Apple Watch sensor has racial bias, claims new lawsuit

2

Comments

  • Reply 21 of 59
    XedXed Posts: 2,568member
    MacPro said:
    zimmie said:
    MacPro said:
    I can't see what Apple could do?  If they improved the sensitivity wouldn't that just make measurements better for light skins too, thus maintaining the differential?  It's physics not bias. 
    Not necessarily. All pulse oximeters use a measurement correction curve to convert from the amount of reflected light to an oxygenation percentage. That curve needs to change based on skin tone and possibly other characteristics we don’t even know about today (since only recently did a million people start carrying an oximeter around with them everywhere). Correcting the curve for darker skin would make readings for lighter skin less accurate, so it would need to be adjustable.
    That sounds not only doable, but automatable.
    It's not easy, which is why it's been an issue for decades.
    entropys9secondkox2
  • Reply 22 of 59
    zimmie said:
    MacPro said:
    I can't see what Apple could do?  If they improved the sensitivity wouldn't that just make measurements better for light skins too, thus maintaining the differential?  It's physics not bias. 
    Not necessarily. All pulse oximeters use a measurement correction curve to convert from the amount of reflected light to an oxygenation percentage. That curve needs to change based on skin tone and possibly other characteristics we don’t even know about today (since only recently did a million people start carrying an oximeter around with them everywhere). Correcting the curve for darker skin would make readings for lighter skin less accurate, so it would need to be adjustable.
    There should be a calibration setting. 
    It may not be that simple. The amount of IR light that gets absorbed in the skin is dependant on different factors including the % of melanin in the skin. Melanin absorbs certain IR frequencies effectively so what ultimately happens is you have a lot less signal to work with. If you have insufficient signal, calibration curves won't help because you simply don't have enough photons to work with. It is unfortunate and in a way, the physics is racially biased (add us soulless gingers with a lot of freckles as well). I guess the choice is to not invent the device at all since it cannot benefit everyone?
    anonymouseiOS_Guy80radarthekat9secondkox2
  • Reply 23 of 59
    zimmie said:
    MacPro said:
    I can't see what Apple could do?  If they improved the sensitivity wouldn't that just make measurements better for light skins too, thus maintaining the differential?  It's physics not bias. 
    Not necessarily. All pulse oximeters use a measurement correction curve to convert from the amount of reflected light to an oxygenation percentage. That curve needs to change based on skin tone and possibly other characteristics we don’t even know about today (since only recently did a million people start carrying an oximeter around with them everywhere). Correcting the curve for darker skin would make readings for lighter skin less accurate, so it would need to be adjustable.
    This cannot be right. Even among the same race, skin tone can vary a lot. Female in general is lighter than males. And skin tone can change after sun tan. Oxygenation level cannot be based solely on light reflection. 
    iOS_Guy809secondkox2
  • Reply 24 of 59
    Welp, looks like the racists (”it doesn't effect me so it can’t be racism”) are out in force.

    They key here is that even though the limits of oximeters are known, Apple went ahead and introduced the feature regardless of how it would impact users’ experience. No disclaimers. No apparent skin tone detection. No on Watch warning or dialogue allowing you to adjust for skin tone. And do on. It’s marketed, engineered, and now defended as working for the default skin color assumption... i.e light-skinned people. It’s not a leap to assume that has there been more people with darker skin tones involved in the decision making, this would have been flagged as a bug or a non-starter. So yes, this is the sort of structural or chain of casual racism that you can encounter.

    The separate case regarding medical treatment outcomes will likely be the more serious and far-reaching. The medical industry has a history of these types of errors in judgment. We’ll have to wait and see how well the device makers trained people, what disclaimers are in training literature, research papers and the like, as well as how well hospitals and other organizations trained their staffs.

    I’ll remind everyone that if the shoe were on the other foot, and light skinned people’s lives had been endangered by a fault in the device or training, this would be a major shitfest in general, in the news cycle for weeks, and likely the target of very public and swift government reaction.

    Also, anyone trotting out “frivolous lawsuit” — have a care. That term has been weaponized to question the legitimacy of valid the legal rights of black and brown Americans for decades. Do some research before you use that as a point of argument.

    There’s also a bunch of folks that mixing up the two cases. The case for medical treatment is being referred to as supporting evidence that Apple management and engineers should or would have known about the technology’s limitations as currently designed.

    For anyone interested in well researched institutional racism, how casually it’s applied, and what cumulative effect it continues to have on generations of Americans, I highly recommend The Color of Law by Richard Rothstein. It’s an excellent example of legal and historical analysis being applied to understand American legal history, and how it’s been used to create and defend racial hierarchies.

    Please think twice before responding to me in anything less than a civil and respectful way. If all you’re going to do is trot out manufactured talking points and threats to society, don’t bother. You’ve had quite enough air-time already. I won’t engage trolls.

    For anyone with a good-faith question or argument, I’ll do my best to respond, given my limited time and energy. (And also limited patience. Educate yourselves, people. If it were an engineering issue, you’d go look up info in technical and science resources — not Faux News and its ilk. Do the same here. Read a book, even if you disagree with it. Or ask me for a link or several.)
    If what you said is correct, you can simply use the $10 oximeters sold on Amazon to prove that they can measure oxygenation level better than the $399 Apple Watch. Have you done this before? Can you show us your result? 
    9secondkox2elijahg
  • Reply 25 of 59
    The world has gone mad. These litigation lawyers lack moral integrity. 
    9secondkox2elijahgiOS_Guy80
  • Reply 26 of 59
    The term "ambulance chasers" comes to mind.
    9secondkox2elijahgiOS_Guy80
  • Reply 27 of 59
    davidwdavidw Posts: 2,053member
    zimmie said:
    MacPro said:
    I can't see what Apple could do?  If they improved the sensitivity wouldn't that just make measurements better for light skins too, thus maintaining the differential?  It's physics not bias. 
    Not necessarily. All pulse oximeters use a measurement correction curve to convert from the amount of reflected light to an oxygenation percentage. That curve needs to change based on skin tone and possibly other characteristics we don’t even know about today (since only recently did a million people start carrying an oximeter around with them everywhere). Correcting the curve for darker skin would make readings for lighter skin less accurate, so it would need to be adjustable.
    There should be a calibration setting. 

    The problem with that is ....... what do you use to calibrate it against? It would take some one in the medical profession using the proper medical device and procedure to verify a person actual reading and then calibrating the Apple Watch to read the same. And this would need to be done every so often, to maintain accuracy. In order to self-calibrate the Apple Watch, one needs to know their actual reading to begin with. Otherwise it's still guess work.

    I guess Apple could provide color swatches of various skin tones, like they do with paint, and the wearer could choose the closest matching swatch and set their watch based on that. But even with that, how would the users know that the setting they chose, was more accurate than just leaving the watch setting at "default".

    Maybe Apple can have a medical profession with the proper medical device set up in an Apple Store and people with Apple Watch can get it calibrated there. But the malpractice insurance would make it prohibitively expensive. Losing one lawsuit could wipe out all the profits  Apple made selling Apple Watches.  
    9secondkox2
  • Reply 28 of 59
    davidwdavidw Posts: 2,053member
    Welp, looks like the racists (”it doesn't effect me so it can’t be racism”) are out in force.

    They key here is that even though the limits of oximeters are known, Apple went ahead and introduced the feature regardless of how it would impact users’ experience. No disclaimers. No apparent skin tone detection. No on Watch warning or dialogue allowing you to adjust for skin tone. And do on. It’s marketed, engineered, and now defended as working for the default skin color assumption... i.e light-skinned people. It’s not a leap to assume that has there been more people with darker skin tones involved in the decision making, this would have been flagged as a bug or a non-starter. So yes, this is the sort of structural or chain of casual racism that you can encounter.

    The separate case regarding medical treatment outcomes will likely be the more serious and far-reaching. The medical industry has a history of these types of errors in judgment. We’ll have to wait and see how well the device makers trained people, what disclaimers are in training literature, research papers and the like, as well as how well hospitals and other organizations trained their staffs.

    I’ll remind everyone that if the shoe were on the other foot, and light skinned people’s lives had been endangered by a fault in the device or training, this would be a major shitfest in general, in the news cycle for weeks, and likely the target of very public and swift government reaction.

    Also, anyone trotting out “frivolous lawsuit” — have a care. That term has been weaponized to question the legitimacy of valid the legal rights of black and brown Americans for decades. Do some research before you use that as a point of argument.

    There’s also a bunch of folks that mixing up the two cases. The case for medical treatment is being referred to as supporting evidence that Apple management and engineers should or would have known about the technology’s limitations as currently designed.

    For anyone interested in well researched institutional racism, how casually it’s applied, and what cumulative effect it continues to have on generations of Americans, I highly recommend The Color of Law by Richard Rothstein. It’s an excellent example of legal and historical analysis being applied to understand American legal history, and how it’s been used to create and defend racial hierarchies.

    Please think twice before responding to me in anything less than a civil and respectful way. If all you’re going to do is trot out manufactured talking points and threats to society, don’t bother. You’ve had quite enough air-time already. I won’t engage trolls.

    For anyone with a good-faith question or argument, I’ll do my best to respond, given my limited time and energy. (And also limited patience. Educate yourselves, people. If it were an engineering issue, you’d go look up info in technical and science resources — not Faux News and it’s ilk. Do the same here. Read a book, even if you disagree with it. Or ask me for a link or several.)
    Are you sure? Or are you being bias against Apple.

    From Apple own support sites.


    >Subject pools included a wide range of skin types and tones to ensure that the sensor platform can accommodate the full range of users and maintain accuracy. At the wavelengths that Apple Watch uses, melanin is a strong light absorber — particularly in the green and red part of the spectrum — potentially making PPG measurements more difficult in users with darker skin tones. To account for this, the

    Apple Watch sensing platform senses the amount of detected light signals, and it automatically adjusts the LED current (and hence the light output), photodiode gain (sensitivity to light), and sampling rate to ensure adequate signal resolution across the range of human skin tones.<


    https://support.apple.com/en-us/HT211027

    >Measurements taken with the Blood Oxygen app are not intended for medical use and are only designed for general fitness and wellness purposes.<




    radarthekat9secondkox2elijahg
  • Reply 29 of 59
    Seems Apple has always made it clear the blood oxygen readings are not meant for medical use. The Apple Watch 6 with the blood oxygen feature shipped in Sep 2020, and as far as I can tell the study with analysis of the racial bias of oximeters was Dec 2020, and that was specific to misdetection of occult hypoxemia, which, again, the Apple Watch use case does not claim.

    However the ECG feature of the Apple Watch has saved people's lives with dark skin, so pinning the racist tag on Apple seems a bit of a reach.
    9secondkox2iOS_Guy80
  • Reply 30 of 59
    radarthekatradarthekat Posts: 3,843moderator
    A lawsuit of this type must show damages.  How was the plaintive damaged?  And to what extent?  Can the damages he suffered be applied equally to all plaintives his suit attempts to cover?  Can a person with healthy heart and blood oxygen levels claim any damages from readings that were not 100% accurate if they had no adverse health outcomes? 

    Side note:  Can I claim damages because the calorie burn counted by Apple Watch is not as accurate for me as it is for some other user, because you do know that Apple says right up front it’s not meant to be perfectly accurate but rather relative within all the readings for each specific user.  Can I claim I would have walked farther or eaten healthier if only I were getting perfectly accurate calorie burn readings, and since those readings weren’t perfectly accurate I ate too much and exercised too little and therefore became obese?  

    Back to the pulse oxygenation feature… how do I show damages, that were clearly imputed by Apple, and how do I measure the extent of those damages when there are many other factors involved with determining a person’s health?

    Right out of the gate I’d have to think that anyone who isn’t able to point to some adverse health effect that can be linked directly to a reading, or misreading by the Apple Watch’s pulse oximeter, is a person who cannot show damages.

    As to racial bias, it’s pretty clear that Apple took steps to be inclusive and work to mitigate issues with accuracy that physics imposes, not biased laws or anything else that @freeassociate2 would like to blame.  Sometimes there’s no conspiracy, no injected bias.  It is what it is and, in this case, Apple has likely done and will likely continue to do its best to serve everyone.   

    edited December 2022 9secondkox213485
  • Reply 31 of 59
    zimmie said:
    MacPro said:
    I can't see what Apple could do?  If they improved the sensitivity wouldn't that just make measurements better for light skins too, thus maintaining the differential?  It's physics not bias. 
    Not necessarily. All pulse oximeters use a measurement correction curve to convert from the amount of reflected light to an oxygenation percentage. That curve needs to change based on skin tone and possibly other characteristics we don’t even know about today (since only recently did a million people start carrying an oximeter around with them everywhere). Correcting the curve for darker skin would make readings for lighter skin less accurate, so it would need to be adjustable.
    There should be a calibration setting. 
    I can't imagine the calibrating process would be "what color shade is your skin: choose 1 to 10" and Apple Watch will adjust LED strength accordingly.
    iOS_Guy80
  • Reply 32 of 59
    Seems Apple has always made it clear the blood oxygen readings are not meant for medical use. The Apple Watch 6 with the blood oxygen feature shipped in Sep 2020, and as far as I can tell the study with analysis of the racial bias of oximeters was Dec 2020, and that was specific to misdetection of occult hypoxemia, which, again, the Apple Watch use case does not claim.

    However the ECG feature of the Apple Watch has saved people's lives with dark skin, so pinning the racist tag on Apple seems a bit of a reach.
    Yep, that’s because it takes way too long compared to a regular pulse oximeter. Plus most of the regular ones vary in their readings by up to + - 5pts in my experience. 

    This douche probably won’t get very far because he can’t prove malicious intent on a device that has had these challenges before they were put in an Apple Watch. I sure hope he has enough money to pay Apple’s legal fees when he loses. 


    radarthekat9secondkox2elijahgiOS_Guy80
  • Reply 33 of 59
    rcomeau said:
    MacPro said:
    I can't see what Apple could do?  If they improved the sensitivity wouldn't that just make measurements better for light skins too, thus maintaining the differential?  It's physics not bias. 
    I don’t know what wavelength of light they use for these but there probably is a range in the non visible spectrum that could be used where skin tone would be irrelevant.
    Pulse oximetry depends on the differential absorbance of 2 (or more) different frequencies of IR light in the blood, specifically the oxygenated and deoxygenated haemoglobin. You can pick 2 IR wavelengths where the absorbance curves of oxi and de-oxi cross each other (typically pick one <700nm and one >800nm. Acquire relative intensity (changes in time) of both these wavelengths and you can fit them to a curve using something called the modified Beer-Lambert law to get an estimated change in blood oxygenation. You have to pick wavelengths that are on the right spots on the curve as well as what wavelengths of LED are actually available and affordable (only a new nm difference can mean a $1 LED or a $100LED due to the material in the LED that emits the light). This is not an easy problem to solve, other than blasting more light in, but you then run into concerns over heating the skin (absorbing IR light means conversion into heat) as well as power consumption and safety (too intense and you run the risk of eye damage if some fool stares into the back of the watch with the LEDs on. Not an easy fix.
    The Beer-Lambert law is pretty much the attenuation due to the imaginary part of the complex (real plus imaginary parts) index of refraction of the material (in this case the human tissue). The real part controlling the change in phase as the light propagates and so the angle of refraction. For a given material the index of refraction will have peaks and valleys in it described empirically by the Sellmeier equation. Basically the engineer would have to find a frequency of light where the imaginary part of the index of refraction isn’t significantly effected by the melanin in the skin. But as you mentioned whether or not it is economically practical depends on the availability of LEDs in that range. Bandgaps of LEDs can be tuned using structures like quantum wells though.
  • Reply 34 of 59
    hodarhodar Posts: 357member
    My camera does not take great picture when it’s dark.   The daytime pictures look great

    How much more racist can you get?

    /sarcasm_flag_off
    algrradarthekat9secondkox2elijahgiOS_Guy80
  • Reply 35 of 59
    algralgr Posts: 27member
    Oh good god, learn some physics.

    - The watch works by shining photons.
    - The point of dark skin is to absorb photons 
    - Less photons means less data.  
    = If it was possible to shine more photons they would say that Apple was racist for the battery not lasting as long.  Maybe that is what is happening now.

    I remember back when racism was an "old people problem" that was not going to be a problem in the future.  Now the problem seems hopeless.
    radarthekat9secondkox2elijahgiOS_Guy80
  • Reply 36 of 59
    While Apple might play their famous sosumi sound, on second thought, shouldn’t they rather sue the inventor of the laws of physics? Who was that again…
    9secondkox2elijahgiOS_Guy80
  • Reply 37 of 59
    MplsPMplsP Posts: 3,931member
    MacPro said:
    I can't see what Apple could do?  If they improved the sensitivity wouldn't that just make measurements better for light skins too, thus maintaining the differential?  It's physics not bias. 
    I don’t know what wavelength of light they use for these but there probably is a range in the non visible spectrum that could be used where skin tone would be irrelevant.
    The wave lengths (more than one is used) are determined by the absorptive spectra of oxyhemoglobin and deoxyhemoglobin. You can't change those based on skin color.


    9secondkox2elijahgiOS_Guy80
  • Reply 38 of 59
    Welp, looks like the racists (”it doesn't effect me so it can’t be racism”) are out in force.

    They key here is that even though the limits of oximeters are known, Apple went ahead and introduced the feature regardless of how it would impact users’ experience. No disclaimers. No apparent skin tone detection. No on Watch warning or dialogue allowing you to adjust for skin tone. And do on. It’s marketed, engineered, and now defended as working for the default skin color assumption... i.e light-skinned people. It’s not a leap to assume that has there been more people with darker skin tones involved in the decision making, this would have been flagged as a bug or a non-starter. So yes, this is the sort of structural or chain of casual racism that you can encounter.

    The separate case regarding medical treatment outcomes will likely be the more serious and far-reaching. The medical industry has a history of these types of errors in judgment. We’ll have to wait and see how well the device makers trained people, what disclaimers are in training literature, research papers and the like, as well as how well hospitals and other organizations trained their staffs.

    I’ll remind everyone that if the shoe were on the other foot, and light skinned people’s lives had been endangered by a fault in the device or training, this would be a major shitfest in general, in the news cycle for weeks, and likely the target of very public and swift government reaction.

    Also, anyone trotting out “frivolous lawsuit” — have a care. That term has been weaponized to question the legitimacy of valid the legal rights of black and brown Americans for decades. Do some research before you use that as a point of argument.

    There’s also a bunch of folks that mixing up the two cases. The case for medical treatment is being referred to as supporting evidence that Apple management and engineers should or would have known about the technology’s limitations as currently designed.

    For anyone interested in well researched institutional racism, how casually it’s applied, and what cumulative effect it continues to have on generations of Americans, I highly recommend The Color of Law by Richard Rothstein. It’s an excellent example of legal and historical analysis being applied to understand American legal history, and how it’s been used to create and defend racial hierarchies.

    Please think twice before responding to me in anything less than a civil and respectful way. If all you’re going to do is trot out manufactured talking points and threats to society, don’t bother. You’ve had quite enough air-time already. I won’t engage trolls.

    For anyone with a good-faith question or argument, I’ll do my best to respond, given my limited time and energy. (And also limited patience. Educate yourselves, people. If it were an engineering issue, you’d go look up info in technical and science resources — not Faux News and its ilk. Do the same here. Read a book, even if you disagree with it. Or ask me for a link or several.)
    No one said wrong things don’t happen or that racism doesn’t exist. 

    This is simply not an example of it. 

    The word “frivolous” applies. It’s not a weapon. It’s an adjective rightfully used. The only thing “weaponized” here is the word “racist” used by you to try to shut down others who don’t agree with you. 

    Your post is literally what some of us were talking about. Stop calling literally everything racist because it isn’t perfect. Save it for actual racism, which we all agree is evil and happens far too often to people of all colors (some more often snd severe than others) around the world. 

    There is no inherent racism, bias, or malice in this technical limitation. Zero. 

    So you have a care for the world around you and stop polluting it with misdirected hate toward those who don’t follow your whims and beliefs and victimhood signaling. There is a world that exists beyond you and the difficult things that may have happened to you. Time to grow up and get along with others. 
    edited December 2022 elijahgiOS_Guy80cg2713485MplsP
  • Reply 39 of 59
    One thing we must not forget - the Apple Watch is not a medical device and should never be relied on for medical readings, such as SpO2, ECG, or anything else remotely health related.

    Technology limitations based on skin tone are physical characteristics, and not something Apple can control. 
    9secondkox2radarthekatelijahg
  • Reply 40 of 59
    zimmie said:
    MacPro said:
    I can't see what Apple could do?  If they improved the sensitivity wouldn't that just make measurements better for light skins too, thus maintaining the differential?  It's physics not bias. 
    Not necessarily. All pulse oximeters use a measurement correction curve to convert from the amount of reflected light to an oxygenation percentage. That curve needs to change based on skin tone and possibly other characteristics we don’t even know about today (since only recently did a million people start carrying an oximeter around with them everywhere). Correcting the curve for darker skin would make readings for lighter skin less accurate, so it would need to be adjustable.
    There should be a calibration setting. 
    I can't imagine the calibrating process would be "what color shade is your skin: choose 1 to 10" and Apple Watch will adjust LED strength accordingly.
    Seems to be the simplest effort toward a solution. And sometimes that’s where the fruit is. Of course it’s not exactly so simple, otherwise it wouldn’t be an issue for so long. But if anyone can figure it out, it’s Apple. 

    Just a few years ago, watches told time and played really bad Atari 2600 style video games. Now they tell the time, control your phone, use the internet,  do email, messsging, manage your fitness, check your pulse, etc. that’s crazy. 

    And now we just expect everything to be perfect on this little watch - even when dedicated industrial grade equipment isn’t doing a better job. 

    And now there are actually people who want to SUE this company that has made such great strides in saving lives with a little watch by throwing the race card up when something doesn’t work perfectly?  The heck? No wonder they haven’t tried blood pressure monitoring yet. 

    So the idea of a calibration setting to get a better reading of blood oxygen seems logical, but actually accomplishing that will take some dedicated time if it’s possible at all. 

    And most importantly, as others have stated, Apple explicitly notes that the feature is not meant for medical purposes. Just to provide a general idea of health and wellness. As the lawsuit seems to be focused on the idea of a medical feature, it’s most likely to be tossed out by any competent judge not totally indoctrinated with critical race theory. 
    edited December 2022 ravnorodomelijahgcg27
Sign In or Register to comment.