top of page

YRT Limited.

Information Provided by Pharos
Singly or in Any Combination

Conventional Sleep Scoring Software  |  Michele Sleep Scoring (MSS)

MSS was introduced in 2013. It measures the following conventional metrics in full-length PSGs:
 

  1. Sleep onset latency, REM latency, Total recording time, Total sleep time, Sleep Efficiency

  2. Times in different sleep stages,

  3. Arousal and awakening index, Number of respiratory-related, PLM-related, and spontaneous arousals, organized per sleep stage and body position. 

  4. Apnea hypopnea index using the recommended and alternate hypopnea criteria,

  5. Oxygen saturation values including average SpO2 and number of desaturations >2%, >3%and >4%.

  6. Body position.

Features

  • Scoring is based on signals from the right and left central electrodes OR right and left frontal electrodes.

  • Reports are provided in graphical and numerical forms.

  • Reports can be customized to user preferences.

  • Scoring is done remotely, but can be reviewed and edited on a custom viewer we provide

  • Editing is facilitated by: 

    • A window that displays ORP and the sum of delta wave durations in each epoch viewed during editing. This information helps distinguish stage wake from sleep and stage N2 from N3 in equivocal epochs using objective criteria 

    • A proprietary software analysis that suggests areas that need specific attention from the editor

  • Rapid turnover

  • Manual editing by certified RPSGTs can be provided

  • Reports can include graphical and numerical results of ORP and its derivatives. MSS is the only sleep scoring system that can provide these values

Advantages Relative to AI-based systems:

MSS’s algorithms were developed to faithfully implement the features indicated by the American Academy of Sleep Medicine (AASM)  that define each stage and each respiratory event. A rigorous, iterative process was followed to ensure that the final scoring results represent a compromise between the scoring of different technologists, from various centers. in equivocal epochs that could legitimately be scored in more than one state. .

We believe that the rigorous process used in MSS development ensures that only electrophysiological features with well established association with each sleep stage are used to identify the stage, that epochs with erroneous manual scoring in the development PSGs are excluded, that the results will compare favorably with manual scoring, and that allowance is made for legitimate differences between scorers when scoring equivocal epochs. That MSS staging results accomplished these goals was confirmed in a study on 70 PSGs not used in development and unknown to the MSS personnel that were scored by MSS in a totally arms-length way and the results were compared with those of manual scoring by 10 RPSGTs from 5  academic institutions (Harvard, UPenn, Stanford, Wisconsin and St. Louis), who were unfamiliar with MSS  

In this study, agreement (intraclass correlation) between MSS and the average of the 10 scorers was well within the range seen between two scorers in the same institution and/or between the two scorers of each institution and the scorers of the other 4 institutions in virtually all scoring categories. Furthermore, the average ICC for MSS vs. the average of 10 scorers was superior to the average ICC within or between institutions in time in stage N2, NREM arousals, and the number of obstructive and central apneas.
 

It would be of interest to determine the performance of AI-based scoring systems using the same rigorous, arms-length validation approach and to compare the results with MSS results when scoring the same PSGs under the same conditions.

Novel EEG Sleep Microstructure Metrics:

What are they and what do they measure?

Odds Ratio Product (ORP) and its Derivatives:

ORP is a measure of sleep depth, which is, in turn, a major determinant of how arousable a person is. There are several cycles of sleep during a typical night. Each cycle begins with an EEG pattern characteristic of wakefulness or light sleep (predominant high-frequency (alpha/beta) activity) with little activity in the slower frequencies (delta/theta). The pattern then gradually progresses into those with less activity in the high frequencies and more activity in the slow frequencies. At the end of the non-REM cycle (≈90 minutes), the EEG pattern is typically one characteristic of deep sleep, with predominant activity in slow frequencies and little high-frequency activity. Following a period of REM sleep, the cycle then repeats. In addition to these slow sleep cycles, the EEG pattern often suddenly and briefly changes to one of light or deep sleep to a pattern of wakefulness in association with arousals and awakenings. 

​

These decades-old observations have two important implications: a) Sleep depth is constantly changing during the night; there is no individual-specific sleep depth. b) Changes in sleep depth are characterized by reciprocal changes in high and low-frequency activities. In some subjects, the changes in high-frequency activity as sleep deepens are small relative to the changes in slow activity, while in other subjects, the opposite is the case. It follows that, for proper assessment of sleep depth, changes in the activity across the entire frequency range must be taken into account.

​

ORP is distinguished from other methods of measuring sleep depth in two ways:
 

  1. It measures sleep depth every 3 seconds. This allows assessment of dynamic changes in sleep depth (see ORP-9, below), as well as wake intrusions that do not meet the conventional criteria of arousals.
     

  2. Sleep depth is based on the changes in activity across the entire relevant EEG frequency range and not on just one frequency range (cf. power in the beta or delta frequency range).

Having a sleep depth value every 3 seconds allows calculation of sleep depth over any number of ranges of interest (different sleep stages, different times, different body positions, when events are present or absent. etc.), as well as the dynamics of changes in sleep depth, for example how fast does deep sleep return following arousals (See ORP-9 below) or how long and to what extent does a disturbance  (for example noise) affect sleep depth (two publications by Smith and one publication by Basner). The new observations that can be made using ORP, along with new non-ORP EEG metrics to be described in section 3, open new avenues of research into the pathogenesis and management of a wide range of clinical disorders. Examples of these applications can be found in the references cited throughout this website.

For comprehensive reviews, see (Younes, The case; Younes, Frontiers; Atlas; Younes Clin Sleep Med).

​

The ORP derivatives listed below are currently available. For more information about what they measure, their theoretical bases, and existing publications utilizing the specific derivation, if any, please click on the Read More icon.  This information serves as a guide into how the derivative may be useful in research on pathogenesis/management of relevant clinical disorders, or for basic studies on sleep physiology. 

​

ORP During Stage Wake (ORPW):

This metric helps distinguish excessive wake time due to low sleep pressure (e.g., hyperarousal) from wake time interrupted by repeated attempts to fall asleep.

ORP During Non-Rem Sleep (ORPNR):

This metric summarizes the highly variable overnight ORPNR values into one lumped value. ORPNR is most useful in identifying associations between sleep depth in cohorts with various clinical phenotypes and physiological conditions. Due to the very wide range of ORPNR in all cohorts, interpretation in single individuals requires consideration of associated sleep metrics.

ORP in Different Non-REM Stages (ORPN1, ORPN2, ORPN3):

Although these are available, they are of no scientific or clinical value. The two approaches to quantifying sleep depth use completely different markers of sleep depth and, while cohort averages of ORP decline as stage progresses from N1 to N3,  the two approaches produce highly uncorrelated results in individual subjects. Reporting these values in individual PSGs only serves to expose substantial differences between the two approaches, which can be confusing. .

ORP in REM Sleep (ORPREM):

ORPREM represents the first and only method available to objectively evaluate sleep depth in REM sleep. ORP levels in REM sleep range from values that are in the awake range (>2.00) to values that are in the range of deep non-REM sleep (<0.50). High ORPREM is associated with reduced REM time and REM sleep fragmentation, and this has possible implications to dream and psychiatric disorders.

Speed of Progression to Deep Sleep Following Arousals (ORP-9):

ORP-9 is the mechanism by which the brain regulates non-REM sleep depth. High values (high ORP nine seconds after the end of arousals) result in high ORP (light sleep) throughout non-REM sleep. ORP-9 is useful in identifying the operation of central mechanisms in the pathogenesis of shallow sleep in general and in OSA in particular.  

Cumulative Sleep Index (CSI):

CSI is the cumulative reduction in ORP from the level associated with full wakefulness (2.50) across the total study. It is intended to provide a quantitative metric of the overall “wakefulness reduction” achieved during each study. Its main use is in the investigation of mechanism of excessive sleepiness and habitual short sleepers.

Wake Intrusion Index (WII):

WII is the number of 3-second epochs with  ORP > 2.00 (the wake level) per hour of sleep. It includes 3-second epochs during conventional arousals but also instances when the increase in ORP is too brief to qualify as arousal. WII far exceeds the conventional arousal index with the ratio reaching 20 in some subjects. Early evidence suggests that it is a more sensitive indicator of sleep fragmentation. WII may also be relevant to clearance of brain waste during sleep via the glymphatic system.

ORP Architecture: 

A breakdown of the ORP into deciles of .25 increments, showing the amount of Total Recording Time spent in each of the deciles. Presence of deepest sleep (deciles 1 and 2) compared to full wakefulness (decile 10) reveals distinct architectures that have been associated with various health factors in the Sleep Heart Health Study (SHHS). One key element of the ORP Architecture is the separation of stage wake into full wakefulness vs. drowsy wakefulness. Full wakefulness is more representative of low sleep pressure, whereas drowsy wakefulness typically results from higher sleep pressure in association with a sleep disorder preventing the onset and maintenance of deeper sleep.

 

More details to come.

​

Non-ORP Digital Biomarkers: Five EEG features are included in this category. Three of these, Delta wave characteristics, EEG spindle characteristics, and alpha intrusion, have been known for several decades and have well-documented functional roles in sleep science but have not been readily available for general use in the clinics or in research laboratories. They are now available on demand through Pharos. The other two (normalized EEG power and arousal intensity) are new, with potential implications for sleep physiology and pathophysiology:

 

  • Delta Wave Characteristics: An emerging metric that is being shown to have an important role in sleep health.  

    • ​More details to come.
       

  • EEG Spindle Characteristics: Density (average number of spindles per minute), power, and frequency of Spindles during stage 2 and stage 3 using a validated spindle detector. 

    • ​More details to come.
       

  • Alpha Intrusion: Spikes of Alpha activity in the EEG have been associated with chronic pain, fibromyalgia, chronic fatigue, and other disorders. We report on the % of sleep that includes alpha intrusions. Measured as % of 3-second epochs during NREM sleep with alpha power >30 μV2, the threshold for visual detection of this phenomenon. 

    • More details to come.
       

  • Normalized EEG Power: In some subjects, power in all relevant frequency ranges is high, while in other subjects it is low. This index is the average of ratios of power in individual frequency ranges to average power in the corresponding frequency ranges among the population being studied when the individual ratios are approximately similar (i.e. amplification is generalized and not limited to specific frequencies). The index was recently found to be a predictor of future development of dementia. 

    • More details to come.
       

  • Arousal Intensity and Heart Rate Response to Arousals: Grading of arousals on a scale of 0-9. There is high variability among even healthy young adults in terms of arousal intensity and heart rate response, though it is stable within individuals. More intense arousals have been shown to have greater risks and can also be a contributing factor to upper airway instability .

    • More details to come.

​

Custom Biomarkers: Pharos is open to incorporating custom software developed by external scientists in Pharos’ scoring program under a confidential non-disclosure agreement. The incorporated software would belong to the scientist who developed it and would be used exclusively to score studies that originate from the scientist providing the external software.

Get in Touch!

Have a question or looking for more information? We’re here to help.

bottom of page