A data collection and AI-assisted coaching framework for fasting athletes — built on real data from 12 athletes across a local gym during Ramadan 2026, and designed to scale with more participants.
Muslim athletes lack practical, accessible guidance for training during Ramadan. Existing research is too scattered to be useful in real time. This project builds a framework that shows how structured daily data collection and AI-assisted coaching could work for this population — prototyped well enough that someone could actually use or extend it.
We collected data from 12 athletes at a local gym during Ramadan 2026. N=12 is small, and we treat these findings as directional rather than definitive — a foundation to map onto published datasets, not a standalone result.
Four consistent patterns emerged across all 12 athletes. These are treated as directional signals to be mapped against published literature, not definitive conclusions.
The sharpest energy drop occurs in the first two weeks as the body adapts to fasting, with the lowest point around Week 3. By Week 4 athletes show measurable improvement — consistent with published research on physiological adaptation during Ramadan.
42% of all diary entries reported noticeable dehydration, peaking at 51% in Week 1. High-intensity athletes hit 58%. The limited hydration window between iftar and suhoor is the structural cause — post-iftar hydration protocols are the highest-impact intervention.
Only 52% of nights met 6+ hours. Week 2 was the worst at 44%. Days with adequate sleep produced recovery scores 0.6 points higher on average — making sleep the strongest predictor of next-day recovery in the dataset.
61% of sessions were completed as planned, 24% were modified, and ~15% were fully skipped. This shows intelligent self-regulation. The implication for coaches is to build structured modification plans into the Ramadan training schedule rather than expecting full compliance.
Even athletes who wanted to participate found consistent daily logging difficult — Tarawih, disrupted sleep, and normal training load all competed with it. This is a finding in itself: future versions need passive data collection (iPhone Health integration) rather than manual entry.
The AI coach gives advice that varies based on your sport, Ramadan week, and what you've logged. The decision logic is documented here so it's fully transparent and reviewable — not a black box.
| Score | What it means | Coach guidance |
|---|---|---|
| 1–2 | Very low | Reduce session intensity ~40%. Prioritize mobility or light skill work. Flag to coach. |
| 3 | Moderate | Maintain planned session with modified rep ranges or reduced volume. |
| 4–5 | Normal | Proceed with full session. Monitor hydration closely post-training. |
| Score | What it means | Coach guidance |
|---|---|---|
| 1–2 | Poor | Rest day or active recovery only. Do not stack consecutive low-recovery sessions. |
| 3 | Partial | Lower-intensity session acceptable. Prioritize sleep window before next session. |
| 4–5 | Full | Normal training load appropriate. |
| Group | Sports | Coach focus |
|---|---|---|
| High intensity | Track, Soccer, Wrestling | Aggressive hydration guidance, stronger intensity scaling, dehydration risk flagged early |
| Moderate intensity | Basketball, Volleyball, Swimming | Balanced guidance, session timing relative to iftar |
| Lower intensity | Tennis | Maintenance-focused, recovery optimization prioritized |
The Python notebook walks through 10 sections of analysis using pandas, matplotlib, seaborn, and scipy. Available in the GitHub repository.
Enter your sport, training intensity, gender, and current Ramadan day.
Rate energy, recovery, training completion, hydration, and sleep. Data saves to your account.
Your data overlaid on group averages — by week, day-by-day, or by sport.
Get advice specific to your sport, week, and what you've actually logged.
Download your full log as a CSV to share with your coach or for further analysis.
Set your profile, log your training days, and get guidance based on your sport, week, and logged data.
Open the Tracker →