Each year, 50% of all Americans over the age of 18 will require physical therapy (PT). Yet, only about 10% of those people will take advantage of physical therapy services! This results in millions of Americans living with preventable issues that reduce their quality of life and increase the cost of healthcare.
Arm Innovator, Laura Kassovic is the CEO and Co-Founder of MbientLab, an organization that uses sensor technology for biomechanics, industrial control, research and product development. Laura has recently developed a new product, MIO, which aims to make physical therapy more affordable and accessible for all Americans. We asked Laura about the development behind the new product and here’s what she had to say:
MIO Therapy is a virtual assistant for physical and occupational therapists. MIO is based on our proven wearable motion sensor technology and when accompanied with an app, the sensors calculate the range-of-motion of any major body joint with great accuracy. The Mbient team worked with Stanford Healthcare, UCSF Medical Center, and Good Samaritan Hospital to build MIO Therapy.
The MIO App is for clinicians, it records patient data during PT sessions to report progress
A patient recovering from a wrist fracture wears the MIO sensors during a visit to a PT clinic. The sensors, combined with the MIO App, offer real-time 3D visualization of the wrist during the session.
Wrist movements and range-of-motion
Flexion, extension, and deviation of the wrist (in degrees), is recorded in the app for performance analysis and treatment optimization by the clinician. The MIO app also offers mini games and targeted exercises to entice patients to complete treatment and enhance the therapeutic experience.
Patient performing wrist exercises using the MIO App
There are two reasons for building MIO:
The MIO App and MIO dashboard allow researchers and clinicians to record and publish scientific data
MIO Therapy is powered by MbientLab sensors, specifically the MetaMotionR board. The board is equipped with the Nordic nRF52, which boasts an Arm Cortex-M4 processor with a Bluetooth transceiver. This allows the sensors to simultaneously and wirelessly communicate with the MIO app. The Cortex-M4 is ideal for sensor technology because it combines high-efficiency signal processing functionality.
The MetaMotionR also includes the Bosch BMI160 and Bosch BMM150 sensors, a 9-axis IMU for 3D motion sensing. The firmware that runs on the Cortex-M4 CPU is compiled with Keil and includes a custom scheduler, the Nordic Bluetooth 4.2 stack, Bosch sensor fusion, and a IIR butterworth filter using Arm DSP extensions for Cortex-M. The addition of DSP extensions to the Thumb instruction set provides the opportunity to perform signal processing operations directly on the Cortex-M4 processor.
The Cortex-M4 processor runs the Bosch Sensortec FusionLib, which combines the measurements from a 3-axis gyroscope, 3-axis geomagnetic sensor and a 3-axis accelerometer, to provide an absolute orientation vector. The FusionLib includes algorithms for offset calibration of each sensor, monitoring of the calibration status, and a Kalman filter to provide distortion-free and refined orientation vectors.
The MIO App calibrates the sensors and displays wrist motion in real-time and 3D
For our application, the magnetometer is left off in order to reduce the possibility of interference with hospital equipment and to improve battery life. The output of the FusionLib without the magnetometer produces relative orientation. A one-time calibration sequence performed by the clinician before a PT session ensures the quaternion outputs of the sensors are aligned in absolute space (i.e. if they are both in the same physical orientation they produce identical outputs). For joints with more variable geometry, such as ankles and foreheads, MIO patients are required to hold a specific position while a baseline is calculated. This allows us to offset for body variability and achieve consistent results.
The live, wirelessly streamed, quaternion data is then used to animate a 3D model in the app. For each major body joint such as the wrist, two sensors are used. One sensor is placed below the joint; the second is placed above the joint. Both sensors simultaneously stream quaternion vectors to the MIO app, which performs additional calculations. The quaternion value of the lower sensor is inverted and multiplied by the upper value to produce a final quaternion representing the rotation of the joint. From here we use various Euler conversions to calculate range of motion in degrees. As a result, the MIO App can calculate the ulnar deviation of the wrist of our patient within +-1 degree of accuracy.
Patient wearing the MIO sensors, one placed on the wrist and another on the hand
We can optionally add more sensors to track additional limb movement and achieve more visibility.
As we embark on this journey, I invite you to think back to a time when physical therapists were a group of women called “reconstruction aides” working to treat the wounded on the battlefields of World War 1. We have come a long way; the physical therapy profession requires a Master of Science (doctoral) degree and therapists are licensed to treat everything from vertigo to tendonitis.
MbientLab have just launched MIO and invite everyone to join in as they take physical therapy into the digital age. To share this story with your friends, family, and colleagues, and to learn more about MIO, visit the website.
To hear more from Laura and the Arm Innovator Program, subscribe below.
Sign up here