FatigueSense app

[Utsikt]

We actually agree that long-term tracking of activity, symptoms, sleep, and perceived wellbeing is valuable. In fact, we’ve conducted a study with people experiencing long COVID over a three-month period, analyzing activity and fatigue patterns. That work is currently under peer review in the journal to be published, and we plan to expand to gain further insights into long-term effects.

Three months is not long term, it’s very much short term. One year is the bare minimum if you want to have any hope of getting data across fluctuations.

Regarding what the app is or isn’t, it really does not matter what you technically say it is. Nobody here has claimed it’s a medical advice according to the law.

It’s presented as, and more importantly, will be interpreted as, a tool that will provide the used with better insights about how to manage their day to day. It’s presented as being based on research, not as a tool for research.

It is also presented as developed by «digital health experts». The only reason to tell users that is to imply that it is endorsed by health experts.

And the website literally says it’s going to provide guidance:

It combines wearable signals, health trends, and AI analysis to deliver practical guidance people can actually use.

So there is some dissonance between what you say here, what the website says, and what the screenshots tell us..

Spoiler: From the website

Your Day at a Glance​

The Dashboard gives you an instant overview of your fatigue and energy levels. See your AI-predicted scores, daily briefing with personalised recommendations, and data quality indicators all in one place.

• 
  AI-predicted fatigue and energy scores updated daily
• 
  Personalised daily briefing with actionable recommendations
• 
  Quick actions for Activity Pacing and AI Insights
• 
  Morning assessment and end-of-day check-in prompts
• 
  Data quality monitoring alerts you to wearable sync issues

Know Tomorrow's Fatigue Today​

FS AI is the heart of FatigueSense. It analyses your wearable data to predict your fatigue outlook whether that's today's forecast or tomorrow's. The prediction adapts to time of day and gets more accurate as it learns your patterns.

• 
  Time-of-day aware: morning shows today's outlook, evening shows tomorrow's forecast
• 
  Top 3 drivers explain what's influencing your prediction in plain English
• 
  Personalised recommendations written in a supportive, non-clinical tone
• 
  Intraday check-in lets you agree or disagree with your predictioins
• 
  Personal model indicator shows when your custom prediction model is active

Your Personalised Activity Budget​

Get a daily activity budget tailored to your predicted fatigue level. FatigueSense tells you how many steps and active minutes you can safely do, broken into morning, afternoon, and evening blocks with built-in rest periods.

• 
  Personalised step and active minute targets based on your fatigue level
• 
  Time-blocked schedule: morning, afternoon, and evening with mandatory rest periods
• 
  Live progress tracking against your daily budget
• 
  Activity Planning tool to check if planned activities fit your capacity
• 
  Pacing history: see how past pacing choices affected next-day fatigue

Discover What Drives Your Fatigue​

AI Insights analyses your history to find patterns you might not notice. It identifies your personal fatigue triggers, ranks them by confidence, and shows you exactly what's behind your good and bad days.

• 
  Personal trigger identification (e.g., "Poor sleep → high fatigue next day")
• 
  Confidence levels and evidence counts for each trigger
• 
  Triggers update as more data is collected
• 
  tVNS correlation analysis for research participants
• 
  Understand the drivers behind your energy patterns

What-If Activity Simulator​

Curious how changing your sleep or activity would affect tomorrow's fatigue? The Activity Simulator lets you adjust variables like sleep duration, steps, and active minutes to preview the predicted impact before you commit.

• 
  Adjust sleep, steps, and active minutes with intuitive sliders
• 
  See a live preview of predicted fatigue based on your changes
• 
  Plan your activities with confidence — know the trade-offs in advance
• 
  Powered by your personal AI model for accurate, tailored results
• 
  Great for planning big days, recovery days, or pacing experiments

A Profile Built Around YOU​

FatigueSense doesn't use population averages — it builds a personal profile specific to your body. Over time, it learns your baseline HRV, typical sleep patterns, normal step counts, and what deviations from YOUR normal mean for YOUR fatigue.

• 
  Personal baselines for HRV, sleep, steps, and heart rate
• 
  Maturity indicator: Building → Emerging → Established → Robust
• 
  Deviation analysis: see how today compares to YOUR normal
• 
  Confidence score showing overall profile reliability
• 
  Profile refreshes automatically as new data arrives

Two Minutes That Make the AI Smarter​

A quick morning check-in and optional evening reflection. Rate your fatigue and energy on a smooth 0-100 scale, note any contributing factors, and help the AI calibrate its predictions to your experience.

• 
  Visual Analogue Scale (VAS) sliders — scientifically validated measurement
• 
  Real-time colour feedback from green (good) to red (severe)
• 
  End-of-day check-in closes the feedback loop for the personal profiling agent
• 
  Free-text fields for contributing factors

 

[Trish]

One thing I find confusing is the use of heart rate variability. It is said that HRV is anindicator of 'autonomic function' as if low variability is a sign of autonomic malfunction. It is also said to be low with deconditioning and higher with fitness. But presumably in the context of predicting tomorrow's fatigue it has nothing to do with these but is a measure of how often there was a bout of activity that put the heart rate up? Is this really usefully called 'HRV' rather than just a measure of tachycardia. I may be missing something here.

An Overview of Heart Rate Variability Metrics and Norms​

Fred Shaffer 1,*, J P Ginsberg 2

PMCID: PMC5624990 PMID: 29034226

Heart rate is the number of heartbeats per minute. Heart rate variability (HRV) is the fluctuation in the time intervals between adjacent heartbeats (1). HRV indexes neurocardiac function and is generated by heart-brain interactions and dynamic non-linear autonomic nervous system (ANS) processes.

 

[MrMagoo]

I was surprised Workwell Foundation wasn’t cited, but maybe that’s because it’s specifically ME/CFS, and they only cited two papers I think. Workwell - that’s the basis for Visible.

 

[Utsikt]

@nanay - would you be able to elaborate on the this from your website?

Edinburgh Venture Builder​

Enterprise and venture support linked to the University of Edinburgh.

If this is this programme,

Venture Builder Incubator is designed for ambitious postgraduate students, early career researchers and academics who are looking to transform their deep tech or data-driven research into a viable business.

How does that align with this previous statement of yours:

Finally, I think it’s important to emphasise context: this is early-stage academic research being translated into a free tool. No one involved is being paid to commercialise this — the aim is to explore, learn, and ultimately contribute to a space where there is still limited quantitative understanding of fatigue.

I guess you might technically not receive a salary from FatigueSense, but do you own any shares in the company? If so - isn’t that a conflict of interest that should be disclosed? Your profile describes it as a «health-tech startup», and there is some work from the Venture Builder from 2024. Yet the feasibility study published in late 2025 does not list ownership of the company as a CoI.

There is also no mention of any company on the website, the text at the bottom says:

© 2026 FatigueSense. An academic research project.

That does not align with what your profile says - where it’s a health-tech startup.

And as others mentioned earlier, it’s emphasised that the beta is free - does that mean that the end product will not be a free tool?

Edit to add: I have nothing against the commercialisation of research-derived tech - I’m just after transparency and clarity.

 

[rvallee]

Welcome @nanay, I hope we will have a productive discussion out of this. We are not a typical patient forum, not all us on here are patients to begin, and we are most of all a forum dedicated to the science of ME/CFS, with a much higher level of rigor.

As some have mentioned, there are many such apps already. Frankly, too many, so one thing that could be useful out of this would be to differentiate it from others, rather than try to implement the same idea. I think you do have some headway into this, but as was pointed out, the evidence and the language of the website don't quite match, and this is something that sets this forum apart: we notice.

As @Jonathan Edwards mentioned, like many such apps you use HRV as a signal, among others. This is mainly based on the deconditioning hypothesis, one that has been thoroughly debunked for many years. While many people with ME do have low HRV, some of us have a very normal, even quite good, typical HRV. This is contradictory data, it conflicts with the popular narrative of deconditioning being highly significant, from which is derived the idea of promoting activity.

Even though most of us are highly sedentary, so this actually even conflicts with the idea of HRV being a critical, objective measure of fitness, as people who have not walked more than 25 minutes at a slow pace for a decade should not have a normal HRV if it is truly a measure of fitness. I don't mean this only about HRV, it's just one of many, there are many such conflicting data, and they are far more interesting than trying to guide people in ways that no one has figured out yet.

Long Covid has thoroughly destroyed this narrative, but of course it will still be years for this basic fact to make its way into the literature. This is definitely something that could help move things along, far better than trying to provide tips and advice on how to move safely. No one has figured that out, in part because conflicting data is simply ignored. There is a lot of value in there. Science thrives on those inflection points, where someone notes "well, this shouldn't be happening", something that doesn't really happen when it comes to chronic illness.

 

[Utsikt]

@rvallee I don’t think the model has any preconceived notions about what the HRV of someone is supposed to look like. It just tracks patterns in an individual, and tries to use that to make predictions. Nana can probably clarify this.

It seems like the app collects a lot of data, and a longer project would be able to show if HRV doesn’t add any value - perhaps because it’s very sensitive to differences in when it’s measured etc. Time will tell.

 

[Utsikt]

Add this from a thread on Visible - their biometrics didn’t really add anything to the model beyond what you’d already get from the self-reports from the day prior.

The paper writes:

When only prior-day symptom predictors (i.e., lag-1 symptom reports) were included,models achieved AUC values were 0.78 for crash, .73 for fatigue, and 0.83 for brain fog. When morning biometric features were added to the prior-day symptom model, AUCs increased modestly across all outcomes, reaching 0.81 for crash, 0.74 for fatigue, and 0.85 for brain fog.

So the conclusion seems to be that prior-day symptoms are useful predictors but that biometric features add almost no additional information to that.

 

[MrMagoo]

Add this from a thread on Visible - their biometrics didn’t really add anything to the model beyond what you’d already get from the self-reports from the day prior.

Yup, Visible runs along these lines.
There have been at least two papers based on Visible and I think last years one was trying to make very strong claims about predictions.

I put my visible data in AI to analyse and I have a clear 2 day lag for PEM. Maybe people really are all individuals! Even AI had the good sense to tell me it gave itself a >65% evaluation of being correct, so it didn’t feel it was groundbreaking accuracy.

I wear visible and Fitbit. The Fitbit won’t let me turn off it’s dumb “hey, let’s have a break from training so hard today” feature, so it’s nice that Visible says “your body is out of balance just now”, which is the same thing but in ME language.

If I worked still, I’d like to be able to show my manager/colleagues that Visible message and pace point usage, I think it would have an impact. I’d worry about an NHS type version of visible being used as evidence that one day a week I’m ok and therefore should get a job.