Our method

Written rules, published studies, and you decide.

Pantari adjusts your training load with rules taken from published studies. Here they are, where they come from, and what has not been checked yet.

Published on 8 October 2026. Reviewed on 9 October 2026.

In short

Your session is built by written, tested rules, never by an AI. No AI judges your discomfort.
These rules come from published studies, cited below with their limits.
You stay in charge: every piece of advice can be turned down with one tap, except after a warning sign.
Pantari makes no diagnosis and does not replace a health professional.
No physiotherapist has reviewed this content yet. It is planned: the date and the findings will be published here.

What the app listens to

Your discomfort during the session, on a 0 to 10 scale.
Your discomfort when you wake up the next morning.
Your perceived effort, and what you actually ran (the app’s timer or your watch file).

How the app decides, after every session

The rules are checked in this order; the first one that applies decides.

A warning sign (discomfort at rest or at night, swelling, tingling)The program stops and the app invites you to see a health professional.
Discomfort during exercise above your return’s thresholdThe session ends walking; the next one starts one stage lower. For a zero-discomfort return, a break until it goes away.
Discomfort rising from one week to the nextOne stage lower.
Waking up above your usual levelThe stage stays the same and the next session is a little shorter. Clearly above, one stage lower and an invitation to see a professional.
OtherwiseThe session is validated. You move up when your return allows it: a number of validated sessions, and for some a minimum time per stage.

When a rule has no dedicated study, it is a cautious choice by Pantari, and we say so.

A threshold for each return

The discomfort you can tolerate is not the same for a bone, a tendon or a joint. Each return has its discomfort threshold and its next-day rule, taken from the studies below. The app applies them for you, after the starting questions: they only make sense with those and the next-day follow-up.

Return	Stages	Studies
Stress fracture	McKay	Warden 2021 · George 2024 · Hoenig 2025
IT band syndrome	McKay	Geisler 2021 · Fredericson 2005
Runner’s knee	McKay	Willy 2019 · Esculier 2018 · Neal 2024
Plantar fasciitis	McKay	Koc 2023 · Riel 2019
Shin splints	Moen	Moen 2012 · Winters 2013
Achilles, at the heel	McKay, at least 2 weeks per stage	Pringels 2025 · Fahlström 2003 · Chimenti 2024
Achilles tendon	McKay	Silbernagel 2007 · Chimenti 2024 · Sprague 2021

Something else stopped you? Pantari offers a general, cautious return, with the strictest thresholds.

Stages, never a date

Running comes back in stages taken from published ladders. McKay et al. (2020): six five-minute blocks where running grows from 30 seconds to 4 minutes 30, then 30 minutes continuous. Moen et al. (2012), for shin splints: six phases, up to 16 to 18 minutes continuous.

You move to the next stage when your sessions have shown it, not because a week has gone by. If you have a race in mind, the app gives you rough guides in ranges of weeks, recalculated after every session, never a date to meet.

Before you start

Eight questions check that Pantari is the right tool for you today. They follow the warning signs published by the French National Authority for Health (HAS, 2019) and the signs of stress fracture described by the American Academy of Family Physicians.

If one of them points to it, the app offers no program: it simply tells you so and gives you a summary of your answers to show a health professional.

The exercises

125 exercises are served today. Each one cites its sources (75 studies and documents in all), says what they support, and what they do not quantify.

7Several studies: Guidelines or several consistent studies.
12One study: At least one study has tested the exercise.
106Common practice: No study has tested this exact exercise; the ones it cites support its principle. The app shows this plainly.

To stay cautious: no jumps or impacts at the first stages, and volumes at the low end of published ranges. When no study quantifies a volume, the exercise says so and stays cautious.

How the exercises are written

Studies are searched in PubMed and OpenAlex, for each need of a return.
An AI drafts a first text from those studies only.
Automatic checks reject an exercise whose contraindication cites no source, or that uses care vocabulary.
Pantari’s founder reads and validates every exercise, one by one, before it is served.
A physiotherapist will review the whole set. This has not been done yet.

AI is only used to write the exercises. Your session, your progression and the reading of your discomfort use none.

What has not been checked yet

No physiotherapist or lawyer has reviewed the rules and texts yet; the reviews are planned.
The thresholds come from studies on groups of runners: they say nothing about your own case.
No study has been run on Pantari itself yet. We will publish no outcome figure without one.

Independent

No advertising, no commercial partner, no data resale. Pantari lives on its subscriptions, once early access is over.

Found a mistake?

A rule looks wrong, a source is miscited? Write to contact@pantari.com. We fix it, and the fix is dated here.

Versions

9 October 2026 · Shin splints: the Winters review (2013) replaces their 2018 study, which is about recognising shin splints, not about the return.
8 October 2026 · First version.