I've had a back and forth e-mail with the company for a while as they answered a few queries I had regarding Halo and what makes them stand out / differentiate. After collating all the information in the various e-mails, I've done a write up here, a lot of it is in their words from the e-mail and some of it gets quite technical and it's not my area of expertise so excuse me if I used any terms incorrectly.
Data Direction MYQ's scan is performed 100% on-device even when white labelled through their partner's product. The entire measurement capture process is handled on the device which results in an almost instantaneous processing time and zero cloud processing which means no data is sent out of the users phone back to MYQ or their partners. The way Amazon functions is that it captures the user data, sends it for processing on the Amazon system through their cloud and then returns it to the user. MYQ differs with an expanded measurement library, as well as its processing capabilities. Not only do the images from MYQ's scan not get sent out of the device, they are also deleted as soon as the measurements have been processed, and all of this happens in less than 1 second.
Data Capture MYQ has developed expansive medical image libraries using iDEXA (Dual-energy absoption measurement) or medical images on which it has developed and perfected its image capturing and analysis systems. All the analytics are based on the data that MYQ captures simultaneously with user-specific images in our global data collections.
MYQ's advantage comes from its collection of proprietary medical imagery libraries and from its rapid, patented capture of the participant's dimensions to enhance the MYQ process by training it with medical imagery. It is likely the world's largest iDEXA imaging library for measuring and analysing the human form in the world. The unique clinical gathering of data and subjects that MYQ has prepared over the past 3 years has a distinct advantage and has been developed to enhance the specific needs required for the measurement protocols of MYQ's technology. For example, MYQ's total body fat calculation is based on the unique data set that it has built globally across males and females of multiple ethnicities, shapes, and sizes. Each individual generates 12,000 body-location data outputs. This model enhances MYQ's capability of returning or measuring all global measurement standards, being ISO, ISAK, ASTM standardized measurements, or other required body shape and composition measurements. The data analysis model has novel machine learning modules that estimate body composition including body fat, bone, and muscle through the understanding of medical imagery. MYQ does not use simple, regression equations nor a regression network that predicts body fat in a manner similar to the doctor's skin-fold style predictions. MYQ has gathered its own unique data, that has been modeled and captured for the specific requirements of its imaging and dimensioning system to enhance the data for on-device protocols and delivery of accurate dimensions and body composition.
In contrast with Amazon's Halo, it seems they use regression and/or congenital techniques to estimate a person's body fat across a machine learning based model built from a pool of previously examined/learned people and shapes. This style of regression formula, equation, or even those done through machine learning is a statistical model that determines the relationship (If any even exists) between the predictor variable (input) and the outcome variable (output). In classical techniques, regression is normally abstracted or represented by a limited number of input parameters, then a regression is assumed to map these parameters to their corresponding outputs. These conventional approaches have limited flexibility and freedom and don't model the physical characteristics of the process accurately and assume the process is either deterministic or semi-deterministic. For example, one may use a person's height, weight, gender, age, neck, wrist, waist circumferences in a regression equation, or even a machine learning model to predict the total body fat. In this case, 7 parameters are assumed to be accurate enough to model the total body fat of all human bodies, which is completely inaccurate and incorrect even if we assume the 7 parameters are accurate.
With MYQ's technology, they do not abstract or represent the human body shape or physique with a handful or a limited number of parameters. They create an ecologically and physically valid model of the human shape where limited, distinctive, or salient features or unique descriptors are extracted from thousands of human representations with their proprietary dataset of 3D human scans and their corresponding images. Each individual in this scenario would have their own unique forensic image and shape features along with their corresponding outputs or labels. Modern artificial neural networks with millions of neurons are then designed, optimized, and trained using this big data to understand and model the relationships between these forensic images and the corresponding outputs. The results of this is no longer a basic regression model with limited degrees of freedom, but a hyperspace style neuron-driven regression and matching mechanism where the number of input parameters goes into the millions. This makes their patented technology statistically meaningful when it comes to a larger widespread population, robust, and achieves their state of the art accuracy and supersedes other competing technologies, even the ones that attempt to replicate MYQ's tech.
Standalone Use vs Specialised External GearMYQ's tech uses existing smartphone sensors and can be integrated with any external sensor such as a Fitbit or a wireless sensor. As such, no additional purchases are needed to make MYQ's software integrate-able since MYQ augments the partner's tech environment rather than disrupting it. MYQ supplies its partners with an SDK to embed the tech into their existing environment, so in this way users are not forced into MYQ's hardware environment and can stay in the partner's ecosystem and their user data/info controlled via his/her consent within the partner's application.
As for pricing, whether it is a subscription basis for repeated use or a single/episodic use, that is agreed individually with each partner, which then facilitates a predetermined pricing structure with MYQ under an agreement that covers payment amount, duration, minimum quantity of users etc. MYQ's service is volume driven and can cost as little as USD $1.50 for unlimited access.
On the other hand, Amazon Halo's offering requires users to come into the Amazon environment, obligates them to buy a sensing band, and requires a further subscription for the measured dimension and movement data to be processed inside the Amazon ecosystem. The Halo Band will cost approximately $99.99 USD and the service that unlocks their more advanced features will cost $3.99 per month. Amazon also intends for the user to leave the Halo Band on at all times.
Furthermore, the newly completed CompleScan application by MYQ will also supply its users with the following information through their smartphone in one session:
Blood pressure
Stress levels
Heart rate
Heart rate variability
Irregular heartbeat
Respiratory rates
Emotion
Total body fat
Waist to height ratio
Waist to hip ratio
Cardiovascular disease risk
Type 2 diabetes risk
Stroke risk
This is all analysed through MYQ's partner app, and it is a non invasive solution that delivers complete privacy to the consumer.
The resulting captures and data allows an individual the opportunity to understand their personal health and exposure to chronic disease risks that form up to 70% of deaths globally every year (Not including the current pandemic, although the app is able to detect Covid-19 based symptoms as well).
MYQ takes the position that Amazon Halo entering the market is not a negative for them, and in fact it is a positive as it stimulates the industry and further validates that MYQ has been on the correct path for the last 7 years. At the end of the day, there are many fits that suit both organisations and the difference processes and different measurement capabilities.
I feel pretty confident that as MYQ says, Amazon's Halo is great for the industry, but it is actually a more expensive offering that offers less data with less accuracy than MYQ's tech. There's plenty of the market to share between the 2, and any other tech giants that haven't gotten into the market yet will likely approach MYQ to partner up or to takeover completely if they want exclusivity of MYQ's tech for themselves. Definitely a game changer in many fields and verticals, we just need MYQ to deliver which with the recent go-lives and go-lives to come it seems they will do without a problem. Question of when not if!
MYQ Price at posting:
$1.05 Sentiment: Buy Disclosure: Held
(20min delay)