1. Guest, the 'News in Brief' for the week beginning 21st September 2020 is here.
    Dismiss Notice
  2. Welcome! To read the Core Purpose and Values of our forum, click here.
    Dismiss Notice

The use of monitoring devices such as actometers to measure outcomes in clinical trials. Discussion thread

Discussion in 'BioMedical ME/CFS Research' started by Sly Saint, Sep 18, 2018.

  1. Sly Saint

    Sly Saint Senior Member (Voting Rights)

    Messages:
    5,117
    Likes Received:
    42,405
    Location:
    UK
    This post was copied and subsequent posts either copied or moved from this thread:
    David Tuller: Trial By Error: So What's Happening with the MAGENTA Trial?


    Re the actometers; I know others have said this but why on earth didn't they use fitbits?
    Crawleys always going on about how she is
    'like down with the kids, yeah', and has teenage children.

    It would have been an obvious choice; plus it would have measured loads of other things that she says are key to the childrens 'recovery' eg sleep patterns.

    Plus on activity "You can record individual activities by putting your Fitbit into Activity/Exercise Mode mode".

    same goes for all the others using the ruddy CFQ
     
    Last edited by a moderator: Sep 24, 2018
    MEMarge and ladycatlover like this.
  2. Invisible Woman

    Invisible Woman Senior Member (Voting Rights)

    Messages:
    6,915
    Likes Received:
    40,630
    I'm guessing here but I know from speaking to a very knowledgeable and helpful chap from their telephone helpline that fitbit (as of circa 18 months ago or so) are not designed to be, and should not be used as a medical device.

    Guessing some more- the accelerometers are presumably calibrated and have defined /limited variation between units so that it's is possible not just to compare the data from a single device, but also across devices.
     
  3. Adrian

    Adrian Administrator Staff Member

    Messages:
    4,755
    Likes Received:
    24,021
    Location:
    UK
    I think there is a difference between a medical device and a device measuring activity for a trial. There will be inaccuracies but for the levels of activity needed for a meaningful improvement I think they would be fine. Certainly better than a diary.

    One of the issues with being a 'medical' device is they need to go through approval steps which can be complex. Also as they do heart monitoring etc they may not be accurate enough to rely on for diagnosis.
     
  4. Invisible Woman

    Invisible Woman Senior Member (Voting Rights)

    Messages:
    6,915
    Likes Received:
    40,630
    They are certainly better than a diary, but I question whether they are accurate enough for use in a trial.

    For some the difference in activity is very small. Many of the devices in the market don't have the tolerances to cope with that. I know I can fool my device quite easily - it sees some arm movements as steps or even cycling!

    I don't know if there is anything governing the use of devices in trials in terms of the standards and accuracy needed.

    I think calibration would be a big problem. The devices would need to be within a given tolerance and that tolerance would need to be known. So lets say if we allowed a tolerance of +/-10%- for someone doing 100 steps a day, at either end of tolerance we have a reading of 90 vs 110 steps. That's quite a big difference.
     
    obeat, ladycatlover, MEMarge and 3 others like this.
  5. obeat

    obeat Senior Member (Voting Rights)

    Messages:
    588
    Likes Received:
    4,112
    I think she said that teenagers preferred to use their phone to record activity. How subjective is that? I imagine getting teens to use an actometer might nalso have problems. After all surely they are in control of turning it on? Or is it like an electronic tag??
     
  6. Invisible Woman

    Invisible Woman Senior Member (Voting Rights)

    Messages:
    6,915
    Likes Received:
    40,630
    My understanding of the actometers is that they measure rest as much as activity and therefore the devices are probably always left on.

    I am sure they record events such as being powered off, powered on and charged etc.

    I know with wrist worn fitness devices like fitbit triggering alerts use most of the battery life. Actometers wouldn't be alerting (as presumably that would affect behaviour rather than record it), so their battery life should be much longer.

    Of course they would also have to store more/longer data files too though.
     
    Last edited: Sep 19, 2018
  7. Invisible Woman

    Invisible Woman Senior Member (Voting Rights)

    Messages:
    6,915
    Likes Received:
    40,630
    Another thing that occurred to me re the use of actometers vs fitbit type devices that communicate with smartphone apps is that you wouldn't necessarily want the interaction.

    The smartphone apps are deliberately written with use of colours etc, to keep us using them. So a kid who is checking the app may change their behaviour based on the readings you see. That is after all what these apps are supposed to be designed to do - encourage you to change your behaviour.

    In the same way that sending a newsletter out to trial particpants extolling the virtues of a trialled treatment arm, you would be biasing the results of your treatment.
     
  8. Adrian

    Adrian Administrator Staff Member

    Messages:
    4,755
    Likes Received:
    24,021
    Location:
    UK
    I'm thinking a bit more about fit bits. The tolerance is important and needs to be much smaller than the effect being measured so if (Crawley in this case) is looking to measure 10% improvements then maybe they are not sufficient. But if a treatment is to have a large effect then the improvements should be bigger. The other thing about calibration is if we are measuring improvement with an individual device so the initial level may not be important but the change does need to be consistent across devices.

    I think there are two different things - there is the accelorometer chip (that detects movement) and then the processing of the results. I think the accelerometer chips in the apple devices are very accurate (I assume fitbit uses similar); one company was going to release a much more accurate accelerometer but dropped it which suggests there is no need for more basic accuracy.

    I did find some data sets collected from smart phones and watches which were being used for ML to interpret the movements that people were making (sitting, standing etc) but I didn't go through the papers to see how accurate they are getting.

    Some people are digging more into fit bit data. For example, I found this blog:


    which is using an undocumented fitbit API to pull out data and do their own analysis, so it could be that detailed data is available and this could be separated from the processing that fitbit do (if that is more dodgy than the other systems). But I suspect the raw accelorometer data never leaves the fitbit (because it would be large) and is preprocessed and this is what is sent to the fitbit cloud. From a quick glance at the API https://github.com/orcasgit/python-fitbit/tree/master/fitbit there seems to be the ability to pull a days data where something is recorded on a per minute basis which is probably not quite good enough.



    I've not used/looked at a fitbit so not sure how they interact with people but I agree here. I assume you can set goals so they don't hassle you?
     
  9. Trish

    Trish Moderator Staff Member

    Messages:
    25,847
    Likes Received:
    126,716
    Location:
    UK
    I have been using a Fitbit that measures movement (''steps'') and heart rate for a few months. It sends information via bluetooth to my computer, updating regularly by itself and when I ask it to.

    It displays various graphs of day, week and month as well as summary data of daily step count, resting pulse and time spent in the different pulse rate zones each day. It also does a sleep analysis of time spent in light, deep, and REM sleep based, I think, on some algorithm combining movement and pulse.

    I think it could be very useful in research for looking at how each patient's daily activity, resting pulse rate, time spent in aerobic pulse rate zone and sleep change over time. I've started tracking that for myself alongside tracking symptoms to see what it tells me. Too early to tell.

    As a research tool for assessing whether a patient is complying with planned activity increases, and seeing whether this can be sustained over time and how it correlates with symptoms, I think it would be enormously valuable.
     
  10. Invisible Woman

    Invisible Woman Senior Member (Voting Rights)

    Messages:
    6,915
    Likes Received:
    40,630
    You can, but would teenagers leave it alone? You might need to be able to lock them out. The interfacd is specifically designed to be attractive and draw you in, join up with other people etc.

    Based purely on my personal experience I disagree. If the patient has any additional issues such as anaemia or hypothyroidism it will skew the data and the data won't show what you think it shows.

    My own device is perfectly good and no less accurate than any of the others. It does not always accurately track sleep etc.

    Then I also worry about their use generally - not from a research gathering point of view, but from surveillance. Recently, one of the larger insurance companies have started to insist that people wear these devices as part of the terms of their policy.
     
    MeSci, MEMarge, Webdog and 1 other person like this.
  11. adambeyoncelowe

    adambeyoncelowe Senior Member (Voting Rights)

    Messages:
    2,192
    Likes Received:
    15,505
    My cheap, non-branded 'Fitbit' tracks all of the things above, however it can't track sleep before 11pm and it doesn't distinguish between stages 3 and 4 of sleep. There's just 'light sleep' and 'deep sleep'.

    Sometimes, I'm aware that it doesn't track steps if I'm stepping on one spot. But it still shows relatively high steps if I do more.

    So while it's not 100% accurate, it can still successfully capture increases in activity, even it's off.
     
    Binkie4, Skycloud, MEMarge and 6 others like this.
  12. NelliePledge

    NelliePledge Moderator Staff Member

    Messages:
    6,060
    Likes Received:
    38,173
    Location:
    UK West Midlands
    Even if not 100% technically accurate if they give people the tracker for 2/years including a reasonable period of time before and after the intervention it will give useful non subjective information on whether the individual participants increased and maintained any increase in activity. I find it hard to accept that a questionnaire or indeed a self recorded daily chart is ever going to be a better option.
     
  13. Amw66

    Amw66 Senior Member (Voting Rights)

    Messages:
    3,852
    Likes Received:
    17,862
    Unless you fill in a chart as you do things, it can be difficult to remember what you did and for how long.
    Then you are prone to making it up.
    Was there not a phone activity app?
     
  14. NelliePledge

    NelliePledge Moderator Staff Member

    Messages:
    6,060
    Likes Received:
    38,173
    Location:
    UK West Midlands
    Yes but I think that’s just an online form for filling in yourself so not objective
     
  15. Amw66

    Amw66 Senior Member (Voting Rights)

    Messages:
    3,852
    Likes Received:
    17,862
    If there was a real time phone app that you could simply hit a coloured symbol as you go through the day, that may make more sense of things .
    The number if apps that are free / peanuts would duggest that it may not be difficult to get a tailor made one for trials.
     
    ladycatlover likes this.
  16. MEMarge

    MEMarge Senior Member (Voting Rights)

    Messages:
    1,454
    Likes Received:
    8,656
    Per info on the Magenta website:

    "Graded Exercise Therapy

    Participants in this group will receive a detailed physical assessment including how far they can walk in 2 minutes and how many times they can move from sitting to standing in one minute. This will help us set a safe exercise programme. Participants will be asked about the exercise they do each day and will be helped to find their exercise baseline. The baseline is the average amount of exercise they do each day. It will be less than they do on a good day. When they have found their baseline, we will ask participants to slowly increase their exercise. When they are able to do 30 minutes each day, we will increase the intensity of exercise. participants will be asked to record exercise using either charts or the iPhone app ActiveME. They will not be asked to record other activities, only exercise. To make sure participants do not over do their exercise, we will ask them to use a heart rate monitor."

    So, @Amw66:
    The participants can use the (Bristol devised) "iphone app Active ME"

    http://www.bristol.ac.uk/ccah/research/childdevelopmentdisability/chronic-fatigue/magenta-trial/
     
    ladycatlover likes this.
  17. Amw66

    Amw66 Senior Member (Voting Rights)

    Messages:
    3,852
    Likes Received:
    17,862
    Not much use ehen it is only for exercise.

    I have a weird memory that the App had been dropped. ( This could be a false memory from lack of sleep )
     
    MEMarge likes this.
  18. Amw66

    Amw66 Senior Member (Voting Rights)

    Messages:
    3,852
    Likes Received:
    17,862

    https://itunes.apple.com/gb/app/activeme/id458308805?mt=8
     
    MEMarge and ladycatlover like this.
  19. Arnie Pye

    Arnie Pye Senior Member (Voting Rights)

    Messages:
    3,087
    Likes Received:
    16,900
    Location:
    UK
    [​IMG]
     
    MEMarge and ladycatlover like this.
  20. MEMarge

    MEMarge Senior Member (Voting Rights)

    Messages:
    1,454
    Likes Received:
    8,656
    Last edited: Sep 22, 2018
    ladycatlover likes this.

Share This Page