Learn how AI and the latest CPM technologies optimize patient rehabilitation

By Avnet Staff   |   February 21, 2024

Continuous passive motion is a technique used in rehabilitation after surgery. Advances are being made in the control systems to deliver better results.

Canadian orthopedic surgeon Robert Salter conceived the use of continuous passive motion (CPM) as a therapy in the early 1970s. In 2018 the CPM market was worth more than $650 million.

Advances in electronic integration and control techniques continue to improve cost and performance. Now artificial intelligence (AI) incorporated into designs helps optimize each patient’s therapy.

CPM therapy works by supporting and moving limbs on behalf of a passive patient within a range of motion prescribed by the therapy.

In the case of knee surgery recovery, CPM helps the joint regain flexibility by mechanically moving the lower leg according to normal human-joint kinematics. This gentle movement promotes the flow of blood and nutrients to the injured area, helps to reduce long-term pain, and has been shown to reduce scar tissue.

Trials of CPM in medical research began with treating joints such as the knee, hip, shoulder and elbow. Salter and others showed CPM promotes the regeneration of damaged tissues, prevents joint stiffness, and promotes earlier use of the joint without the patient having to engage the joint actively during recovery.

Getting to know the CPM market

Medical practitioners are turning to CPM to treat burn victims, neurological damage caused by strokes, as well as brain and spinal cord damage. Athletes are using the technology to recover from less serious injuries, returning to peak performance quickly.

In its analysis of the market, Grand View Research found the market reached $650 million worldwide in 2018, and the market continued to grow at over 6% annually. Underlying trends point to long-term growth, particularly in the technology's original target market of post-surgical treatment.

The number of hip replacement surgeries across more than 30 Organisation for Economic Cooperation and Development (OECD)-member states increased 30% from 2007 to 2017. Knee surgeries saw 40% growth. The number of knee-replacement operations could grow sixfold in the U.S. between 2022 and 2030, according to some projections.

The growing number of joint replacements

Knee-replacement surgery growth trends shown from 2009 to 2020 for key OECD countries. (Source: OECD Health Statistics 2021)

In parallel, as electronics and embedded computing technology evolve, more personalized programs for rehabilitation are developing. CPM devices can use measurements and analysis in concert with input from physicians to support specific motion patterns suited to each user at each point in their treatment. Often, CPM is used in therapies that also involve manual physiotherapy and other treatments.

To avoid burdening hospital resources, the systems are increasingly being used in patient homes, particularly during the later stages of treatment. That trend drives the development of lower-cost, more portable designs.

Though the components are readily available and can be inexpensive, building a CPM device for physiotherapy is a complex project that requires a good understanding of electronics, software development, and the relevant medical standards and regulations.

Understanding the core design considerations

Safe operation is paramount in CPM design. The knee joint is a complex structure. Its combination of bone and muscle tissue allows for both translation and rotation. This can complicate the task of creating control algorithms that promote healing without causing injuries through undesired manipulation of the joint. However, safe control can often be assisted using mechanical structures that allow a limited range of manipulations, such as a single-axis joint placed under the knee.

A common electromechanical design for knee-oriented CPM systems is to couple a motor to a slider that changes the angle of a revolute joint through a lead screw. Often, designs use stepper motors as these support precise and gradual control of the mechanical joint. With advanced motor-control algorithms, it is possible to use a brushless motor in place of a stepper motor. This achieves finely tuned motion behaviors needed for CPM.

A typical knee-joint CPM

Graphic shows the structure of a typical knee-joint CPM system with leadscrew mechanism. (Source: Trochimczuk 2014)

Control over the stepper motor is often implemented by firmware running on a microcontroller (MCU) that interfaces to a suitable motor driver circuit. For a stepper motor, a suitable circuit would be based around an integrated dual full-bridge driver, a type of device produced by many power-semiconductor specialists.

In many cases, the control loop will use a proportional-integral-derivative (PID) algorithm to give stability. Typically, one or more position sensors will report how far the actuator has moved since the last movement command. A rotary encoder can often provide accurate data, though more sophisticated systems may include additional forms of distance or position sensors.

Using ultrasonic or infrared reflections can help gauge how the assembly is moving in space. Designers may also employ force sensors to check that the correct torque is applied. For example, if the patient is attempting to prevent the exoskeleton from moving, the correct action may be to halt the operation.

Under normal conditions, the PID algorithm will use the sensor inputs to adjust the torque and speed applied by the motor. Unexpected changes may be reflected in anomalous position readings. In more extreme cases, safety sensors may detect motion outside the operating range of the device. In these cases, the firmware will need to halt operation predictably and reliably to avoid injuring the patient.

Because of the need for reliably safe operation, the system may include hardware interlocks that override the MCU and suspend operation if the machine is found to be operating outside its safe operating zone.

A closed-loop control circuit for CPM

This block diagram shows how a CPM design comprises drive and control closely linked in a closed loop. (Source: Rattarojpan 2011)

Using sensor feedback to support AI in CPM

CPM devices use therapeutic programs that follow a set sequence of operations over weeks and assume a predictable rate of improvement. OEMs are now incorporating more real-time feedback into the control algorithms, not just from the motor-related sensors but from physiological sensors. This feedback can be combined with machine-learning techniques.

One of the earliest uses of machine learning in motion control lay in the application of iterative learning control (ILC) to PID control loops to tune and optimize motion profiles. The ILC process helps overcome non-linearities and other problems with basic PID control that can occur in electromechanical systems involving complex linkages between rotational and linear elements.

More recently, CPM designers have looked to add machine learning to support a range of needs. Some applications are offline, where data records from and between sessions are analyzed by a model that determines how well treatment is progressing and what settings should be used in the next session.

Advanced AI algorithms such as deep learning can help control real-time behavior, particularly when systems are used in the home and away from a skilled therapist. These systems can also help monitor sensor inputs for safety. For example, failure to attach the limb to the device correctly may cause abnormal motion that is difficult to detect using conventional sensor-analysis algorithms. It may also be trained, with appropriate sensor inputs, to pick up conditions such as excessive swelling that may result from over-exertion of the target muscles.

To detect conditions such as abnormal swelling or muscle activity, various sensors that the patient wears during treatments will be required in concert with the position-measurement sensors directly attached to the machine. Pressure sensors on a worn strap around the limb can help detect swelling, possibly in combination with temperature sensors.

Another potentially important sensor modality for detecting physiological changes during treatment is surface electromyography. Though the electrical signals from the muscles will typically be lower in intensity during passive-motion treatment cycles compared to active physiotherapy, the signals can provide useful data on how recovery is progressing around the affected area.

Making the interface to medical equipment

As well as the core electronics and software used to control the CPM-based therapies, the human-machine interface is just as important to the overall design. The interface needs to support the fine-tuning of parameters that healthcare professionals will want. At the same time, home-use systems will need to be easy to use without medical supervision.

For this reason, implementors have turned to graphical touchscreen interfaces because they provide a large amount of useful information in an accessible manner. The touchscreen also makes it easy to control different segments of the therapy and see potential issues during treatment.

Internet connectivity is an increasingly important element. New connected devices must now also support data security to comply with regulations such as the Health Insurance Portability and Accountability Act (HIPAA). As with all medical equipment for clinical use, it is vital that design teams consult with healthcare professionals and potentially seek legal advice before entering the market.

SEE AVNET SILICA'S HEALTHCARE SOLUTIONS

About Author

Avnet Staff

We use Avnet Staff as a collective byline when our team of editors and writers collaborate on the co...

View Bio