Sick or Under-Recovered? A Wearable-Based 48-Hour Decision Framework for Athletes (HRV, Resting HR, Temperature + Symptoms)

Short answer: if your HRV drops and resting heart rate rises, do not decide from one morning score. Use four signals together—HRV trend, resting HR drift, temperature/respiration trend, and symptoms—then re-test at 24-48 hours.

That is the practical decision logic SensAI uses to turn uncertainty into an action: Train Easy, Modify Session, Indoor-Only, or Full Rest. It is not a diagnosis. It is a safer way to make training decisions while your data and symptoms evolve.

Why athletes misread low recovery scores (and why one metric is not enough)

Most athletes misclassify recovery in two directions: they train hard into early illness, or they fully stop when they are only under-recovered.

Both errors come from metric tunnel vision. A low readiness score can reflect hard training, poor sleep, alcohol, menstrual phase shifts, travel stress, or an emerging infection. One metric cannot separate those states reliably.¹²³

Stanford researcher Tejaswini Mishra summarized the upside and the limit clearly: wearable data can help with early detection, but only when interpreted as a pattern, not as a single alarm. Her team concluded that “data from consumer smartwatches can be used for the pre-symptomatic detection of COVID-19.”³

SensAI’s approach is simple: trend first, context second, action third.

The 4-signal model: HRV trend + resting HR drift + temperature/respiration + symptoms

Use this sequence every morning:

1. HRV trend: compare to your personal baseline (not population norms).
2. Resting HR drift: check whether RHR is elevated vs your own baseline.
3. Temperature/respiration trend: look for sustained upward drift from your normal.
4. Symptoms: classify mild local symptoms vs systemic or chest/cardiac red flags.

Why this cluster works: multiple studies show wearables can detect presymptomatic physiology changes, but signal strength improves when combining streams.¹²³⁴⁵

Evidence snapshot (why 24-48h retesting is worth it):

• In the WHOOP/PLOS dataset (271 symptomatic individuals; 2,672 day-level samples), the model identified 20% of COVID-positive cases in the 2 days before symptoms and 80% by day 3 of symptoms.¹
• In COVI-GAPP (1,163 participants; ~1.5 million recorded hours), test-set sensitivity for COVID-19 detection up to 2 days before symptoms was 0.68.²
• In Stanford’s smartwatch cohort, 26/32 COVID cases (81%) showed physiological/activity alterations, and 22/25 were detected before or at symptom onset.³
• In controlled viral-challenge data, infection classification reached 92% for H1N1 and 88% for rhinovirus; severity prediction 24h pre-symptom reached 90% and 89%.⁴

Dean J. Miller’s WHOOP/PLOS analysis put it plainly: “changes in respiratory rate could serve as a leading indicator of SARS-CoV-2 infections.”¹

Minimum baseline requirements (7-14 nights) and what counts as a meaningful change

You need a personal baseline before thresholds are useful.

• Minimum baseline: 7 nights usable; 14 nights is better.
• HRV meaningful shift: roughly >8-12% below your rolling baseline for 2 consecutive mornings.
• RHR meaningful drift: roughly +3 to +7 bpm above baseline, especially if persistent.
• Temperature/respiration drift: sustained upward change for >=2 nights is stronger than a one-night spike.

The exact cutoff should be personalized, but the principle is stable: persistent deviation beats one-off noise.⁶⁷

The SensAI 48-hour decision algorithm (morning Day 1 -> re-test Day 2)

Day 1 is a triage day. Day 2 confirms direction.

Decision gates and thresholds: likely under-recovered vs possible illness

Use these gates in order.

Gate 1 — Physiology cluster

• HRV down >8-12% vs baseline and
• RHR up >=+3 bpm and/or
• temperature/respiration above normal trend

If 2+ are present, move to Gate 2 with caution.

Gate 2 — Symptoms

• Likely under-recovered: fatigue, heavy legs, poor sleep, mild soreness, no fever/chills, no chest symptoms.
• Possible illness: sore throat progression, chills, body aches, unusual fatigue, GI symptoms, new cough, or feeling systemically unwell.

Gate 3 — Confounders

• Hard block in last 72h?
• Alcohol above your norm?
• Menstrual phase where your baseline commonly shifts?
• Travel, heat load, or major sleep disruption?

If a confounder plausibly explains physiology and symptoms are mild, default to conservative Modify and re-test in 24h.⁸⁹

Gate 4 — 24-hour trend test (Day 2 morning)

• If metrics normalize and symptoms improve -> likely under-recovery.
• If metrics worsen or symptoms spread/systemic -> possible illness, reduce load further.

Martin Risch’s COVI-GAPP group reached the same practical direction: “Wearable sensor technology can enable COVID-19 detection during the presymptomatic period.”²

Pathway outputs: Train Easy / Modify Session / Indoor-Only / Full Rest

1) Train Easy (Green)

Use when: one mild metric deviation, no systemic symptoms, no red flags.

• Reduce intensity 10-20%
• Keep session short and technical
• Avoid maximal intervals

2) Modify Session (Yellow)

Use when: 2-signal deviation or mild symptoms without red flags.

• Cut volume 30-50%
• Keep HR zones low/moderate
• Prefer skills, mobility, aerobic base

3) Indoor-Only (Orange)

Use when: symptoms suggest possible infection but stable breathing/chest status.

• No group training
• No high ventilation sessions
• Keep effort easy; stop if symptoms escalate

4) Full Rest (Red)

Use when: strong multi-signal divergence plus systemic symptoms, fever trend, or red flags.

• No structured training
• Recovery-only day
• Reassess in 24h or seek medical advice sooner

SensAI uses these four outputs because athletes need a clear action, not just a score.

Confounder checks before you call it “sick” (hard training block, alcohol, menstrual phase, travel/jet lag, heat, poor sleep)

Confounders regularly mimic illness in wearables.

• Menstrual phase: In a large wearable cohort (n=11,590), mean cycle-related amplitude was 2.73 bpm for RHR and 4.65 ms for RMSSD, enough to look like “poor recovery” if phase is ignored.⁸
• Alcohol: In 20,968 participants and >5 million person-days, one extra drink above personal norm was associated with nocturnal RHR increases (+2.8 bpm women, +2.4 bpm men) and HRV reductions (-3.8 ms women, -3.3 ms men).⁹
• Travel/jet lag/heat/sleep disruption: common sources of transient HRV and RHR drift.

The practical rule: if confounders are high and symptoms are low, do not label it illness on Day 1. Use the 24-hour retest.

Device-specific translation (WHOOP Recovery, Garmin Training Readiness, Oura Readiness) into the same framework

Device scores are different interfaces for similar biology.

• WHOOP Recovery: color bands are useful only when paired with HRV/RHR and symptom context.¹⁰¹¹
• Garmin Training Readiness: treat score drops as prompts to inspect trend + symptoms, not standalone stop signs.¹⁰
• Oura Readiness: same rule—one low day is less important than 2-day direction.

SensAI normalizes these into one decision layer:

• Score down but symptoms absent + confounder present -> Modify + retest.
• Score down + symptom progression + worsening trends -> Indoor-Only or Full Rest.

This keeps decision quality consistent regardless of wearable brand.

Red-flag symptoms and medical escalation rules (not a diagnosis)

This framework is for training decisions, not diagnosis.

Escalate to medical care (and avoid training) for:

• Chest pain, chest pressure, palpitations, syncope/presyncope
• Shortness of breath out of proportion
• Persistent fever, severe worsening symptoms, or neurologic symptoms
• Cardiac involvement symptoms after viral illness

As Ruuskanen and co-authors emphasize, “symptoms of cardiac involvement … are absolute contraindications for exercise.”¹²

Context matters: in a 6,138-athlete review, clinically diagnosed myocarditis after COVID-19 was 1.2%, and myocarditis-attributable sports sudden cardiac death incidence was estimated at 0.047 per 100,000 person-years.¹² Rare does not mean ignorable.

Practical implementation: 5-minute morning checklist + coach notes + 48-hour follow-up

Use this daily checklist before deciding:

1. Open wearable trend view (HRV, RHR, temperature/respiration).
2. Mark symptom status (none, mild local, systemic, red-flag).
3. Log confounders (hard session, alcohol, menstrual phase, travel, heat, sleep).
4. Choose pathway (Train Easy / Modify / Indoor-Only / Full Rest).
5. Set Day 2 review note (what must improve to progress).

Coach note template:

• “Day 1: HRV -11%, RHR +5 bpm, temp up, mild sore throat, no chest symptoms, travel yesterday -> Indoor-Only, retest tomorrow.”

If Day 2 markers improve, progress one level. If they worsen, de-load further and consider medical input.

SensAI differentiated angle: personalized baselines, context-aware prompts, and adaptive thresholds from your wearable + training log

Most tools stop at descriptive scores. SensAI is built for prescriptive decisions.

SensAI combines:

• Personalized baselines from your own trend history
• Context-aware prompts (sleep, menstrual phase, travel, alcohol, hard blocks)
• Adaptive thresholds linked to your training log and symptom trajectory
• Action outputs that coaches and athletes can execute immediately

That combination is why SensAI can reduce false alarms while still catching meaningful risk windows.

If you coach athletes at scale, this is the practical edge: fewer overreactions, fewer missed warning signs, and clearer day-to-day training calls under uncertainty.

Internal resources and service links

• SensAI Home
• About SensAI
• SensAI FAQ
• Contact SensAI
• Race-week readiness framework
• DOMS vs injury framework

FAQ (direct answers to athlete search queries)

Should I train when HRV is low and resting HR is high?

Usually Modify first, not full stop, unless symptoms are systemic or red-flag. Re-test in 24 hours. Persistent divergence plus symptom progression shifts to Indoor-Only or Full Rest.⁶⁷

Low HRV + elevated RHR: am I getting sick?

Possibly, but not always. Hard training, poor sleep, alcohol, and menstrual phase can create the same pattern. Use confounder checks plus Day 2 trend direction before concluding illness.⁸⁹

Workout when getting sick using wearable data: what should I do?

Use the 4-pathway model. If symptoms are mild and stable, choose a reduced, easy session. If symptoms escalate or chest/cardiac signs appear, rest and seek care.¹²¹³

WHOOP low recovery: sick or overtraining?

Treat WHOOP color as a starting point. Add symptom pattern, RHR drift, and temperature/respiration trends. SensAI’s rule: one low score is caution; multi-signal worsening is action.¹¹¹

Garmin Training Readiness low: should I skip training?

Not automatically. If the low score aligns with worsening physiology and symptoms, de-load. If confounders explain it and symptoms are absent, modify and re-test.¹⁰

Oura readiness low but I feel fine. Train or rest?

If you feel well, symptoms are absent, and the next 24-hour trend is stable, Train Easy or Modify is usually reasonable. Avoid maximal efforts until the trend confirms recovery.

How many bpm above resting HR means rest day?

There is no universal number. For most athletes, sustained +3 to +7 bpm above personal baseline is meaningful, especially when paired with HRV suppression or symptoms.⁶⁷

Does HRV drop before illness in athletes?

Often yes. Multiple cohorts show wearable signal changes can appear before symptom onset, but predictive certainty improves when combining HRV with other signals and symptom tracking.¹²³⁴

Is this a diagnosis tool?

No. This is a training decision framework. Diagnosis and treatment decisions belong to licensed medical professionals.

---

Footnotes

1. Miller DJ, Capodilupo JV, Lastella M, et al. “Analyzing changes in respiratory rate to predict the risk of COVID-19 infection.” PLOS ONE (2020). https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0243693 ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶

2. Risch M, et al. “Wearable sensor technology for early detection of COVID-19: COVI-GAPP study.” BMJ Open (2022). https://bmjopen.bmj.com/content/12/6/e058274 ↩ ↩² ↩³ ↩⁴ ↩⁵

3. Mishra T, Wang M, Metwally AA, et al. “Pre-symptomatic detection of COVID-19 from smartwatch data.” Nature Biomedical Engineering (2020). https://pubmed.ncbi.nlm.nih.gov/33208926/ ↩ ↩² ↩³ ↩⁴ ↩⁵

4. Natarajan A, Su H-W, Heneghan C, et al. “Assessment of physiological signs associated with influenza and rhinovirus infection from wearable sensors.” JAMA Network Open (2021). https://pmc.ncbi.nlm.nih.gov/articles/PMC8482058/ ↩ ↩² ↩³

5. Smarr BL, Aschbacher K, Fisher SM, et al. “Fevers and the online tracking of body temperature with wearable sensors.” Scientific Reports (2020). https://www.nature.com/articles/s41598-020-78355-6 ↩

6. Sanches CA, et al. “Wearables and HRV for health monitoring: systematic review and meta-analysis.” Journal of Medical Internet Research (2023). https://www.jmir.org/2023/1/e47112 ↩ ↩² ↩³

7. Leitner J, et al. “HRV and orthostatic responses in an elite endurance athlete during upper respiratory tract infection.” Frontiers in Sports and Active Living (2021). https://www.frontiersin.org/journals/sports-and-active-living/articles/10.3389/fspor.2021.612782/full ↩ ↩² ↩³

8. Moreno-Pino F, et al. “Cardiovascular fluctuations across the menstrual cycle in wearables.” npj Digital Medicine (2024). https://www.nature.com/articles/s41746-024-01394-0 ↩ ↩² ↩³

9. Grosicki GJ, et al. “The impact of alcohol consumption on wearable-derived physiology and behavior.” PLOS Digital Health (2026). https://journals.plos.org/digitalhealth/article?id=10.1371/journal.pdig.0001284 ↩ ↩² ↩³

10. Garmin. “How getting sick might change your heart metrics.” Garmin Blog (2025). https://www.garmin.com/en-US/blog/fitness/how-getting-sick-might-change-your-heart-metrics/ ↩ ↩² ↩³

11. WHOOP. “What does an infection do to your respiratory rate?” WHOOP Locker (2022). https://whoop.com/en-gb/thelocker/what-does-an-infection-do-to-your-respiratory-rate ↩ ↩²

12. Ruuskanen O, et al. “Sport and exercise during and after viral acute respiratory illness.” (2024). https://pmc.ncbi.nlm.nih.gov/articles/PMC11282332/ ↩ ↩² ↩³

13. Gluckman TJ, Bhave NM, Allen LA, et al. “2022 ACC expert consensus decision pathway on cardiovascular sequelae of COVID-19.” JACC (2022). https://www.jacc.org/doi/10.1016/j.jacc.2022.02.003 ↩