Your Fitness Data Belongs to You — But It Doesn't Yet

The Invisible Lock-In

Consider what happens when you train on a cable machine at a commercial gym. The machine has sensors. It knows the exact height of the pulley, the weight you selected, the number of reps you completed, and the arc of your movement. That data is generated by your body. It describes your body. And yet, in virtually every case, it is owned by the equipment manufacturer — if it is stored at all.

This isn't a conspiracy. It's simply the natural result of how fitness technology evolved. Each manufacturer built their own ecosystem, their own app, their own data format. Peloton stores your data on Peloton's servers. Technogym has its own cloud. Your physical therapy clinic records your range-of-motion assessments in a paper chart or a proprietary EMR field. Your trainer tracks your programming in a spreadsheet that lives on their phone.

None of these systems talk to each other. More importantly, none of them were designed to hand your data back to you in a format that another system could understand.

What You Actually Lose

The practical consequences are easy to underestimate until you experience them directly.

Starting from zero, again and again

You switch gyms, and your training history doesn't come with you. You see a new physical therapist after moving to a different city, and they have no data from the last eighteen months of rehabilitation work you did with your previous provider. You get a new fitness tracker, and your previous metrics are stranded in the app of the device you replaced. Every transition means starting over — not just in terms of records, but in the sense that whoever is now helping you has no objective baseline to work from.

This isn't just inconvenient. For someone recovering from injury, the loss of accurate prior movement data can meaningfully slow rehabilitation. For a competitive athlete, losing training history breaks the continuity that coaches rely on to understand how a body adapts over time.

The AI problem

The inability to move fitness data across systems matters even more now that AI coaching and analysis tools are becoming genuinely capable. These systems are only as good as the data they have access to. A personalized AI coaching tool needs longitudinal movement data — not just last week's workout, but years of how your body responds to loading, recovers between sessions, and compensates for previous injuries.

That data exists, scattered across six apps, two gyms, a physical therapy clinic, and an old wearable you replaced three years ago. But it might as well not exist, because no AI system can reach across those silos to assemble a coherent picture of how you move.

Clinical disconnect

The gap between athletic training data and clinical data is perhaps the most consequential. A sports medicine physician evaluating your knee can look at imaging. They can do a physical assessment. What they almost certainly cannot do is pull up the last three months of load data from your training, cross-referenced with the range-of-motion measurements from your physical therapy sessions, displayed alongside your own subjective notes from training days when you felt something wasn't right.

All of that data was generated. Most of it was even recorded, somewhere. But it exists in formats that weren't designed to be combined, held by organizations that weren't incentivized to share it, and owned by parties who are not you.

Why This Happened: The Silo Problem

Understanding why fitness data became this fragmented requires understanding that fitness technology didn't develop as a unified field. It developed as a collection of separate industries that happened to share a subject — the human body in motion — but were solving different problems for different markets at different times.

Gym equipment manufacturers were building physical products. Their innovation priority was the hardware. Digital features were secondary, often added as afterthoughts or marketing differentiators rather than core functionality. Data formats weren't standardized because there was no competitive reason to standardize them — interoperability would have benefited users but reduced platform stickiness.

Sports science and biomechanics research developed its own set of file formats — C3D for motion capture, MATLAB-based workflows for force plate analysis, proprietary EMG software. These formats are rigorous and precise, but they were designed for laboratory researchers, not for a gym member trying to share data with their physical therapist.

Healthcare and physical therapy operate under their own constraints. HIPAA compliance requirements shaped how patient data is stored and shared. Electronic Medical Record systems were built to satisfy regulatory requirements and billing workflows, not to serve as interoperable data platforms for movement professionals.

Wearable technology came along later still, with its own generation of apps and ecosystems built around subscription revenue and hardware lock-in.

None of these fields planned to create a fragmented mess. They each solved their own problems well. The fragmentation was an emergent property of parallel development without coordination.

What an Open Standard Would Change

The solution is not a new app. It is not a cloud platform that promises to connect everything. Those approaches have been tried, and they tend to reproduce the same lock-in problem in a new wrapper — now your data lives in the aggregator's silo instead of the equipment manufacturer's silo.

What's needed is an open notation standard — a plain-text, human-readable, machine-parseable format for describing human movement that any system can write and any system can read, without requiring permission from a vendor or payment of a licensing fee.

This is the model that worked for the web. HTML is an open standard. Email is an open standard. These technologies created enormous value precisely because they didn't require you to use one company's product to communicate with another company's product. The format itself was the commons.

A movement notation standard built on the same principle would let your cable machine's firmware write a standard movement record, your wearable's app export in the same format, your physical therapist's assessment software produce a compatible file, and your AI coaching tool ingest all of it without custom integration work for each data source.

What MNN Looks Like in Practice

This is the problem that motivated the development of MNN — Muscular Neuro Notation — within the AIUNITES project. MNN is a plain-text notation format for human movement that attempts to encode the information that actually matters: which muscles are active, how active, at what joint positions, against what resistance vectors, and with what compensation patterns.

A single MNN string for a cable row looks like this:

This string contains the movement pattern (horizontal pull), the muscle activation profile (rhomboids at high activation, biceps and paraspinals supporting), the joint position (shoulder at 45 degrees of flexion, elbow at 90 degrees), the resistance vector (horizontal, mid-height, cable source), and the loading parameters (60kg, time under tension 3 seconds eccentric, 1 second pause, 3 seconds concentric).

That single line of text could be generated by equipment firmware, written by a trainer in a training log, parsed by a rehabilitation software system, processed by an AI analysis tool, or used to drive a virtual avatar. It doesn't require any of those systems to know about each other. It just requires that they all speak the same language.

What makes it different from existing formats

Existing movement data formats tend to be either too specialized or too general. Motion capture formats like C3D are precise but describe kinematics (where body segments are in space) rather than the physiological and training context that matters for coaching and rehabilitation. General-purpose health data formats like HL7 FHIR can technically represent exercise data but weren't designed for it and don't have the vocabulary to express meaningful details like muscle activation levels or resistance vector angles.

MNN is designed specifically for the intersection of training, rehabilitation, and digital representation of human movement. It was built to be rich enough to carry the information that coaches, clinicians, and AI systems actually need, while remaining simple enough to be written by hand, read without a special viewer, and parsed with a straightforward grammar.

Data Portability as a Design Principle

The fitness industry is slowly moving toward greater data portability, partly in response to regulatory pressure and partly because users are increasingly sophisticated about demanding it. Apple Health and Google Fit have created some degree of aggregation at the wearable layer. The European Health Data Space regulation is pushing healthcare systems toward greater interoperability.

But these developments tend to create portability at the level of simple metrics — steps, heart rate, sleep duration — rather than at the level of rich movement data that would actually serve training and rehabilitation goals.

True movement data portability requires a format expressive enough to capture what a skilled coach or clinician would actually care about. Steps and heart rate are useful for general health monitoring. They are not sufficient for understanding why someone's left shoulder is compensating on overhead pressing movements, or whether the load progression over the last twelve weeks has been appropriate given their recovery patterns.

That level of data representation requires a notation system designed for the purpose — and a commitment from the fitness and healthcare technology communities to treat that notation as a commons rather than a competitive asset.

The Path Forward

Open standards succeed when they reach critical adoption mass — when enough implementations exist that the cost of not supporting the standard becomes higher than the cost of implementing it. Getting there requires the standard to be genuinely useful, freely available, and easy to implement.

MNN is designed with all three constraints in mind. The full specification is publicly available. The format is plain text with a defined grammar, not a binary format requiring a special library. The notation is designed to be writable by humans as well as machines — a trainer who learns MNN can write movement records by hand in a text file, and those records are immediately compatible with any software that implements the parser.

The goal is not to replace existing fitness data systems. It is to give those systems a common language for the parts of their data that describe human movement — so that a cable machine, a clinical assessment tool, a training log, and an AI coaching platform can all speak fluently about the same body moving through the same exercises, without any of them needing a custom integration with any of the others.

Your fitness data belongs to you. Getting it to actually behave that way is an engineering problem with a clear solution — and open notation is the core of that solution.