Piotr Sędzik

Your Whoop says recovery 72%. What does that actually mean?

When your fitness app shows "Recovery Score: 72", do you know how it got that number?

With Whoop, Oura, Fitbit you don't. Black box. Can't audit it, can't tune it, can't verify it works for your users.

And it matters more than you'd think:

  • An algorithm tuned for 25-year-old athletes gives bad scores to 65-year-old cardiac rehab patients

  • Sleep scores built for monophasic sleepers break for shift workers

  • HRV baselines vary wildly between populations

With Open Wearables every algorithm is open. Read the code. Fork it. Adjust thresholds for your specific users. Building for clinical? Your medical team can verify every line.

Should health algorithms be auditable by default? Curious what you all think.

We launched on Product Hunt today. If you think open health scoring matters, support us with an upvote:

https://www.producthunt.com/products/open-wearables

68 views

Add a comment

Replies

Best
Yosun Negi

For me, this is exactly the problem with most wearables. I rely on them daily, but I’ve never been able to understand or question how my recovery score is calculated.

Assi Mahmood

Health scores shape behavior, so black-box algorithms shouldn’t be treated as unquestionable truth especially in sensitive or clinical contexts. If users are making decisions from those numbers, auditability and context should be standard, not optional.

Kamil Maksymowicz

@assi_mahmood Completely agree, and this is the argument that resonates most with clinical teams. A score that influences a recovery plan or training load needs to be explainable, not just accurate. That's why every algorithm in OW can be audited, forked, and tuned, and why we built HIPAA-ready architecture from the start.

Thami Benjelloun
Been on Whoop for almost 2 years now, and honestly at this point I can predict my recovery score most mornings before I even check 😅 Which probably says something about how “personalized” these models actually are vs. just learning your routine. Where it really starts to feel off is with non-standard lifestyles as you said, shift work, lots of travel, inconsistent sleep. The model assumptions break pretty fast there. It’s the case of my friend, Pilot 👨🏻‍✈️ The black-box issue is the bigger thing though. If people are adjusting training (or eventually health decisions) based on these scores, not being able to inspect or tweak the logic feels like a real limitation. What I’m curious about with Open Wearables: How do you think about baseline adaptation over time? Is the goal that teams fully own and tune scoring per user segment, or do most people start from a shared model and adjust? Feels like opening this layer up is the right long-term move.