Bio-signals Collecting System for Fatigue Level Classification, 2023, Younggun Lee et al

Dolphin

Dolphin

Senior Member (Voting Rights)
https://arinex.com.au/EMBC/pdf/full-paper_317.pdf

Bio-signals Collecting System for Fatigue Level Classification*

Younggun Lee1 , Yongkyun Lee2 and Dongsoo Kim3

Abstract—

Fatigue is a risk factor that reduces quality of life and work efficiency, and threatens safety in a high-risk environment.

However, fatigue is not yet precisely defined and is not a quantified concept as it relies on subjective evaluation.

The purpose of this study is to manage risks, improve mission efficiency, and prevent accidents through the development of machine learning and deep learning based fatigue level classifier.

Acquiring true fatigue levels to train machine learning and deep learning fatigue classifier may play a fundamental role.

Aims of this study are to develop a bio-signal collecting device and to establish a protocol for capturing and purifying data for extracting the true fatigue levels accurately.

The biosignal collection system gathered visual, thermal, and vocal signals at the same time for one minute.

The true fatigue level of the subjects is classified through the Daily Multidimensional Fatigue Inventory and physiological indicators related to fatigue for screening the subjective factors out.

The generated dataset is constructed as a DB along with the true fatigue levels and is provided to the research institutions.

In conclusion, this study proposes a research method that collects bio-signals and extracts the true fatigue levels for training machine learning and deep learning based fatigue level classifier to evaluate the fatigue of healthy subjects in multi-levels.
 
I've added breaks to the quoted sections below to make them easier to read

In this paper, the fatigue level of the subjects is classified with the proposed Daily Multidimensional Fatigue Inventory (DMFI) and four physiological indicators related to fatigue.

Physiological indicators include reaction time and success rate of

Psychomotor Vigilance Test (PVT) as an indicator of acute fatigue,

blood lactate level as an indicator of physical fatigue,

salivary C-Reactive Protein (CRP) level as an indicator of cumulative fatigue,

and salivary cortisol level as an indicator of mental fatigue.

In order to determine the true fatigue level, the DMFI score is first used to screen for discrepancies between subjective fatigue levels and evaluated fatigue levels. Furthermore, training data are filtered out when physiological values exceed the normal range.

The main contribution of this paper is the development of a data collection system for training classifiers that can measure daily integrated fatigue in a short time via remote sensors for healthy general populations.
Click to expand...


We aim to develop a real-time fatigue level classifier based on artificial intelligence to overcome the limitations of subjective fatigue level evaluation. By measuring the fatigue level instantly in the field, it can prevent safety accidents and improve the quality of task performance. To develop the machine learning (ML) and deep learning (DL) fatigue level classifiers which utilize visual, thermal image, and vocal signals,
Click to expand...

I find this quite an exciting development. I don't know how effective and valid the measures of fatigue are, but this is the first time I've seen attempts to find objective measures of fatigue. I do recognise this is about fatigue in healthy people, which is mostly about tiredness and healthy muscle fatigue, but I think there are overlaps with unhealthy fatigue in ME/CFS.

I hope this will be explored by ME/CFS researchers as part of development of better outcome measures. If it were found that this sort of approach could really be helpful in disease as well as in healthy people, it could be really useful, I think. Maybe one day we can dispose altogether of the Chalder Fatigue questionnaire!
 
Last edited:
I just tried the PVT ( http://www.sleepdisordersflorida.com/pvt1.html#responseOut ) and got a result of 283 ms, which was rated "excellent". That's despite having written "groggy, sluggish" in my journal this morning. So, based on that I don't think PVT is a good measure of ME's "fatigue-like" symptom.

CRP seems to be a marker for systemic inflammation. Exertion would cause that to rise, but PWME feel "fatigued" even without exertion, so that seems like it won't work well for ME. Hmm, this choice seems misguided, since exertion = fatigue increase might be valid, and exertion = inflammation increase might be valid, and inflammation = increased CRP might be valid, but that doesn't mean that fatigue = CRP increase. Logical fallacy there.

Blood lactate probably won't work either, since PWME are generally less likely to be that physically active, and again, ME's symptom isn't related to actual physical fatigue.

I just checked for "salivary cortisol fatigue" and found a Harvard paper with this "The cortisol level, when checked four times in a 24-hour period, was no different between fatigued and healthy patients in 61.5% of the studies." That doesn't sound useful for measuring ME's symptom either.

So, excellent goal, but the method seems based on ignorance of what fatigue really is ... and what ME's "fatigue-like" symptom is. I think ME's symptom is more likely neurological, and so will either have to be measured from CSF samples (risky and not guaranteed to reveal localized factors), or via brain scanning of some form. Maybe there's a measurable difference in firing rates or numbers of neurons firing when attempting to accomplish a task. Maybe the beta/delta wave ratio is different when "fatigued". Maybe glucose or fatty acid metabolic rate in glial cells is reduced when "fatigued". Maybe fatigue can be measured by the ratio of butyrate consumption rate of neurons in a certain volume of the brain vs spinal neuron pulse width.

If the goal is to prevent workers from handling critical tasks when not really up to the challenge, then specific tests along the lines of the PVT would probably be better. One test for attention span for flight controllers; one eye-hand (and memory test) test for surgeons, etc.
 
You must log in or register to reply here.
Back
Top Bottom