# US 2022/0044820 A1 · Methods and apparatus for injury prediction based on machine learning techniques

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

- Number: US 2022/0044820 A1
- Document ID: US20220044820A1
- Status: Published application
- Kind: A1
- Application number: 17/395,152
- Filed: 2021-08-05
- Published: 2022-02-10
- Earliest claimed: 2020-08-05
- Assignees: StrongArm Technologies, Inc.
- Applicants: StrongArm Technologies, Inc.
- Inventors: Michael Kim; SivaSankara Reddy Bommireddy; Michael Spinelli; Sean Petterson
- Web page: https://patentgrep.com/patent/US20220044820A1

## CPC

- G16H 50/30: for calculating health indices; for individual health risk assessment
- A61B 5/1118: {Determining activity level}
- A61B 5/1121: {Determining geometric values, e.g. centre of rotation or angular range of movement}
- A61B 5/1126: {using a particular sensing technique}
- A61B 5/486: {Biofeedback (using electroencephalography [EEG] A61B5/375)}
- 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}
- G06F 18/24: Classification techniques
- G06F 18/2433: Single-class perspective, e.g. one-against-all classification; Novelty detection; Outlier detection
- G06V 40/20: Movements or behaviour, e.g. gesture recognition (recognition of facial expressions G06V40/16)
- G16H 50/20: for computer-aided diagnosis, e.g. based on medical expert systems
- A61B 2562/0219: Inertial sensors, e.g. accelerometers, gyroscopes, tilt switches
- G16H 10/00: ICT specially adapted for the handling or processing of patient-related medical or healthcare data (for medical reports G16H15/00; for therapies or health-improving plans G16H20/00; for the handling or processing of medical images G16H30/00)
- G16H: HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- A61B: DIAGNOSIS; SURGERY; IDENTIFICATION (usefulness limited to only animals A61D)
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING

## Abstract

Systems and methods of the present disclosure enable injury prediction using one or more processors for receiving a time-varying signal of sensor measurements from a sensor device associated with a user. The processor(s) generate time windows of the time-varying signal, including a series of the sensor measurements across a predetermined time period, and generate motion features based at least in part on the series of the sensor measurements of the time windows. The processor(s) utilize an injury risk classification machine learning model to predict an injury risk during each time window based at least in part on the motion features. An injury alert message is generated based at least in part on the injury risk being predicted; and transmitting the injury alert message to at least one user computing device.

## Claims

### Claim 1

A method comprising:
receiving, by at least one processor, a time-varying signal of sensor measurements from at least one sensor device associated with at least one user;
generating, by the at least one processor, at least one time window of the time-varying signal, wherein the at least one time window comprises a series of the sensor measurements across a predetermined time period;
generating, by the at least one processor, at least one motion feature based at least in part on the series of the sensor measurements of the at least one time window;
utilizing, by the at least one processor, an injury risk classification machine learning model to predict an injury risk during the at least one time window based at least in part on the at least one motion feature;
generating, by the at least one processor, an injury alert message based at least in part on the injury risk being predicted; and
transmitting, by the at least one processor, the injury alert message to at least one user computing device.

### Claim 2 (depends on 1)

The method as recited in claim 1, wherein the at least one sensor device comprises at least one accelerometer configured to detect the sensor measurements comprising yaw, pitch, roll, acceleration and gyroscopic measurements.

### Claim 3 (depends on 1)

The method as recited in claim 1, wherein the injury risk classification machine learning model comprises at least one novelty detection algorithm trained to detect anomalous motion features of the time-varying signal.

### Claim 4 (depends on 3)

The method as recited in claim 3, further comprising:
receiving, by the at least one processor, at least one historical motion feature for at least one historical time window;
receiving, by the at least one processor, at least one non-injury event label associated with the at least one historical motion features; and
training, by the at least one processor, the novelty detection algorithm with the at least one historical motion features and the at least one non-injury event label in order to predict whether a non-injury event is in the time-varying signal.

### Claim 5 (depends on 4)

The method as recited in claim 4, further comprising:
determining, by the at least one processor, the injury risk based at least in part on the novelty detection algorithm not predicting the non-injury event in the at least one time window.

### Claim 6 (depends on 3)

The method as recited in claim 3, further comprising:
receiving, by the at least one processor, at least one first historical motion feature for at least one historical time window;
receiving, by the at least one processor, at least one non-injury event label associated with the at least one first historical motion features;
receiving, by the at least one processor, at least one second historical motion feature for at least one historical time window;
receiving, by the at least one processor, at least one injury event label associated with the at least one second historical motion features; and
training, by the at least one processor, the novelty detection algorithm with the at least one first historical motion features, the at least one non-injury event label, the at least one second historical motion features, and the at least one injury event label in order to predict a non-injury event, an injury event, or both in the time-varying signal.

### Claim 7 (depends on 1)

The method as recited in claim 1, wherein the injury risk classification machine learning model comprises at least one of:
density based machine learning models,
tree-based machine learning models, and
variance based machine learning models.

### Claim 8 (depends on 1)

The method as recited in claim 1, wherein the at least one motion feature comprises at least one of:
sagittal angle,
twist angle,
lateral angle,
sagittal velocity,
twist velocity, and
lateral velocity.

### Claim 9

A system comprising:
at least one processor; and
a non-transitory memory storing instructions which, when executed by the at least one processor, causes the at least one processor to:
receive a time-varying signal of sensor measurements from at least one sensor device associated with at least one user;
generate at least one time window of the time-varying signal, wherein the at least one time window comprises a series of the sensor measurements across a predetermined time period;
generate at least one motion feature based at least in part on the series of the sensor measurements of the at least one time window;
utilize an injury risk classification machine learning model to predict an injury risk during the at least one time window based at least in part on the at least one motion feature;
generate an injury alert message based at least in part on the injury risk being predicted; and
transmit the injury alert message to at least one user computing device.

### Claim 10 (depends on 9)

The system as recited in claim 9, wherein the at least one sensor device comprises at least one accelerometer configured to detect the sensor measurements comprising yaw, pitch, roll, acceleration and gyroscopic measurements.

### Claim 11 (depends on 9)

The system as recited in claim 9, wherein the injury risk classification machine learning model comprises at least one novelty detection algorithm trained to detect anomalous motion features of the time-varying signal.

### Claim 12 (depends on 11)

The system as recited in claim 11, wherein the instructions which, when executed by the at least one processor, further cause the at least one processor to:
receive at least one historical motion feature for at least one historical time window;
receive at least one non-injury event label associated with the at least one historical motion features; and
train the novelty detection algorithm with the at least one historical motion features and the at least one non-injury event label in order to predict whether a non-injury event is in the time-varying signal.

### Claim 13 (depends on 12)

The system as recited in claim 12, wherein the instructions which, when executed by the at least one processor, further cause the at least one processor to:
determine the injury risk based at least in part on the novelty detection algorithm not predicting the non-injury event in the at least one time window.

### Claim 14 (depends on 11)

The system as recited in claim 11, wherein the instructions which, when executed by the at least one processor, further cause the at least one processor to:
receive at least one first historical motion feature for at least one historical time window;
receive at least one non-injury event label associated with the at least one first historical motion features;
receive at least one second historical motion feature for at least one historical time window;
receive at least one injury event label associated with the at least one second historical motion features; and
train the novelty detection algorithm with the at least one first historical motion features, the at least one non-injury event label, the at least one second historical motion features, and the at least one injury event label in order to predict a non-injury event, an injury event, or both in the time-varying signal.

### Claim 15 (depends on 9)

The system as recited in claim 9, wherein the injury risk classification machine learning model comprises at least one of:
density based machine learning models,
tree-based machine learning models, and
variance based machine learning models.

### Claim 16 (depends on 9)

The system as recited in claim 9, wherein the at least one motion feature comprises at least one of:
sagittal angle,
twist angle,
lateral angle,
sagittal velocity,
twist velocity, and
lateral velocity.

### Claim 17

A non-transitory computer readable medium storing instructions which, when executed by at least one processor, are configured to cause the at least one processor to perform steps comprising:
receiving a time-varying signal of sensor measurements from at least one sensor device associated with at least one user;
generating at least one time window of the time-varying signal, wherein the at least one time window comprises a series of the sensor measurements across a predetermined time period;
generating at least one motion feature based at least in part on the series of the sensor measurements of the at least one time window;
utilizing an injury risk classification machine learning model to predict an injury risk during the at least one time window based at least in part on the at least one motion feature;
generating an injury alert message based at least in part on the injury risk being predicted; and
transmitting the injury alert message to at least one user computing device.

### Claim 18 (depends on 17)

The non-transitory computer readable medium as recited in claim 17, wherein the injury risk classification machine learning model comprises at least one novelty detection algorithm trained to detect anomalous motion features of the time-varying signal.

### Claim 19 (depends on 18)

The non-transitory computer readable medium as recited in claim 18, further comprising instructions which, when executed by the at least one processor, are further configured to cause the at least one processor to perform steps comprising:
receiving at least one historical motion feature for at least one historical time window;
receiving at least one non-injury event label associated with the at least one historical motion features; and
training the novelty detection algorithm with the at least one historical motion features and the at least one non-injury event label in order to predict whether a non-injury event is in the time-varying signal.

### Claim 20 (depends on 19)

The non-transitory computer readable medium as recited in claim 19, further comprising instructions which, when executed by the at least one processor, are further configured to cause the at least one processor to perform steps comprising:
determining the injury risk based at least in part on the novelty detection algorithm not predicting the non-injury event in the at least one time window.
