes

The first pass reads your data. It compares each latest value against its reference range, compares it against the reading before it, and notices measures that have not been rechecked in two years. Anything far outside its range is escalated.

The second pass walks a list of adult screening intervals drawn from USPSTF, ADA, ACS and CDC guidance. It keeps only the ones that apply to your age, sex, smoking history and stated conditions, then works out when each is next due. Where a result already in your record can stand in for the screening, that date is used automatically, so a cholesterol panel you imported counts as the cholesterol screening.

The two passes together produce one recommendation at the top of the page: nothing asking for attention, due for a check-in, worth booking an appointment, or contact a clinician promptly. Every part of it names the rule and the numbers it used.

What it deliberately does not do

• It does not diagnose. Out of range is not a diagnosis, and in range is not a clean bill of health.
• It does not connect to your clinic or your insurer. Nothing here is verified against a clinical system.
• It does not decide for you. Every screening it lists is a starting point for a conversation, and several of them, prostate testing in particular, are explicitly meant to be decided jointly with a clinician.