Review by Richard K. Bogan, MD, FCCP
Chairman and Chief Medical Officer
INTRODUCTION TO AUTOMATED PROCESSING OF THE BIOLOGICAL SIGNAL
In sleep medicine we monitor multiple signals. It is important for clinicians to understand the state of the individual, monitor what stage of sleep they are in, whether they are awake or not, and to also understand processes. The American Academy of Sleep Medicine (AASM) has recently revisited this because there have been many changes in the methodology for scoring, and in this specific case, human visual scoring. As humans, we are accustomed to seeing science advance and move towards more automated processing and I think that Morpheus® makes a great stride towards automated processing and scoring.
Morpheus, created by WideMed, Ltd., is the signal processing that monitors biological signals and it allows for an automated analysis of the biological signal. It also has the capability of allowing human supervision so that a human can look at the biological signal and critique studies performed by Morpheus. It provides the best of both approaches. It can incorporate up to 100% human supervision depending on the quality of the signal and the quality of the signal processing.
morpheus is an evolving software specializing in advanced signal processing that enhances the resolution. You can compare Morpheus to the amount of pixels there are in a digital photograph. The way I look at Morpheus is that there are a lot of pixels and therefore the resolution is very high in terms of looking at the biological signal, and also attached are the benefits of having R&K as well. R&K is basically a human analysis of biological signals that is based on the visual recognition of frequency and amplitude characteristics. Morpheus utilizes mathematics models and algorithms to provide a consistent automated scoring.
SleepMed is a diagnostic and therapeutic sleep company that currently operates approximately 150 sleep labs and will process over 70,000 sleep studies this year. WideMed created Morpheus to assist sleep centers in managing, streamlining and enhancing the data collection and analysis process.
WideMed and SleepMed have worked closely together for a number of months with the goal of providing a method of automated signal processing with human supervision. The new method SleepMed employs using Morpheus streamlines the process much more rapidly allowing more studies to be analyzed in the same period of time. A database is also available in Morpheus so that certain fields of information could be categorized, quantified and searched for future studies in order to understand biological processes and the state of individual.
WideMed’s engineers have expertise in mathematical models of signal processing. WideMed developed a multi-dimensional mathematical analysis of biological signals and by using this multi-dimensional mathematical model they are using robust engineering methodologies such as adaptive segmentation, fuzzy logic and hidden Markov models.
The artificial intelligence methodologies used include fuzzy logic algorithms that characterize the adaptive segments into different biological processes. So here, rather than the human looking at a biological signal we have a mathematical model that allows us to perform signal processing. And because of this signal processing and the mathematical model, we can compartmentalize the biological signal into segments, which is known as “adaptive segmentation”.
WideMed’s engineering skills and SleepMed’s clinical expertise developed algorithms to help interpret the signal processing. So again, you have multi-dimensional mathematical analyses of biological signals that are applied clinically, and by using various algorithms, what Morpheus does is it takes these segments and compartmentalizes them into 30-second chunks that are translated into the traditional R&K. It is like taking a foreign language and translating it into a language that we now know as R&K language.
MORPHEUS’ MULTI-DIMENSIONAL MATHEMATICAL MODEL
Sleep systems with automated scoring typically will use a mathematical model called Fast Fourier Transformation (FFT). Morpheus uses a different mathematical model that is all adaptive segmentation. Morpheus goes back and forth through the signal and identifies areas in the signal that have mathematical properties that are similar throughout the signal tracing. It will identify high frequency areas and low frequency areas, much as a human eye does.
So rather than just giving you a spectral analysis of the frequency and the amplitude characteristics and using a fake denominator, which is what FFT does, Morpheus uses this multi-dimensional mathematical model of adaptive segments so that it replicates what the human does in terms of looking at frequency and amplitude characteristics. At the same time, Morpheus will monitor the EMG of the electromonogram and eye movements, calculate the probability that there are changes in EMG activity and rapid eye movements (REM), and then correlate that data to determine the current state of the individual in terms of wake or sleep.
The benefits that we gain through Morpheus are understanding processes about how the brain organizes sleep, controls the breathing, controls the legs and the periodic leg movements, controls the heart, and having the ability to look at the individual’s sleep state. We can translate all of that into the traditional human scoring where we have the automated processing as well. Ultimately, we are able to study sleep states and processes from a different language and a different prospective from a mathematical scientific methodology using human R&K scoring because it is human-supervised.