What I read, and what this site leans on: the guide, the studies, and the clinical
rules behind the findings. Ordered by how close each sits to the project: the guide
and its lineage, then the work behind our deeper questions, then the wider
background. Plain summaries on the ones that matter most; every link goes to the
original.
The guide, and what it builds on
The guide this site tests, and the clinical pacing tradition it grows from.
The patient-authored guide this whole site tests. Wiggers, living with ME/CFS and POTS, sets out how to read an ordinary Garmin for the signs of a post-exertional crash. It is the primary source for every pattern I tested.
The clinical lineage behind staying under a threshold, the resting-heart-rate ceiling and the two-day exercise test that showed the abnormal recovery in ME/CFS. The 'RHR + 15' kind of rule the guide builds on.
A specialist clinic's guidance on ME/CFS, pacing, and the 'energy envelope', crash prevention written for patients and clinicians, the clinical backdrop to the guide.
A patient-built method, in the same heart-rate-pacing tradition as the guide, for staying under an anaerobic threshold (a resting-heart-rate ceiling, often RHR + 15) to avoid post-exertional crashes. It carries a caveat that matters for this project's own pipeline: Garmin and Polar report resting heart rate as the night's *lowest* HR, which for this patient group may not be true resting heart rate; they recommend using your *average* overnight / sleeping heart rate as the proxy instead.
Work that goes past what the guide covers: the studies behind the deeper
questions we ask in beyond the guide (for instance,
whether a watch can predict a crash, not just recognise one after).
paper
Wearable biomarkers predicting same-day symptoms in long COVID
Aitken et al. · 2026 · npj Digital Medicine 9:257 (Visible app, n=4,244)
What it is & why it matters
The closest peer-reviewed parallel to this project: across 4,244 people, the Visible app's morning heart-rate and HRV readings helped predict that same evening's crash and brain fog. (Authored by the app's company, read with that in mind.)
paper
Feasibility of modelling next-day fatigue and sleepiness from sleep-tracker data (IDEA-FAST)
Zhai et al. (Newcastle · IDEA-FAST consortium) · 2026 · Frontiers in Digital Health 8:1752629 (n=134, 3,062 nights)
What it is & why it matters
The soberest benchmark for 'can a wearable forecast tomorrow's fatigue'. Across 134 people and 3,062 nights, using the honest test where whole people are held out (leave-one-subject-out), next-day physical-fatigue prediction reached only a modest AUC around 0.75 in healthy adults and fell to about 0.62 in neurodegenerative disease, and the authors call it exploratory. The realistic ceiling this project measures itself against: even a large, well-funded team gets weak, not clinical, prediction, which is why this site claims a weather report and not an alarm.
Recovery from exercise in ME/CFS (2-day CPET; symptom recovery duration)
Moore et al. · 2023 · Medicina 59(3):571
What it is & why it matters
The closest thing to a hard number on how long a post-exertional crash lasts. After a standardised two-day exercise test, 80 people with ME/CFS took a mean of about 12.7 days to recover (range 1 to 64) against about 2.1 days for controls, with a modelled 'extremely prolonged' decay. The literature prior behind this site's 'the feeling heals faster than the body' finding: a days-to-weeks recovery is the expected shape, not an anomaly.
Prolonged physiological and behavioural changes after COVID-19 (wearables)
Radin et al. · 2021 · JAMA Network Open 4(7):e2115959
What it is & why it matters
The clean population-scale evidence that the autonomic channel is the slowest to settle. Tracking wearable data through COVID-19, resting heart rate took on average about 79 days to return to baseline (after a brief early dip), while behavioural channels — step count and sleep — returned in about 32 and 24 days. The same-direction dissociation behind this site's recovery finding: the heartbeat-derived signal lags the behavioural ones by weeks.
What heart-rate variability is, what normal looks like, and how the body's autonomic balance shifts.
paper
An overview of heart rate variability metrics and norms
Shaffer & Ginsberg · 2017 · Frontiers in Public Health 5:258
What it is & why it matters
A standard reference for what heart-rate variability is and what normal looks like, the foundation for reading HRV at all, and for the sleep-stress proxy this site relies on.
Inter- and intra-individual variability in daily resting heart rate
Quer et al. · 2020 · PLOS ONE 15:e0227709 (n=92,457)
What it is & why it matters
Resting heart rate varies widely from person to person and day to day, across 92,000 adults. It is the baseline for judging what an unusual personal-RHR shift even means, and why this site reads signals against your own normal.
Pools 64 case-control studies and finds a reproducible autonomic shift in ME/CFS, higher resting and orthostatic heart rate, a sympathetic tilt, and lower vagal (parasympathetic) HRV. This is the source for the site's line that these bodies tend to sit tilted toward 'activated', with less beat-to-beat variability and slower to settle.
The circulation-on-standing side of the story: what POTS is, how it's diagnosed, and why a wrist watch can flag a pattern but never see the thing that actually defines it.
paper
Postural Orthostatic Tachycardia Syndrome: Mechanisms and New Therapies
Mar & Raj · 2019 · Annual Review of Medicine 71:235-248
What it is & why it matters
A clear map of what POTS actually is: an excessive heart-rate rise on standing, driven by one (or more) of three mechanisms, low blood volume, over-active sympathetic drive, or partial nerve damage. It is why 'POTS' on this site means a circulation-on-standing problem, and why a single watch signal can't tell the subtypes apart.
Why the site is careful with the word 'POTS.' In POTS the resting autonomic balance tips toward stress, heart-rate variability sits lower, not higher. The watch pattern the guide calls a 'U-dip' is a brief move the other way (a calming blip), so it can't be read as a POTS signal; it's a pattern the participant learned to manage as if it were one.
The honest limit, in one paper: in ME/CFS and long COVID, blood flow to the brain can fall on standing even when heart rate and blood pressure look completely normal. A wrist watch reads heart rate, so it is blind to this whole class of orthostatic trouble, the reason this site never claims the watch 'sees your POTS.'
How a consumer Garmin actually computes stress and heart rate, and how well it holds up.
paper
Validity of the Garmin stress score against ECG-derived HRV
Rosenbach et al. · 2025 · Stress and Health (Wiley)
What it is & why it matters
An independent, pre-registered, peer-reviewed study validating Garmin's stress score against gold-standard ECG heart-rate-variability. The evidence that the sleep-stress HRV proxy used here stands on something real.
Stress and Recovery Analysis Method Based on 24-hour HRV
Firstbeat Technologies · 2014 · White paper (Firstbeat, now Garmin)
What it is & why it matters
The white paper for the algorithm Garmin uses, how a 0–100 stress score is computed from beat-to-beat heart-rate-variability. The reason stress and HRV are two views of one underlying signal.
The biology and the lived experience behind the numbers.
paper
Mechanisms of long COVID and the path toward therapeutics
Peluso & Deeks · 2024 · Cell 187:5500–5529
What it is & why it matters
A flagship synthesis of what is known about Long COVID's biology and the search for treatments, the big-picture context for why a body might behave the way this one does.
A landmark synthesis of long COVID. It anchors the site's framing that post-exertional malaise is reported by a majority of patients, and that dysautonomia and POTS are commonly found, while noting that large-population, diagnosis-specific prevalence figures are still sparse.
The source for the site's line that standing intolerance is common while a confirmed POTS diagnosis is a smaller subset: this review reports that roughly 2–14% of COVID survivors develop POTS and 9–61% experience POTS-like symptoms in the months after infection, the figures swinging widely with who is studied and whether POTS is confirmed by a standing or tilt-table test.
Muscle abnormalities worsen after post-exertional malaise in long COVID
Appelman et al. · 2024 · Nature Communications 15:17
What it is & why it matters
Muscle biopsies before and after exertion in long COVID patients show real, physical damage after a crash, among the strongest evidence that post-exertional malaise is not simply deconditioning.
Long-term COVID-19 impact on heart rate variability: a systematic review
Suh, Kwon & Lee · 2023 · Healthcare 11:1095
What it is & why it matters
Pulls together the studies on how Long COVID affects heart-rate variability, the autonomic-dysregulation picture in one place, instead of reading each cohort.
The episodic nature of disability in long COVID (qualitative)
O'Brien et al. · 2023 · BMJ Global Health 8:e011276
SSRIs, the nervous system & long COVID
Why an antidepressant was tried for long COVID, and what SSRIs do (and don't reliably do) to heart-rate variability and the watch's numbers.
paper
Impact of antidepressant use on the autonomic nervous system: a meta-analysis and systematic review
Fiani et al. · 2023 · European Neuropsychopharmacology 71:75–95
What it is & why it matters
The best current synthesis of whether antidepressants move the autonomic nervous system. Across 30 studies it found SSRIs have no or inconclusive effects on most heart-rate-variability measures, and the one measure that did move pointed in opposite directions depending on study design. The anchor for this site's 'the HRV story is mixed' line, and for the distinction that the strongly HRV-lowering drugs are the older tricyclics, not SSRIs.
Differential associations of specific SSRIs with resting-state heart rate and heart rate variability
Kemp et al. · 2016 · Psychosomatic Medicine 78(7):810–8 (n≈10,500)
What it is & why it matters
The strongest single signal that SSRIs can lower HRV, and it names citalopram. In over 10,000 adults, users of every SSRI except fluoxetine had lower heart-rate variability than non-users, with citalopram's effect milder than paroxetine's. Cross-sectional, so it can't prove cause, but it's the closest the literature comes to a citalopram-specific result.
Serotonin reduction in post-acute sequelae of viral infection
Wong et al. · 2023 · Cell 186(22):4851–4867
What it is & why it matters
The paper that energised the 'give an SSRI for long COVID' idea: it proposed that post-viral inflammation depletes serotonin and thereby dampens vagus-nerve signalling and memory. Influential and ambitious, and, as the commentary filed next to it argues, methodologically contested. Read the two together.
Effect of early treatment with fluvoxamine on risk of emergency care and hospitalisation among patients with COVID-19 (TOGETHER trial)
Reis et al. · 2021 · Lancet Global Health 10(1):e42–e51
What it is & why it matters
The strongest randomised evidence for an SSRI in COVID. Fluvoxamine, started early in acute infection, cut hospitalisation, probably through an anti-inflammatory (sigma-1) action rather than its serotonin effect. Crucially this is about the acute illness, not treating established long COVID.
Carla Rus's study, the work that put this intervention on my radar. 95 post-COVID patients treated with SSRIs, most of whom improved, brain fog and sensory overload most of all. Exploratory by its own description: a questionnaire study with no control group, so it raises the hypothesis rather than settling it.
Are all antidepressants the same? The consumer has a point
Camino et al. · 2022 · Psychological Medicine 53(9):4004–4011
What it is & why it matters
Evidence that emotional blunting is a real, recognised SSRI effect, reported more often with serotonergic antidepressants than others, citalopram among them, and one of the side effects most tied to dissatisfaction. The backdrop to my own experience of feeling toned-down.
Resting heart rate, fitness, infection & what moves it
Why a trained body runs a low resting heart rate, how an infection and lost fitness move it, and how little of it is just age, the evidence behind the resting-heart-rate driver ledger.
paper
Effect of endurance exercise on autonomic control of heart rate
Carter, Banister & Blaber · 2003 · Medicine & Science in Sports & Exercise / Sports Medicine 33(1):33–46
What it is & why it matters
Why an athlete's resting heart rate runs low, endurance training shifts autonomic balance toward the parasympathetic and lowers resting heart rate (training bradycardia), and the effect fades when training stops. The mechanism behind this record's fitness driver: a trained-low resting heart rate that drifts back up as activity falls away, and that masked the infection's heart-rate rise in the COVID check.
Pre-symptomatic detection of COVID-19 from smartwatch data
Mishra et al. · 2020 · Nature Biomedical Engineering 4:1208–1220
What it is & why it matters
External evidence that a consumer wristband's resting-heart-rate signal really does move for an infection, here, shifts around COVID-19 that were often detectable before symptoms. The population-scale backdrop to this site's single-person COVID check: the watch moving for a known infection is an established phenomenon, not a one-off.
Doing a single-person study credibly, and other studies doing something similar.
paper
CONSORT extension for reporting N-of-1 trials (CENT)
Shamseer et al. · 2015 · BMJ 350:h1738
What it is & why it matters
The reporting standard for single-person (n-of-1) trials, the discipline for doing a study of one body credibly. The pre-registration practice on this site follows it.
These are background and foundation: surveys and studies of the field, plus the
guide and the clinical rules, not findings of this study. The method that turns the
data into findings is in the workings; the variables are in
the data dictionary.