What We Are Building Next, and What We Will Not Claim Yet
Force and load estimation without a force plate, and predictive early-warning. Both are in development. Neither is shipped, and we label them accordingly.
Motion analysis marketing has an honesty problem. Injury prediction in particular is routinely presented as a finished feature when, in most products, it is a research direction. We prefer to draw the line in public: this is what TropX does today, and these are two capabilities we are building but do not yet sell.
Today, TropX measures. Two sensors on each leg capture motion at high frequency. The pipeline reconstructs the legs in 3D, segments reps, scores sessions with OPI, tracks progress over time, and puts a whole caseload on one board. The AI explains what was measured and recalls comparable past sessions. All of that is the shipped product.
In development: force and load estimation
Ground reaction force is what a force plate measures, and force plates stay in the lab. We are working on estimating force and joint load from the motion signal itself, so that a squat in a clinic or a living room carries load information as well as kinematics.
The status is: in development. Estimating force from kinematics alone is a hard and well-studied problem. It becomes a clinical feature in TropX when it holds up against reference equipment, and not before.
In development: predictive early-warning
The measurement foundation makes a predictive layer possible. Session-over-session trends in asymmetry, smoothness, and load tolerance are the signals that should move before a setback turns into an injury. Our cross-session analysis already flags persistent asymmetries and declining metrics as patterns for the clinician to review.
Turning those patterns into calibrated risk forecasts is the part we refuse to rush. A risk score that fires too often gets ignored, and one that stays quiet before an injury is worse. Until the numbers earn it, this capability stays labeled as in development.
If you evaluate motion analysis tools, one question cuts through most of the noise. Which numbers are measured, which are estimated, and which are predicted, and how is each of them validated? We built TropX to have good answers to that question, and we would rather grow the list of answers slowly than blur it.