What happened before the better nights?

The comparison is the feature

This result found night sleep was about two hours longer when left-to-right breastfeeding happened in the two hours before bedtime: 5h 59m across 22 nights, compared with 4h 2m across 13 nights without it.

It does not stop at "Strong pattern." Open the card and Rhythm shows what it compared, the time window, the averages on both sides, and how the result held up after checking every other possible pattern at the same time. This scan checked 108.

This is not another totals chart

The baby trackers I checked chart routines, summarize totals, or predict the next nap. All useful. I could not find another one that checks how one part of your baby's day changes with another and shows the comparison behind it.

• What happened in the two hours before longer or better-rated sleep?
• What tended to follow a feed, activity, or nap?
• Which events came up more often on the days you rated higher?

The questions grow with your tracker

Built-in feeds, sleep, solids, and activities are only the start. Every custom event can join the same analysis. Add Reading, Stroller Walk, Screen Time, Daycare, or Grandma Visit and Rhythm can look for what changed around it once you have logged it often enough.

Fields count too. If you add a Music event with an Artist field, Rhythm can compare nights with one artist against the rest of the music you logged. A food tag can be compared with the other solid foods, instead of pretending all lunch entries were the same.

The boring rows stay visible

Check enough combinations and something will look interesting by accident. Rhythm corrects for that before a result becomes a card. It also keeps the full list of what it checked: strong patterns, weak ones, and the rows that still need more data.

Those rows matter. They tell you what Rhythm tested and chose not to turn into a claim. A new event needs repeat logs across several days before it enters the analysis, and both sides of a comparison need enough examples to be useful.

No AI, and no server reading your baby's history

Rhythm uses ordinary statistical tests over the last 90 days of logs. The scan runs locally on your device in milliseconds. Your history is not uploaded to an AI model, and the result is not generated by a chatbot trying to sound confident.

Dismissals and feedback stay on that device too. The analysis runs over data Rhythm already has locally; turning on household sync does not move the math to a server.

A pattern is a place to look, not a command

Longer sleep after a routine does not prove the routine caused it. Bedtime, illness, teething, or a dozen things that never made it into the log may explain the difference. Rhythm calls it a pattern and stops there. It does not tell you what you should do next.

If a pattern involves medicine or supplements, the card says that your logs are not medical research. Your pediatrician still gets the last word.