# US 12,690,995 B2 · Anterior cruciate ligament protector

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 12,690,995 B2
- Document ID: US12690995B2
- Status: Granted
- Kind: B2
- Application number: 19/092,886
- Filed: 2025-03-27
- Published: 2025-10-02
- Granted: 2026-07-28
- Earliest claimed: 2024-03-27
- Assignees: Princeton Satellite Systems, Inc.
- Applicants: Princeton Satellite Systems, Inc.
- Inventors: Michael Paluszek; Joyce Mo
- Web page: https://patentgrep.com/patent/US12690995B2

## CPC

- A61F 5/0123: {for the knees (A61F5/0106 takes precedence)}
- A61B 5/1123: {Discriminating type of movement, e.g. walking or running (A61B5/1116, A61B5/112 take precedence)}
- A61B 5/389: Electromyography [EMG]
- A61B 5/4585: {Evaluating the knee}
- A61B 5/486: {Biofeedback (using electroencephalography [EEG] A61B5/375)}
- A61B 5/6828: {Leg}
- A61B 5/7264: {Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems}
- A61B 5/7267: {involving training the classification device}
- A61B 5/7275: {Determining trends in physiological measurement data; Predicting development of a medical condition based on physiological measurements, e.g. determining a risk factor}
- A61B 5/742: {using visual displays (displays for heart-related electrical signals, e.g. ECG, A61B5/339)}
- A61B 2505/09: Rehabilitation or training
- A61B 2560/0238: Means for recording calibration data
- A61B 2562/0219: Inertial sensors, e.g. accelerometers, gyroscopes, tilt switches
- A61F 2005/0132: {Additional features of the articulation}
- A61F: FILTERS IMPLANTABLE INTO BLOOD VESSELS; PROSTHESES; DEVICES PROVIDING PATENCY TO, OR PREVENTING COLLAPSING OF, TUBULAR STRUCTURES OF THE BODY, e.g. STENTS; ORTHOPAEDIC, NURSING OR CONTRACEPTIVE DEVICES; FOMENTATION; TREATMENT OR PROTECTION OF EYES OR EARS; BANDAGES, DRESSINGS OR ABSORBENT PADS; FIRST-AID KITS (dental prosthetics A61C)
- A61B: DIAGNOSIS; SURGERY; IDENTIFICATION (usefulness limited to only animals A61D)

## Abstract

A wearable knee brace system integrates inertial measurement unit (IMU) and electromyography (EMG) sensors with machine learning models to prevent anterior cruciate ligament (ACL) and posterior cruciate ligament (PCL) injuries. The system collects real-time biomechanical and neuromuscular data and analyzes the data using supervised, personalized, or federated learning techniques to identify high-risk movement patterns. Upon detecting elevated injury risk, the system may issue real-time alerts or activate a hybrid actuation system comprising high-force, low-displacement actuators and low-force, high-displacement actuators to reduce joint loading. The system further supports personalized model adaptation using calibration activities and transfer learning, as well as privacy-preserving performance improvements through federated learning. Feedback is provided through visual, auditory, or haptic interfaces and may be integrated with rehabilitation tools or mobile applications. The system may be used in athletic, dance, clinical, or rehabilitative environments to enhance performance, optimize recovery, and reduce the risk of knee ligament injuries.

## Claims

### Claim 1

A method for preventing anterior cruciate ligament injury in a user wearing a knee brace, comprising:
receiving motion data from at least one inertial measurement unit (IMU) sensor and neuromuscular activity data from at least one electromyography (EMG) sensor positioned on the user's leg;
analyzing the motion data and neuromuscular activity data using a machine learning model trained to identify movement patterns associated with increased risk of ligament injury;
predicting, based on the analysis, an imminent high-risk movement likely to cause excessive force on the anterior cruciate ligament;
generating a feedback signal in response to the prediction; and
transmitting the feedback signal to at least one of a visual, auditory, or haptic feedback device to alert the user to modify their movement.

### Claim 2 (depends on 1)

The method of claim 1, wherein the machine learning model comprises a convolutional neural network (CNN), a recurrent neural network (RNN), or a long short-term memory (LSTM) network.

### Claim 3 (depends on 2)

The method of claim 2, wherein the prediction is generated using a model trained on labeled datasets comprising time-series IMU and EMG data collected during movement activities such as walking, jumping, squatting, or pivoting.

### Claim 4 (depends on 1)

The method of claim 1, wherein the EMG sensor is positioned over the user's hamstring muscles and the prediction is based at least in part on reduced hamstring activation during high-impact motion.

### Claim 5 (depends on 1)

The method of claim 1, further comprising dynamically adjusting the feedback signal based on user response history or rehabilitation progress.

### Claim 6 (depends on 1)

The method of claim 1, wherein the feedback signal is delivered via a mobile application that displays a real-time biomechanical risk score.

### Claim 7 (depends on 1)

The method of claim 1, wherein the IMU comprises a tri-axial accelerometer and gyroscope configured to compute jerk, linear acceleration, and angular velocity of the user's knee joint.

### Claim 8 (depends on 7)

The method of claim 7, wherein the IMU is positioned on both a femoral segment and a tibial segment of the knee brace to capture relative joint movement.

### Claim 9 (depends on 1)

The method of claim 1, wherein the steps are implemented using instructions stored on a non-transitory computer-readable medium and executed by at least one processor integrated into the knee brace system.

### Claim 10

A method for dynamically reducing knee joint loading in a wearable brace, comprising:
detecting, using at least one IMU sensor on a knee brace, relative motion between a femoral attachment point and a tibial attachment point;
detecting, using at least one EMG sensor, muscle activation levels in at least one of the quadriceps, hamstring, or gastrocnemius muscles;
comparing the detected motion and muscle activation to one or more predetermined thresholds indicative of excessive force or torque on the anterior cruciate ligament;
generating a control signal in response to detecting a value that exceeds the threshold; and
activating one or more actuators coupled to the knee brace to apply tension to at least one strap or artificial muscle, thereby offloading force from the knee joint.

### Claim 11 (depends on 10)

The method of claim 10, wherein the one or more actuators comprise a piezoelectric actuator configured to apply high-force, low-displacement tension.

### Claim 12 (depends on 11)

The method of claim 11, wherein the one or more actuators further comprise a pneumatic actuator configured to apply low-force, high-displacement movement correction.

### Claim 13 (depends on 10)

The method of claim 10, wherein the control signal is generated based on detecting anterior tibial translation exceeding a defined threshold in conjunction with insufficient muscle activation.

### Claim 14 (depends on 10)

The method of claim 10, further comprising deactivating the actuator after a predetermined duration or upon resolution of the high-risk condition.

### Claim 15 (depends on 10)

The method of claim 10, wherein the prediction and actuator activation are performed in real time with a latency of less than 100 milliseconds.

### Claim 16

A method for dynamically reducing knee joint loading in a wearable brace, comprising: detecting, using at least one IMU sensor on a knee brace, relative motion between a femoral attachment point and a tibial attachment point; detecting, using at least one EMG sensor, muscle activation levels in at least one of quadriceps, hamstring, or gastrocnemius muscles; comparing the detected motion and muscle activation to one or more predetermined thresholds indicative of excessive force or torque on the anterior cruciate ligament; generating a control signal in response to detecting a value that exceeds the threshold; and activating one or more actuators coupled to the knee brace to apply tension to at least one strap or artificial muscle, thereby offloading force from the knee joint.

### Claim 17 (depends on 16)

The method of claim 16, wherein the movement recommendation includes a corrective exercise to improve joint stability or muscle coordination.

### Claim 18 (depends on 17)

The method of claim 17, wherein the graphical user interface displays a trend analysis of the user's joint angles or muscle activation patterns over time.

### Claim 19 (depends on 16)

The method of claim 16, wherein the model adapts using a transfer learning process that fine-tunes model parameters based on calibration movements performed by the user.

### Claim 20 (depends on 19)

The method of claim 19, wherein the machine learning model is further refined using a federated learning framework that aggregates encrypted model updates from multiple devices without sharing raw data.