--- title: "Exoplan energy algorithm · Notes · Mike Lvoff" description: "How Exoplan finds your energy peaks: 8 weeks of Apple Health data, cleaned, pooled by weekday and fitted into a curve. With the six studies behind it." url: https://mikelvoff.com/notes/exoplan-energy-algorithm author: Mike Lvoff published: 2026-10-12 updated: 2026-10-12 --- # Exoplan energy algorithm [Mike Lvoff](https://mikelvoff.com/) Research note · 3 min read ![The Apple Health and Exoplan icons above a day of readings drawn as points, with the high-energy hours glowing green.](https://mikelvoff.com/_astro/exoplan-energy-algorithm.DrfFgu6f_Z2q5EbV.webp) TL;DR Exoplan estimates your energy from your heart rate at rest, which rises and falls through the day. The algorithm takes 8 weeks of Apple Health data, drops workouts and odd readings, and builds a typical curve for each weekday. The hours when the curve runs high become your energy peaks, and your sleep quality that morning sets how long they are. ## Goal My original desire was to plan my days as efficiently as possible to achieve maximum productivity. ## Research foundation - Schmidt et al. Cognitive Neuropsychology · 2007 ### [Focus has peak hours](https://doi.org/10.1080/02643290701754158 "A time to think: circadian rhythms in human cognition") Attention and memory change through the day. Chronotype shifts the timing. - Dijk, Duffy and Czeisler Journal of Sleep Research · 1992 ### [Alertness has two drivers](https://doi.org/10.1111/j.1365-2869.1992.tb00021.x "Circadian and sleep/wake dependent aspects of subjective alertness and cognitive performance") Alertness depends on the body clock and time awake. - Vandewalle et al. Journal of Sleep Research · 2007 ### [Heart rate follows the body clock](https://doi.org/10.1111/j.1365-2869.2007.00581.x "Robust circadian rhythm in heart rate and its variability: influence of exogenous melatonin and photoperiod") Measured around the clock in a lab, heart rate peaked in the afternoon. - Bowman et al. Cell Reports Methods · 2021 ### [Wearables can read your body clock](https://doi.org/10.1016/j.crmeth.2021.100058 "A method for characterizing daily physiology from widely used wearables") In more than 130,000 days of data from doctors in training, people’s rhythms varied a lot. - Persson et al. IEEE EMBC · 2019 ### [Sleepiness shows up in heartbeats](https://doi.org/10.1109/EMBC.2019.8857229 "Heart Rate Variability for Driver Sleepiness Classification in Real Road Driving Conditions") On real roads, heart rate variability sorted drivers into sleepiness levels with 61% accuracy. - Luo et al. Digital Biomarkers · 2020 ### [Heart data helps predict fatigue](https://doi.org/10.1159/000512166 "Assessment of Fatigue Using Wearable Sensors: A Pilot Study") In a 27-person pilot, wearable data predicted physical fatigue with about 70% precision. ## High-level overview ### 1\. Collect For each day: heart rate, active energy, and when you fell asleep and woke up. Time counts from wake-up, so early and late mornings line up. ![One synthetic Monday from 6:00 to midnight. Heart-rate readings sit mostly between 55 and 75 bpm and jump during walks and a lunchtime run. Active energy bars below spike at the same times.](https://mikelvoff.com/_astro/chart-collect.D-K-YJTx_ZklIHl.png) One Monday, as the app sends it. ### 2\. Clean Days with an unusual wake-up time are dropped. So are the first hour after waking, the last hour before bed and any time you were moving. Odd readings go too, and the rest is averaged in 20-minute steps. ![The same Monday in hours since waking up. Grey bands cover the first hour, the walks, the run and the last hour. A few readings are circled in amber as dropped, and green dots show the 20-minute averages.](https://mikelvoff.com/_astro/chart-clean.fiKunb_1_1VKriS.png) The same Monday after cleaning. Grey bands are cut out and amber rings are dropped. ### 3\. Fit Each day is shifted to the same average, and the same weekdays from every week are pooled. A smooth curve runs through each pool, and the hours above a sloped line become the peaks. ![Dots from eight synthetic Mondays with a smoothed curve through them and a dashed line sloping down. The curve turns green where it rises above the line, around 4 hours and from 9 to 12 hours after waking up.](https://mikelvoff.com/_astro/chart-fit.y4qbkxWo_Z1ssixP.png) Eight Mondays pooled. The curve sits above the line for 4 of the first 16 hours. ### 4\. Output Your Exo Score each morning sets how long the peaks are. The app then lines them up with when you actually woke up. Until your watch data comes in, Exoplan places the peaks from your chronotype. ![One synthetic Monday at Exo Scores 30, 60 and 90, waking up at 7:00. The higher the score, the longer the energy peaks: at 30, one peak from 15:30 to 19:15, 3 hours 45 minutes in all; at 60, 10:00 to 12:00 and 15:30 to 19:30, 6 hours; at 90, 9:45 to 12:15 and 15:15 to 19:45, 7 hours.](https://mikelvoff.com/_astro/chart-output.BWUgSAV3_Z1i4Esh.png) A Monday at three Exo Scores, waking up at 7:00. ## Caveats The curve shows a typical Monday, so a hard Monday and an easy one start from the same peaks. Heart rate also rises with stress, and the model reads that as energy. Days with an unusual wake-up time are dropped, so travel leaves gaps. I started with 12 weeks of data and moved to 8. I also tried lining days up by shifting them in time. That one stayed an experiment. [Exoplan](https://apps.apple.com/app/apple-store/id6739639089?pt=127298084&ct=mikelvoff-energy-note&mt=8): Planning, tasks, AI assistant ## Contact - [mike.lvoff@gmail.com](mailto:mike.lvoff@gmail.com) - [LinkedIn](https://www.linkedin.com/in/mike-lvoff/) - [GitHub](https://github.com/mike-lvoff)