Movement analysis in physiotherapy: What real-time feedback can do

Digital movement analysis is increasingly finding its way into physiotherapy and rehabilitation. Camera systems, motion sensors and other digital solutions can record movement characteristics during an exercise and give immediate feedback on how it is performed.

In everyday practice, however, the question is less whether movement can be captured technically. What matters is how reliable the respective measurements are for a specific task, how they are interpreted over time and which decisions still require therapeutic judgement.

How digital motion capture works technically

Several technical approaches are available for movement analysis. They differ considerably in terms of accuracy, effort and field of application.

Marker-based optoelectronic systems are used in biomechanics laboratories, among other settings, and are considered the reference method for many kinematic measurements. They allow detailed analyses, but require special cameras, markers on the body and a comparatively high amount of preparation and evaluation work.

Inertial sensors (IMUs) are worn directly on the body and record accelerations and angular velocities, among other things. They can be used on the move, but need to be positioned and calibrated correctly. Depending on the system and the measurement duration, measurement drift can also occur.

Markerless camera-based systems estimate body positions and movement patterns from image or depth information. They require no markers on the body and are therefore easier to integrate into training routines. Their measurement accuracy depends, among other factors, on the system used, the movement, the joint under observation, the plane of movement and the recording conditions.

For practice, this means: a camera-based system on the training floor should not automatically be equated with a biomechanical laboratory analysis. What counts is which movements and measurements the respective system is suitable for and how consistently the measurement conditions are maintained.

Which parameters can be useful for training control

Not every automatically generated measurement is automatically relevant in therapeutic terms. Particularly helpful are data that can be captured reliably and that support a specific decision during the course of training.

Depending on the system, these may include, for example:

• Repetition count and hold duration: They describe how often or for how long a defined task was performed.
• Movement tempo: With standardised exercises, it can indicate how execution changes over time.
• Range of motion in defined tasks: Larger movements, for example in squats or arm movements, can be observed over time under suitable conditions.
• Gross deviations from a defined target movement: For example, a marked compensatory shift of the upper body or a range of motion that is not reached.
• Development across several training sessions: Values recorded repeatedly under comparable conditions can make changes in the training process visible.

Caution is advised with very small angular differences, complex rotational movements and measurements in difficult planes of movement. Differences between the right and left side of the body should also not be interpreted on the basis of an automatically calculated value alone. What matters is whether the measurement accuracy of the system used is sufficient to distinguish relevant changes from normal measurement variation.

Equally important is the boundary between kinematics and kinetics. A camera can describe, or rather estimate, how body segments move. From this, however, it is not readily possible to derive which forces act on joints and tissue or which muscle activity underlies a movement.

Which measurements are genuinely helpful for digital training control is described in more detail in the article Digital exercise therapy: Which measurements really hold up.

Single value or trend: Why measurement conditions are decisive

A measurement can only be interpreted meaningfully if it is known under which conditions it was produced and how large the measurement uncertainty of the method used is.

Even small changes in camera position, starting position, clothing or movement sequence can influence the result. Measurement conditions should therefore remain as comparable as possible during follow-up checks.

In practice, the following points, for example, can be standardised:

• the patient's position in the room
• distance from and alignment to the camera
• starting position of the exercise
• exercise variant used
• load level or resistance
• relevant aids
• timing within the training session

A single value should therefore rarely be viewed in isolation. For training control, values recorded repeatedly under comparable conditions are often more helpful, because they allow developments to be observed across several sessions.

That said, the following still applies: a visible change in the data set is not automatically a clinically relevant change. Putting it into context remains part of therapeutic judgement.

Real-time feedback and motor learning

In addition to capturing movement data, digital systems can give feedback while an exercise is being performed. Both the type and the timing of the feedback play a role here.

A well-known principle from research into motor learning is the focus of attention. An external focus directs attention to the outcome or effect of a movement. An internal focus, by contrast, directs it more strongly towards individual body parts or movement components.

An example:

• internal focus: "Straighten your knee."
• external focus: "Push the floor away from you."

Studies on motor learning suggest that an external focus can offer advantages for execution and learning in many movement tasks. However, this does not mean that every visual display automatically creates a meaningful external focus. What matters is how the feedback is designed and which training goal is being pursued.

There is also no universally valid rule for feedback frequency. Frequent feedback can be helpful when learning a new movement, while in other situations reduced, summary or patient-requested feedback may be appropriate.

For everyday practice this means: feedback should not be given as often as possible, but in a way that suits the task and the patient. Particularly when a movement is later supposed to work without technical support, it can make sense to deliberately plan phases without immediate feedback and to check the transfer.

Movement analysis in physiotherapy with camera-based training systems

One example of digitally supported training with real-time feedback is the Pixformance Station.

According to the manufacturer, an integrated 2D full HD camera with depth sensor captures 25 joint points during training. A virtual trainer demonstrates the respective exercise on the display and gives visual and text-based corrective feedback during execution.

Training parameters are documented automatically and can then be viewed via the online platform. The therapist still defines the training plan and decides on exercise selection, load and any necessary adjustments.

This means recurring components of training such as exercise instruction, immediate feedback and training documentation can be supported technically. Automatic motion capture is not, however, the same as a complete therapeutic movement analysis.

Pixformance has also been used in clinical research projects. A controlled, non-randomised study published in 2025, for instance, investigated a technology-supported fall prevention programme with Pixformance in early geriatric rehabilitation. Among other things, the focus was on feasibility and usability. Such a study provides insights into the use of the system in a particular clinical setting, but is not the same as a general validation of all automatically recorded movement parameters.

Criterion	Pixformance Station	Strength equipment without automated feedback	One-to-one supervision by a therapist	App-based home programme
Feedback during execution	automated visual and text-based feedback	depends on equipment and supervision	individual feedback from the therapist	depends on the respective application
Motion capture	camera-based capture of defined movement points	depends on the respective device	therapeutic observation	depends on the technical capabilities of the app
Training documentation	training data are recorded automatically	depends on equipment and practice processes	therapeutic documentation required	depends on application and data capture
Individual adjustment	training plan can be adjusted therapeutically	depends on equipment and supervision	possible immediately and individually	depends on application and supervision
Manual and tactile assessment	not possible	possible by the therapist	possible immediately	not possible
Training outside the practice	possible via supplementary digital offerings	generally not intended	depends on the agreed self-directed programme	often part of the concept

The different approaches fulfil different tasks. Automated real-time feedback can be particularly supportive with recurring, clearly defined exercises. With complex movement problems, unclear symptoms or situations in which several influencing factors have to be taken into account, immediate therapeutic judgement remains central.

How the training floor and equipment can be integrated organisationally into existing workflows is described in the article Organising equipment-based physiotherapy: Space and scheduling.

Where therapeutic judgement remains irreplaceable

Digital motion capture describes certain visible or measurable characteristics of a movement. It does not, however, automatically explain why a deviation occurs.

Three areas therefore remain particularly closely tied to therapeutic judgement:

• Pain and response to load: An altered movement pattern can have various causes. Pain, uncertainty, strength deficits or other factors cannot be derived from a joint position alone.
• Manual findings and contextual factors: Clinical findings, neurological symptoms, comorbidities, medication and other individual factors can be relevant to the training decision, but are not fully reflected by camera-based motion capture.
• Goal setting and prioritisation: Which movement quality is aimed for at a given point in time is a therapeutic decision. A deviation from a technically defined target movement therefore does not automatically constitute an error.

A clear division of tasks makes sense: the therapist defines the training goal, exercise selection, load and limits. The digital system can give feedback during the performance of defined exercises and record training data. These data are then interpreted in the therapeutic context.

Implementation in everyday practice

If you want to integrate digital movement analysis into everyday training, you should first determine what the data collected are actually to be used for.

One possible approach:

1. Select a few suitable reference exercises: Use exercises that are relevant for the respective patient group and can be repeated under comparable conditions.
2. Standardise measurement conditions: Define position, exercise variant, relevant settings and other framework conditions as consistently as possible.
3. Define relevant measurements: Record only values that answer a specific therapeutic question or support a decision during the course of training.
4. Use feedback deliberately: Check when immediate feedback is helpful and when the patient should deliberately perform an exercise without technical correction.
5. Review trends regularly: Look at changes across several training sessions and not exclusively at individual measurements.
6. Define limits within the team: Clarify which data serve as orientation and when a renewed therapeutic assessment is required regardless of the technical feedback.

The organisational and financial side should also be taken into account when introducing such systems. How equipment and floor space costs can be calculated is described in the article Training equipment in physiotherapy: Calculating costs realistically. You will find supplementary key figures on utilisation and capacity in the Therapy Efficiency Report 2026.

Conclusion: Measurements complement therapeutic judgement

Real-time feedback can make movement patterns immediately visible, support patients during defined exercises and make changes across several training sessions documentable. Its value, however, depends on which parameters are recorded, how reliably the respective system measures them and how the data are interpreted therapeutically.

What matters is therefore not the largest possible number of measurements. Above all, information is useful when it is collected under comparable conditions and supports a specific decision during the course of training.

Digital movement analysis thus expands the options for training control – but it replaces neither assessment nor clinical judgement.