Physical Therapy AI Moonshot

“We choose to go to the moon in this decade and do the other things, not because they are easy, but because they are hard, because that goal will serve to organize and measure the best of our energies and skills, because that challenge is one that we are willing to accept, one we are unwilling to postpone, and one which we intend to win.”

John F. Kennedy

What’s a physical therapy moonshot?

It’s a big goal that seems impossible but only because no one has done it, yet.

For example, our Machine Learning Engineers tell us we need 500,000 data samples to validate the first mobile app in the world to predict Falls Risk with an AI algorithm.

500,000 data points!

Collecting that amount of data seems impossible.

But, only because no one has ever collected 500,000 samples of patients performing the Clinical Test of Sensory Integration and Balance (CTSIB) using a visual image obtained with Computer Vision.

But, if you listen to Kennedy’s speech (and you should if you’re a PT clinic leader) you’ll hear how he vividly describes, using metaphor and comparisons, and breaks the steps down to go to the Moon:

But if I were to say, my fellow citizens, that we shall send to the moon…

• 240,000 miles away… a giant rocket more than 300 feet tall
• the length of this football field
• made of new metal alloys, some of which have not yet been invented,
• capable of standing heat and stresses several times more than have ever been experienced,
• fitted together with a precision better than the finest watch,
• carrying all the equipment needed for propulsion, guidance, control, communications, food and survival,
• on an untried mission, to an unknown celestial body,
• then return it safely to Earth, re-entering the atmosphere at speeds of over 25,000 miles per hour
• causing heat about half that of the temperature of the sun

And do all this, and do it right, and do it first before this decade is out — then we must be bold. 

JFK

So, what are the steps to achieve the Physical Therapy AI Moonshot?

How will VisualPT.ai EVER get 500,000 images for the AI model to analyze?

Here are our steps – each subsequent steps proceeds from the one before it.

Data Acquisition Breakdown
30	second video
2	frames per second
60	frames per video
6	video test conditions per CTSIB
360	Total # images per Test Subject
500,000	Machine Learning Engineers estimate # of Images to achieve valid and reliable algorithmic assessments
1,389	Estimated number of Test Subjects needed
10	Requested # of Test Subjects per DPT/PTA Student enrolled in Data Collection Experience
139	# of DPT/PTA Students needed

---

Now using Google’s MediaPipe BlazePose GHUM 3D pose detection model to capture the human figure in the CTSIB test.

Technical Specifications