Disorder & Data

Sleep Disorders

Introduction

Although there are several factors to my sleep, Non-24-hour sleep-wake disorder (shortened to N-24) is generally the most prominant effector. In the briefest explination, it means that one's circadian rythm isn't 24 hours long (usually it's about 25). It's about rare enough in sighted people that there's not a real statistic for how common it is, and no particular reason for why it develops. I'll link here to a better explanation.

I've found that trying to explain a sleep disorder to most people doesn't communicate very well. I assume it's because most people can take the ability to sleep for granted, minus occasional insomnias. I'm not particuarly even worried about stigma, but I still end up having to describe this to people as insomnia if I want to communicate it briefly in a way they'll easily understand. Even when I take the time to explain it, I sometimes still see the other person caught off gaurd once some symptom shows, and it can still take further attempts to communicate how it actually effects me. So, what I want to say here is that I end up caring a lot about this, because to not be sleeping properly is to be dealing with it constantly.

This section is going to cover sleep in the context of both data and described experience.

Living With

Consequences of irregular sleep

I want to start with the more personal account before jumping into numbers. When I've slept on any given day can vary heavily and unpredictably, or it can stagnate at inconvenient times. 

I have a difficulty to keeping scheduled time commitments. This will manifest as either sleeping through an appointed time frame, or as having to postpone sleep until after an event and attending it while sleep deprived. These behaviors can lead to being perceived (respectively) as irresponsible or unengaged and dim. Because it's a constant issue, irregular sleep makes it difficult to attend school or hold a job with scheduled shift hours. I tend to describe myself as being 'just barely over the line' of being abled enough to do these things. I've gotten fairly used to functioning (or appearing to be so) at levels of sleep deprivation that feel barely lucid. Sometimes my sleep will stagnate at a mostly nocturnal place, and most institutions and businesses will become inaccessible to me at these times. This is the same with people, and I'll be put out-of-contact with friends. It would be another section to explain better what being stuck nocturnal really feels like, as a second way of living. There's also a difficulty with keeping some sort of body-routine, or even a daily routine to speak of. No set meal times, or gaurenteed time to get ready in the morning before being straight out the door.


Timeframes data functions on

In practice, I find that the previous 3-4 days of sleep are the ones that are relevant to a current day's “alertness”. If I didn't have a log to reference I would be relying on memory (which will only be one or two days plus the current). Without log or memorization I would be struggling harder for holding down a semi-rhythm, or explaining to myself if I'm sleep deprived, psychotic, or just needing a sip of caffiene. 

On a timescale of 1-2 weeks, I can capture some larger trends and make predictions on where the circadian rhythm will land me. I can usually expect my circadian rhythm to be around 25 hours long, so in theory I would only need to know yesterday’s sleep and would expect to sleep and wake an hour later than that. This would be true without the interference of schedules, various incentives to be diurnal vs. nocturnal, and energy or mood levels. A week or more is enough to balance out smaller or daily upsets in the pattern.


Mind states and patterns

As mentioned earlier, sleep doesn’t exist in a vacuum without schedules and time commitments. A second factor would be moods- energy levels, stress, episodes, whatever you call it. However, being sleep deprived, nocturnal, or on a schedule made irregular from time committments will end up effecting these right back. Mood is a lot more subjective than "hours you were asleep for", and so I've never tracked it in a similar way. The most important bit from this is that days that aren't logged are usually because I was feeling particuarly lethargic and would've likely been sleeping more than average.


Overall timeline

The data I have logged breaks itsself into 4 main chunks. I'll give a brief overview of what these correspond to:

The first is from 09/20/20 to 02/01/21. My highschool went fully-online for quarantine lockdown in Feb '2020. I know that during in-person highschool I hadn't noticed N-24, but there was insomnia and a delayed sleep phase. This means the N-24 develops sometime between that Febuary and when the first log is made. Since school is online and low-concequence, there's very minimal pressure to be awake at certain hours non-natural to my body rythm. I also feel unnaturally lethargic and disconnected in this time.

The second is 08/30/21 to 05/03/22. This is the first year of college spent at The Cooper Union, which I moved to straight out of being in quarantine for a year and a half. From pretty much the first to final day I'm in some level of psychosis (possibly accompanied with bits of hypomania), so this chart probably has the most unhealthy and most varried sleep patterns. I'm also mostly still trying to be awake during classes, so it's not a free schedule. I believe I noticed the first symptoms of prodome shortly before beginning lockdown, so this is extra context to if the development of the N-24 is also schizophrenia onset, and not just a response to lockdown.

The third is from 08/06/22 to 04/29/23. This is during the gap year following being at Cooper, for a portion of time where I'm working an irregular job shift in midday-evening. There's no psychosis here (or really over the gap year at all), and I'm in a similar state to quarantine except for a stimulant perscription and real concequences to sleeping through a shift (so doing better emotionally compared).

The fourth is from 07/17/23 to 12/26/2023 (current). This begins right before starting the previous semester at CMU. There's some small bits of psychosis/hypomania upticks in this period, but I could sort of reset an issue if I caught it immediatley and dropped everything until it passed, so it's no where near as active as at Cooper.

So, to describe this in the way I would expect it to show up in numerical data: the first and third chunks would categorize together as "fatigued and bored, prodome/remission", and the second and fourth as "high stress, active phase/possibility for it to be". The first and second chunks should be more severe versions of their respective third and fourth chunks.

Charts

Getting into drawing some charts

This section deserves a preface: I haven't really been in a math class since the beginning of 11th grade, and I'm not exactly sure of where to take things. I'm going to explain the process I'm going through with trying to figure things out mainly for myself here. Drawing the graphs with Matplotlib in Python.


The first graph here draws a day number (days since 09/20/20) to the corresponding number of hours I was asleep on that day for. The grey area is a line at 7.5 hours, the average number of hours slept in a given day. I find that division useful to watching the weekly average vary across it.


The second graph makes each chunk of being awake then asleep a bar on the x-axis, regardless of if either of those passed across midnight. The x-axis is sorted chronoligically. I was hoping it would be to somehow possible to correspond this to the previous chart so I could view "circadian rythm length" compared to "average amount slept". 

After seeing it , I realized an issue (minus that it's gibberish) with continuing with this is that taking a nap, or waking up for a moment and immediatley going back asleep resets the day. One hint of this is it'll return saying the average day length for me is 22.3 hours long.

So, the next step would be to figure out what counts as a nap (or as waking up just to go back to bed). 

Let me try to see how I'd describe a nap colloquially, without doing calculations. I've called something a nap up to about 4 hours. 4 hours is definatley a bit long, but here's a (made up) example where I think 4 hours could still be a nap:

I'd say it's a nap because the amount of time awake before and after it is so small, and neither I'd really count as long enough to be it's own 'full day'. So it seems more like falling asleep (albeit for a long time) in the middle of a day.

However, using "any amount of sleep under 4 hours" as critera for 'nap' wouldn't work. Here's a made up) example where I'd count 2 hours as a full night's sleep:

I'd count this as 'sleep' rather than 'nap', because otherwise I wouldn't have slept in 44 hours. So, "amount of time awake before and after a block of sleep" may be a qualifier.

Let's look at this then:

Colloquially, I would describe this as '4 hours of sleep that took 7 hours to get'. There's 3 missing hours there, so where do they go? I'm used to losing a lot of time to insomnia, and taking a few hours to fall asleep. In my head I like justifying to myself that an hour spent trying to fall asleep is equal to half an hour worth of sleep. However, in the charts I don't make that calculation, and those hours are all counted as 'awake'. So, then would these 3 missing hours get given to the preceeding and/or proceeding days' awake times, or do they just dissapear? Even if I counted the 3 hours as 1.5 hours of sleep, I'd still have 1.5 missing hours.

Let me get back to that qualifier definition. Let's say "a chunk of sleep must be both preceeded and followed by a chunk of time awake that's longer than 6 hours" for something to be 'sleep' and not a nap'. That makes this 4 naps.

Let's say "a chunk of sleep must be either preceeded or followed by a chunk of time awake that's longer than 6 hours" for something to be 'sleep' and not a nap'. That makes this 1 hour of sleep, 2 naps, and then another hour of sleep.

Let's say "a chunk of sleep must be either preceeded or followed by a chunk of time awake that's longer than 6 hours, and hours inbetween all count as sleep". That makes this 7 hours of sleep. This could be rephrased as "something can be counted as a period of sleep only once I've already been awake for >=6 hours, and it stops being counted once the following amount of awake time is >= 6 hours." And then add "x happens to all hours awake in the period of time counted as sleep". This makes it so I don't have 'nap' as a seprate category from 'sleep' anymore.

This example day is made up to illustrate a point, but there's pleanty of real examples that have a similar pattern, let me bring in two of those:

I color coded these to show how that definition would label them (blue = asleep according to definition). The right one turns out fine, but the left shows 31 hours asleep. So lowering the arbitrary 6 hours to 4 hours could fix that (breaks into 14 asleep, 4 awake, 13 asleep). 

Except that's not a full solution, I'd want this solitary 1 hour chunk to count as a nap, and not break up the portion of time where I'm awake. So, this means I have to be defining 'awake chunks of time that can bridge across naps' and 'asleep chunks of time that can bridge across waking up'. Okay- keep the definition for sleep, and add "something can be counted as a period of awakeness only once I've already been asleep for >=x hours, and it stops being counted once the following amount of asleep time is >= x hours".

I went to check the dataset just to make sure there's not a point where I'm neither awake or asleep for more than 4 (or something reasonable and low) consequitive hours for any signifigant amount of time, and here's probably the most troublesome chunk of time for that I could find:

Asleep for 7 hours, awake 50 (including 6 hours of naps), asleep 7. =2 periods of sleep, 1 of awake. 

Asleep for 7 hours, awake 17, asleep 2, awake 5, asleep 2, awake 10, asleep 1, awake 6, asleep 1, awake 6, asleep 7. =6 periods of sleep, 5 of awake. 

Which says that there's 50 hours awake without a nap of longer than 2 hours. Keep in mind, the max amount of time without any sleep (including naps) is 52 hours. ** I'm adding an ammendment here for myself that "I'd count 2 hours as 'sleep' rather than 'nap', because otherwise I wouldn't have slept in 44 hours" is possibly more of an unhealthy barganning tool rather than a good statistical measurement for when a day resets, and should be reexamined. ** The one on the right is true according to my rule (assuming 5 hours awake is enough to start a count for a new day), but if I care about calculating circadian rythm, the difference between waking up and falling asleep 1 time vs. 6 times matters. To myself, I would describe this bit as "getting by off naps, without any real sleep", so the left is more accurate to experience (although, not sure 'experience' as a measure is best practice). That'd mean ammend it to "something can be counted as a period of sleep only once I've already been awake for >=x hours, and it stops being counted once the following amount of awake time is >= x hours. It also must be >= y hours in duration." And the same for the 'awake' clause.


Okay, so. The other thing to do could be to figure out what "happens to all hours awake in the period of time counted as sleep". Take that '31 hours asleep' from earlier. 

Here's the two different splittings of this- the only thing I know I want from this is for it to not return only "31 hours asleep". (The max hours asleep without waking up is 31 hours, and that's only once, hence why this 31 should probably go. I'll note, in the initial charts, tracking 'woke up briefly, was immediatley back to sleep' is a logged a bit inaccuratley. That was a heavy med and not a coma). The left could return "chunk of sleep : 22 hours asleep, 9 hours awake". This feels fine to me and it'd work with the chart on the right too, so lets say the answer is 'insomnia' and 'nap' categories (awake hours in a sleep chunk and sleep hours in an awake chunk) and I deal with it later.

So, what I'm left with is "how many hours until I split to a new chunk?" and "what's a minumum duration for a chunk?"

I'll admit, finding that cut off point using a calculation (rather than just eyeballing it) is above my knowledge.

I know that I'm probably looking for standard deviation and/or outliers. In the top 'awake' chart (to the left) I can definatley see a dip around that 4 or 6 hour mark I was estimating earlier- times too long to be just waking up from insomnia, but too short to be a full day.

I'm throwing in a second distribution graph below (just for fun). Like I said earlier, I wanted to try to compare "circadian rythm length" to "average amount slept".  (Why are those the two I focus on? They're probably the ones you could get a prediction out of.) It's easy to see how insomnia and hypersomnia can relate to health, but I've definatley seen variations with circadian rythm/irregularity correspond to wellbeing. 

The leftmost is active psychosis (Jan '22), the right is prodome over quarantine (Jan '21). (Some charts measured to the half hour, some only to the hour. Half hours got rounded back, pushed to the left. So, this might return that I sleep a little later than I actually do.)

"Irregularity" might be a better measure to try to define- basically "how much does one day overlap with the days before and after it?"

The other, as just mentioned in that parenthesis, would be "what hours am I most commonly asleep during?"

Attempts at displaying those two shown below:

Most common hours to be asleep from all days.

Attempt at "irregularity"/ overlap. 

I'm gonna go draw up some better charts or something. Maybe split it into differences in the four sections.

Otherwise, let me leave you with this aside while I do that: