Drug efficacy varies by up to 40% depending on when in the circadian cycle it is administered. Croniixx calculates the right time from your patient's actual biology.

Circadian phase +01:12 against a standard 24 hour cycleWake 01:20 to 01:45Light 01:45 to 02:36Deep 02:36 to 03:33Light 03:33 to 04:06REM 04:06 to 04:45Light 04:45 to 05:30Deep 05:30 to 06:09Wake 06:09 to 06:18Light 06:18 to 07:00REM 07:00 to 07:51Light 07:51 to 08:21REM 08:21 to 09:05Wake 09:05 to 01:2000061218Prednisolone 20 mg 09:05 to 11:05 target 09:35 optimalCapecitabine 1500 mg 03:12 to 07:12 target 05:12 optimalCapecitabine avoid 11:05 to 19:05 target 15:05 contraindicatedSimvastatin 40 mg 23:20 to 01:50 target 00:35 optimalRamipril 5 mg 21:12 to 00:12 target 22:42 acceptableDLMO estimate at 22:12Midsleep at 05:12Wall clock 09:00Circadian position 07:48PHASE OFFSET+01:12DELAYEDBIO 07:48CLK 09:00
SleepDeepREMLightWake
Windowoptimalacceptablesuboptimalcontraindicated

Sample patient · delayed phase · four agent regimen

The outer ring is one night of this patient's own sleep architecture, staged from a wearable. The violet needle is where their internal clock currently sits. The dashed needle is the wall clock. The angle between the two is the phase offset, and it is the reason a printed schedule and this schedule disagree.

The inner arcs are dose windows for the regimen, placed against biological anchors rather than clock times. Green is the calculated window. Red is a period the agent should not be given in. The white tick inside each arc is the target moment.

Phase offset
+01:12 delayed against the population reference
Midsleep
05:12 local, from fourteen nights of wearable data
DLMO estimate
22:12, projected from sleep timing
Largest dose shift
Capecitabine moves 2 h 48 m earlier than the printed 08:00

How the timing is derived

  1. 01Ingest

    Terra streams sleep staging, heart rate variability, and movement from Oura, Apple Watch, Garmin, and Whoop. The differences between those four are reconciled before anything is stored, because Apple reports SDNN where the others report rMSSD and Whoop samples inside slow wave sleep rather than across the night.

  2. 02Profile

    Fourteen days of normalized data produce a circadian profile: sleep midpoint and its variability, cosinor fits for heart rate variability and skin temperature, and the standard nonparametric actigraphy measures. Nothing at this stage claims a phase position; it describes what was observed.

  3. 03Position

    The profile is scored into a signed phase offset with a confidence attached. A positive offset means a delayed clock. This is the step that carries the proprietary coefficients, and a public checkout of this repository runs a transparent reference estimator instead.

  4. 04Schedule

    Each agent in the regimen has timing windows expressed relative to biological anchors rather than clock times. Resolving those anchors against the patient's own phase produces dated windows, which is the adaptive schedule the app and the reminder queue consume.