Measuring Fatigue

You can't analyze what you can't measure. If you can't analyze, you can't predict. And, if you can't predict, you can't manage. So, measuring the fatigue has to be the first step in managing the fatigue.

Problem is, there is no "thermometer" out there we can use to measure fatigue. All we have is tedious questionaires like SF-36, and it is subjective since it is based on how the patient feels. It is also impractical for everyday use because it takes too long to complete even when automated. A more practical alternative is a simple rating. You can rate your fatigue from 1 to 9, for example. It only takes a minute or so to think about how you felt on a given day and come up with a number.

It is surprisingly accurate, it turned out. I have been rating my fatigue and the subjective number has a strong correlation to a physical measure. (More on this later).  But this type of rating suffers from the problem of perception drift. What you think is 5 today may not be 5 a year from now. Instead, it could be 4. That would make the subjective fatigue ratings uncomparable over a long period time.

So, what we really need an objective measure. Once I meticulously logged the time I spent resting for a period of about 6 weeks. (As you can see from the example below, I'm horizontal about half of my waking hours.) And it turned out that my subjective fatigue rating had over 80% correlation to the time spent resting on a given day for the first 4 weeks. (It mysteriously dropped to 20% for the last 2 weeks though. I am guessing that it was because I slacked off with logging. I forgot to record the ending times at several occasions. I had to ignore those since the startig time alone is useless without the ending time.)

7/22/2015	10:43	11:17	0:34	7:38
	11:40	12:11	0:31
	12:16	12:34	0:18
	13:47	13:48	0:01
	13:48	15:20	1:32
	15:25	17:31	2:06
	18:35	20:20	1:45
	21:02	21:53	0:51
7/23/2015	11:17	11:28	0:11	7:24
	11:40	12:43	1:03
	13:05	13:32	0:27
	16:00	16:44	0:44

               restingMinutes    fatigue
restingMinutes      1.0000000 -0.8248002
fatigue            -0.8248002  1.0000000

Another way is to measure the total number of steps that you take. Presumably, you are less likely to move when you are fatigue, and less movement  means less steps. The problem is that you may force yourself to move despite fatigue because you have things to do. Likewise, you may not move when you are not tired, because you are working sitting down. So, the total number of steps probably are not going to correlate to fatigue level too well unless you are moving when and only when you feel like. The morning till 1PM usually is such time for me; I don't take a walk or do work till after lunch, so I'm more likely to move when I feel like to in the morning.

I use R with zoo (time series) package to analyze my data and the code to do this is below. The correlation between number of steps and fatigue only came out to be 40%, so this approach is not good enough. We need a strong correlation in order to use it as the proxy for fatigue level.

# 'fintradays' is the time series data containing steps, heart rate,
# etc, in 1 min resolution. 'daily' is the daily
# summary including the next day's fatigue level.
#
steps=by( 
    fintradays, as.Date(index(fintradays), tz="US/Pacific"),
    function(fintraday){sum(fintraday$steps[1:(60*13)])}
)
steps=zoo(steps, as.Date(names(steps), tz="US/Pacific"))
steps=merge(steps, fatigue=lag(daily$fatigue,-1), all=F)
cor(steps,use="complete")

            steps   fatigue
steps   1.0000000 0.4021578
fatigue 0.4021578 1.0000000

Yet another way is to measure the pace. People walk slower when fatigued or sick. If you can measure the number of steps per minute, that could serve as a proxy for fatiguee level. The amble around the house however usually last less than a minute, so it is hard to measure the speed that way. Measuring the speed while walking outside won't work either because I deliberately control the walking speed. That is because walking at above certain pace, 98 steps per minute in my case, causes post-exertional sickness. (Yes, I'm calling it sickness rather than vague-sounding malaise). But, if you can figure out how to measure the natural pace, it just might serve as the measure of fatigue. (I may figure this out eventually, so check back again later).

At the end of the day though, measuring the "fatigue" itself may not matter. The ultimate goal is to maximize the functioning. For that, you only needs to measure the output, whatever that may be, and manage your activities to maximize it. For most CFS patients like myself who spends inordinate amount of waking hours in horizontal position , that goal usually is to stay upright long enough to take care of the activities of daily living and get some work done. So, the amount of time spent vertically/horizontally is probably the best performance measure anyway. And it just happened to be strongly correlated to the subjective feeling of fatigue. The number of steps, on the other hand, often serves as the performance measure for healthy people. Doctors prescribe them 10,000 steps per day.