\n<\/p>\n
Self quantification is the<\/span> trend that just keeps on going. There\u2019s an ever-expanding world of wellness wearables and fitness trackers targeting consumers with shiny promises of the personal value to be had if they monitor stuff like their heart rate, activity and sleep \u2014 from smart watches, bands and rings, to smart scales, CGMs (continuous glucose monitors) and more.<\/p>\n Some of the fitness & wellness devices marketed to consumers have features that claim to be able to detect medical conditions or disease risks. Apple, for one, has made much of its FDA-cleared ECG (elctrocardiogram) and atrial fibrillation (irregular heartbeat) detection features over the years.<\/p>\n But medically cleared features remain the exception in the space. Most of the data being output by category products hasn\u2019t been reviewed by regulators so it can be tricky for consumers to know how accurate\/useful these \u201cpersonalized\u201d assessments really are. And what, if anything, they should make of all the \u201cgeneral wellness\u201d metrics and \u201chealth\u201d scores appearing in their apps.<\/p>\n Fortunately, there are some solid rules of thumb to get the most out of produces in the fitness and wellness category without losing sight of their limitations.<\/p>\n Researchers in the field we spoke to for this article are also quietly optimistic that tenacious personal health monitors, with their ability to go the distance by producing a longitudinal view of what we\u2019re up to, will, down the line, be able to deliver on the promise of preventative healthcare and help many more of us step away from bad habits that risk a long, slow slide into chronic disease.<\/p>\n We\u2019re also told there\u2019s lots of research being done to extract better signal out of noisy real-world data, including by using AI. And if we had to make a bet on where the category is headed, we reckon in-ear monitoring looks particularly interesting (see the last section of this article for more on this) \u2014 after all, rumors continue to suggest<\/a> Apple is looking at adding health features to its AirPods \u2014 so we\u2019re generally bullish on the longer term direction of travel for consumer health tech. But, as it stands, there are still some risks and pitfalls to avoid.<\/p>\n Did you find a fitness tracker under your Christmas tree this year? Are your New Year resolutions to be more active making you wonder if it\u2019s worth investing in a wellness wearable? Read on for help to keep your head amid the hype and find the best signal in noisy data. Finally, we also take a peek forward at where this evolving category could be headed\u2026<\/p>\n The cardinal rule is to approach all fitness\/wellness products with a critical eye \u2014 and be very wary of stuff that overpromises. In short: Read the small print, not the marketing.<\/p>\n Getting you to upload your data might be the main aim of a new and unproven product, especially if it\u2019s making eyebrow-raising claims of accurately assessing your risk of almost every condition\/disease under the sun. Using such a product is likely to help someone else a lot more than it helps you.<\/p>\n Case in point: A pitch for a smart mirror that was being shown off at CES this year sure raised our blood pressure: NuraLogix, maker of a device it brands as a \u201cMagicMirror\u201d, claims the product can produce all sorts of vital\/physical \u201cmeasurements\u201d and \u201chealth risk assessments\u201d off of a 30-second facial scan \u2014 including blood pressure, BMI, Type 2 diabetes risk, 10 year cardiovascular disease risk, hypertension risk and even anxiety and mental health risk, among a list that\u2019s too long to reproduce here in full. Per its press release, it\u2019s using \u201cpatented Transdermal Optical Imaging\u201d tech \u2014 which it dubs \u201ca novel form of Remote Photoplethysmography (rPPG)\u201d \u2014 and proprietary AI to produce \u201caccurate health data\u201d via lingering selfie.<\/p>\n However, if you scroll a further down the company\u2019s PR you\u2019ll find a small-print disclaimer at the bottom of each page \u2014 which states: \u201cIn the United States, this product is for Investigational Use Only. The performance characteristics of this product have not been established.\u201d (Rough translation: Those \u201caccurate\u201d measurements of health\/vital signs\/disease risk assessments we mentioned earlier are of unknown accuracy; this is not a diagnostic medical device.)<\/p>\n A year ago, NuraLogix claimed to be applying for FDA clearance, per a colleague\u2019s report on an earlier iteration of its selfie scanning tech which was being shown off at CES last year<\/a>. But, evidently, it hasn\u2019t managed to get sign-off on any of its myriad \u201chealth solutions\u201d yet, despite the heavy hype in its marketing.<\/p>\n Maybe this company\u2019s smart mirror will gain regulatory clearance for some functionality in future. But hype-first products like this, which arrive in the market prior to their \u201cperformance characteristics\u201d being \u201cestablished\u201d \u2014 accompanied by flashy marketing that touts multi-faceted utility \u2014 just don\u2019t pass the sniff test.<\/p>\n There can be a \u201cblurring of the lines\u201d that enables unsubstantiated health claims to be made, or at lest implied, by device makers, says Ahmar Shah, a senior research fellow in the University of Edinburgh\u2019s Medical School. \u201cUsually the reason people are able to get away with this is they market their device as a \u2018wellness\u2019 device, or say they\u2019re doing \u2018wellness monitoring\u2019. And when you do that then it\u2019s not considered a medical device,\u201d he tells TechCrunch.<\/p>\n This can lead to sweeping suggestions of health utility that haven\u2019t actually been substantiated, so long as device makers carefully frame and\/or caveat their claims. \u201cIt\u2019s an unregulated environment. You will see good stuff \u2014 but you will also see stuff that\u2019s just more marketing than some solid evidence base,\u201d he warns.<\/p>\n If the line blurring continues, pressure may grow to bring \u201cwellness\u201d kit under formal medical device regulatory frameworks \u2014 and at least require a certain standard of proof for any claimed health benefits. But, for now, the game \u2014 and the claims \u2014 continue.<\/p>\n Of course companies are always hungry for data to feed their product development, so they can turn out better, more effective products. But when it comes to consumer health tech, exaggerated marketing claims are a particularly dubious tactic as they may trick people into handing sensitive information to a commercial entity and getting nothing much in return (well, except the risk of their personal data being misused<\/a> \u2014 see Rule Six).<\/p>\n Another growing wave of startups in this space are devising and marketing \u201cgeneral wellness\u201d products that are getting even more up-close-and-personal with the user as tracking hinges on wet testing samples of bodily fluids to serve \u201cpersonalized\u201d insights to an app. (See, for example, Vivoo<\/a>\u2018s urine analysis for suspected UTI testing<\/a> or vitamin deficiency detection; or Zoe<\/a>\u2018s glucose monitoring for metabolism analysis and stool sampling for quantifying gut microbiome, to name two.)<\/p>\n While such products look novel (and may have promise), there\u2019s a risk of unproven assessments misleading users about their health and\/or making them anxious \u2014 especially if people assume outputs are more meaningful than they actually are. At this highly experimental end of the tracking and quantified self category, reading both the small print and<\/em> any research product makers have published to support their claims, so you can make your own assessment of their credibility, is a must.<\/p>\n Bottom line: Companies exploiting health concerns to grab people\u2019s data is not a good look but it happens a lot. (Think 23andMe\u2019s genetic testing as the classic example \u2014 a category veteran whose marketing is suggestive of some general health utility from information it will send you if you send it your spit in a tube so it can extract your DNA but whose small print disclaims its tests as diagnostics and denies they\u2019re capable of producing individual disease risk assessments. So, really, you\u2019re paying to donate your genetic data. With all the risk that entails<\/a>.)<\/p>\n A generous interpretation in cases where product makers may be promising more than they can prove is that those behind the data grabs genuinely believe they\u2019re onto something that will end up being useful in the future. They just need to do more R&D. So this is about getting (your) data to further their research. And, sure, a lot of \u2018innovation\u2019 happens that \u2018backwards\u2019 way round. But, in the meanwhile, since product outcomes aren\u2019t robustly verified, the companies behind these experiments should be making it crystal clear to users they\u2019re the guinea pigs.<\/p>\n As noted above, NuraLogic\u2019s small print concedes its smart mirror is for \u201cInvestigational use only\u201d \u2014 most likely a reference to its own product R&D, frankly. But if you failed to read the small print and have ended up the (unhappy) owner of such an unproven device and it\u2019s too late to send it back \u2014 and now it\u2019s churning out less than ideal assessments of your vital signs \u2014 definitely don\u2019t panic. The data is probably junk. But, remember, (professional) investigation is always an option \u2014 see: Rule Five.<\/p>\n When it comes to contactless health monitoring \u2014 such as the use of cameras (and selfies) for tracking biomarkers like heart rate or blood pressure, as with the aforementioned smart mirror \u2014 Shah\u2019s view is signal quality looks like a major challenge, even before you consider the overarching difficulty such products face of needing \u201clots of good data\u201d to train AIs to perform all the claimed health detections without their outputs being riddling with bias and inaccuracies.<\/p>\n \u201cEngineers are working on trying to come up with better algorithms to counter those signal quality issues etc. But I suspect \u2014 my hunch \u2014 is that these contactless [approaches] will be hard,\u201d he suggests. \u201cJust because they often depend on the ambient lighting conditions and you have less control. So the less control you have of the environment that you\u2019re measuring, generally, you can expect it will be more hard to get a reliable measure.\u201d<\/p>\n \u201cThe deployment of AI in health requires quite a few years,\u201d predicts Danilo<\/span> Mandic, a professor of machine intelligence in Imperial College\u2019s department of engineering \u2014 also pointing to problems with bias and other data quality and accuracy issues linked to recording noisy, moving targets like human bodies.<\/p>\n There are, he suggests, no AI-powered shortcuts to be had here, whatever the hype may imply. Instead, device makers will need to access proper background research and biophysical groundwork to support the development of quality measurements and credible AI models. \u201cThe problem with AI is, at least, many people just say give me data and I\u2019ll do something \u2014 it doesn\u2019t work like that!\u201d he warns. \u201cIt requires domain knowledge and biophysics models.<\/p>\n \u201cIn a way, there\u2019s no replacement for \u2018banging your head against the wall\u2019, as it were, for many years, going into uncharted territory.\u201d<\/p>\n This rule sounds super basic but it\u2019s actually fundamental. Because if a device contains any features that have<\/em> been cleared by medical regulators they will only have been demonstrated as effective and approved for the specific use-case and usage protocol. Stray from the required protocol and you\u2019re not going to get the benefit of the verified assessment \u2014 which, doubtless, got given prominent positioning in the product marketing. Indeed, you might not even get an output. (And, if you do, it\u2019s probably not going to be reliable if you failed to follow the instructions.)<\/p>\n So when, for example, you use the Apple Watch to access its ECG function and it tells you to try not to move your wrist and\/or make sure the watch is snug on your arm when you make the recording \u2014 and when it informs you the feature never checks for heart attacks \u2014 you should really listen to these essential details.<\/p>\n You also need to pay attention to usage instructions and the specifics of what such a regulator approved feature does (and doesn\u2019t) measure to avoid the risk of being misled (either by bad data or your own incorrect assumptions).<\/p>\n Again, in the case of the Apple Watch\u2019s AFib notification feature, the clearance received from the US Food and Drug Administration (FDA) is for users who are \u201c22 years and older with no prior history of the condition\u201d, per Apple\u2019s website<\/a>. If you fall outside those limitations the level of efficacy the company was able to demonstrate to the regulator won\u2019t apply.<\/p>\n Device makers offering products with regulator approved features will typically require users to read and agree to dedicated T&Cs before they can access the specific function \u2014 exactly in order to instruct on correct usage. But we all know how much attention the average tech user pays when they\u2019re faced with yet another screen of small print standing in the way of the thing they\u2019re trying to do. So it pays to be reminded to actually focus on the detail.<\/p>\n A product\u2019s standard user manual should provide details of what has and hasn\u2019t been regulated, per Imperial College\u2019s professor Esther Rodriguez-Villegas, director of the university\u2019s Wearables Technologies Lab and founder of a sleep tracking medical device startup called Acurable<\/a>. So her concise advice is: Read the manual.<\/p>\n \u201cWhat the manual will say is how that the [device] needs to be used for that [regulated feature] to be valid,\u201d she explains, noting the documentation should set out any accuracy limitations related to usage. \u201cIt might say the result should not be taken for any clinical diagnosis,\u201d she also warns. \u201cThere are actually devices that have been regulated and when you look at their user manual they do say that. So that\u2019s why none of these devices should be used without looking at the user manual. Or having a clinician recommendation \u2014 in the sense that the clinician would have read the manual.\u201d<\/p>\n Another signal she suggests consumers may be able to factor in is to look for instances where public healthcare services may be adopting consumer devices, such as to provide to patients for home monitoring. \u201cIf the NHS is using them \u2014 genuinely using them, in clinic, not [having] a clinician to do marketing for them \u2014 so if there are hospitals where this [wearable device] is how they are checking on patients \u2014 then it\u2019s a different story,\u201d she says. \u201cBecause I can guarantee you, you don\u2019t get to the NHS easily. There is a lot \u2014 a lot \u2014 of scrutiny.\u201d<\/p>\n The existence of any features on a consumer device that have been reviewed and signed off by a medical device regulator as effective for a specific purpose is a credibility signal in itself, too, to a degree. Generally speaking, if a product includes such features\/functionality it\u2019s a positive sign about the company behind it \u2014 meaning it\u2019s put in the time, effort and resources to demonstrate to an expert public authority that its product can meet a standard and perform as claimed.<\/p>\n Applications for regulatory clearance are a major undertaking \u2014 and all the work involved in obtaining sign-off can take years, plural. So it\u2019s another rule of thumb when shopping in this category to look for products with approved features. It\u2019s a mark of commitment that you may be able to use to filter between different device makers.<\/p>\n That said, it\u2019s important to remember that regulatory vetting is limited. It only covers whatever specific function the feature was approved for. Whereas consumer health tech products may combine one or more cleared features with (many) others that haven\u2019t been reviewed for efficacy by a public standards body \u2014 meaning, overall, most of the functionality hasn\u2019t had to prove its utility.<\/p>\n Consumers shouldn\u2019t lose sight of these distinctions and just assume all outputs on a particular device are credible because the FDA approved one of the features. An increasing number of products targeting consumer demand in the health\/wellness category are crossing the streams by combining some evidence-backed functionality which has gained a regulatory stamp of approval, with far more features that haven\u2019t been vetted. However legit\/fancy\/impressive these other bells & whistles may seem their output is unproven and may be totally inaccurate.<\/p>\n \u201cA lot of the wearables that are in everyday use are consumer wearables. They\u2019re specifically advertised as \u2018wellness and fitness\u2019. So they don\u2019t have to go through medical device regulation,\u201d explains Dr Gerard Cummins, an assistant professor in engineering at the University of Birmingham who also heads up its Medical Microsystems lab. <\/span>\u201cMedical regulation is there for a reason. If you\u2019re making a device and you\u2019re marketing it as a medical device it needs to have a higher level of quality \u2014 in terms of accuracy and precision and safety \u2014 because people are going to be making life changing decisions based on that. You wouldn\u2019t make a life changing decision based on something that your Apple Watch says just in general.\u201d<\/p>\n In the case of a regulator cleared feature, users may have confidence that, if used correctly, the tech can do what\u2019s claimed. But there will still be a margin of error and the possibility of the device failing to record a strong-enough signal to serve a good result every time. Still, as the user, you\u2019re in charge of controlling usage conditions to ensure the best recording possible (so we\u2019re back to the importance of following instructions).<\/span><\/p>\n It\u2019s also worth taking care not to get tripped up by certain category marketing tactics that can make it sound as if a product has undergone regulatory review when, in fact, the reference is more tangential \u2014 to only a component of what\u2019s being used in it, say, rather than the main output they\u2019re trying to sell you.<\/span><\/p>\n For example, over the last several years we\u2019ve seen a wave of startups building fitness & wellness businesses by commercializing CGMs (continuous glucose monitors). This is a type of sensing hardware that\u2019s been cleared by regulators for use in diabetes management. However what these startups are selling is something else \u2014 often their services are marketed as being for fitness\/wellness support or metabolic tracking.<\/p>\n The algorithms and software they\u2019ve developed, to process and present the signals obtained via CGM, so they can offer \u201cpersonalized\u201d insights for their (non-diabetic) users, has not been approved by regulators. <\/span>So while the presence of \u201cmedical grade\u201d sensing hardware might make these products seem credible, it\u2019s not applicable for what they\u2019re using the tech for<\/span>. The only benefit that bleeds across is that users of these metabolic tracking services can at least be confident it\u2019s safe to apply the (partially invasive) CGM sensor to their arm (again, though, carefully following instructions is a must!).<\/span><\/p>\n Bottom line: It\u2019s important to pay careful attention to what a tracking product is and isn\u2019t; and make sure you understand what it does and doesn\u2019t do to avoid being misled.<\/p>\n At the end of the day, it\u2019s either a diagnostic medical device or it\u2019s not. And most consumer tech is not, regardless of how much its marketing may be banging on about your health.<\/p>\n There\u2019s a big difference between a snapshot and a video. The same is true when it comes to the outputs of fitness trackers: Single data points served up by wearables probably aren\u2019t going to tell you anything useful, not least given the potential for errors and inaccuracies. But lots of snapshots over time can stack up to a story about what might be going on with your body that\u2019s worth listening to.<\/p>\n The way to get the most out of noisy, imperfect data from wearables is to let the tech establish a baseline for you and then monitor this to see how your metrics are moving. In other words, focus on trends, not individual data points. For example, a\u00a0resting heart rate that\u2019s trending lower over time may say something positive about your lifestyle. Ditto the reverse.<\/p>\n Apple\u2019s smart watch can do this step for you by tracking and serving up what it calls \u201chealth trends\u201d. So better devices will aim to do this leg work for you \u2014 you just need to pay attention to the direction of travel over time.<\/p>\n \u201cIf you start establishing a baseline of yourself on one of these [tracked metrics], let\u2019s say your resting heart rate, and you see how that changes over time, this is of value because if there is an error \u2014 in the calculation, in the algorithm, in the sensor, because it\u2019s cheap \u2014 the error will be there constantly. So, somehow, the fact that you have a lot of data about yourself, longitudinally, I think allows you to start building something quite useful,\u201d argues Cecilia Mascolo, a professor of mobile systems at the University of Cambridge.<\/p>\n \u201cThese devices, for as little precision they can have, they can give you years of quite fine grained data. And if you talk to a neurologist even just the information about your sleeping pattern, when you go to bed, when you wake up, is already very indicative over time. So this to me is the most tangible, simple example that I give to say, well, okay, yeah, we\u2019re still working on the precision of this stuff but the data is flowing. And if you establish \u2014 especially if you establish your own baseline and you see how this changes, what the trends are in your own data \u2014 I think this is of value. The longitudinal aspect, I think, is very important.\u201d<\/p>\n \u201cThere are advantages with longitudinal data,\u201d agrees the University of Birmingham\u2019s <\/span>Cummins<\/span>. \u201cIf you go to a GP, you\u2019re just getting a snapshot of your health [maybe] once every two years, whereas [with wearables] you have this rich, longitudinal trend data. The accuracy of individual data points may be up for debate but the trend itself would be quite useful. So you get more of an overview of how things develop in your body over time.\u201d<\/p>\n A number of researchers we spoke to for this article highlighted how the wrist, especially, is a poor place for measuring heart rate given how much we move our arms around in daily life. Blood oxygen levels measured by optical sensors applied to human skin are also notoriously inaccurate \u2014 especially on darker skin. So consumer wearables will undoubtedly capture poor quality data sometimes or even a lot of the time. But a baseline that\u2019s increasing or decreasing (or even just staying stable) may still be able to tell you (or your doctor) something useful. Whether it\u2019s a trend in your resting heart rate, total time asleep or respiration.<\/p>\n The big promise of trackers is of course exactly that they keep a record, and let users access a longitudinal view of bodily signals, which creates the possibility of spotting changes that might otherwise be too subtle\/gradual for a person to notice. So \u2014 tl;dr \u2014 lean into your trends.<\/p>\n Having trends surfaced handily in an app, which may also be encouraging the user to track various aspects of their lifestyle and activities, creates the opportunity for people to join the dots between changes in their data and their habits. This is about helping users get a handle on factors under their control or influence which might, possibly, have health implications.<\/p>\n Let\u2019s say your app notifies you about a drop in your average resting heart rate. Did you starting doing yoga to combat stress, or make the effort to clock up eight hours of quality sleep (not five), or start drinking less alcohol? Maybe you can link a directional shift in your data with a particular lifestyle tweak and make an intelligent guess at what\u2019s working for your health. That\u2019s the power of a good tracker.<\/p>\n While there\u2019s a growing number of fitness\/wellness products that propose to let you track more complex\/less well understood aspects of biology \u2014 such as metabolism, or the gut microbiome \u2014 or which claim to be able to quantify more nuanced metrics like heart rate variability (HRV), or offer to segment your sleep stages (deep, REM, core etc), the value proposition for the average consumer of trying to self-quantify at such granularity is: A) not clear-cut because so much of what\u2019s going on with human biology at these deeper levels still isn\u2019t well understood; and B) probably pointless (for now), as our current gadgets and gizmos aren\u2019t able to capture accurate enough data for such tracking to be meaningful.<\/p>\n So the rule here \u2014 for the general consumer \u2014 is that the cost-benefit analysis of shelling out for unvalidated deeper self-quantification (or, indeed, a full-body \u2018scan\u2019<\/a>) probably doesn\u2019t stack up yet.<\/p>\n On the flip side, where the value is more clear-cut is in tracking simpler stuff. Tracking basics like your movement (e.g. steps) and sleep (i.e. total time asleep) might not sound super exciting these days but such self quantification can be surprisingly powerful and positive for your health. Because we know that staying active and getting enough sleep are great lifestyle interventions that can absolutely improve our well-being and foster better health.<\/p>\n As the University of Birmingham\u2019s Cummins points out, studies have shown wearable devices <\/span>can be positive for behaviour change \u2014 because of the \u201cfeedback loop\u201d they set up between the user and the device. (\u201cThey see their activity increasing and might notice a trend in, for example, their cardiovascular health over time. So there are benefits,\u201d he suggests.)<\/p>\n Trackers that encourage users into being more active, such as by cheerleading when you hit a daily step goal, or nudging you to get more hours of quality sleep by gently suggesting you wind down in the evening, can thus have clear value \u2014 even if these features may seem pretty unremarkable at this point in time, a little shy of two decades since the advent of the humble Fitbit. But forget fancy bells & whistles; when it comes to health potential, simple stuff can be powerful.<\/p>\n That said, it\u2019s really the effectiveness of the product experience \u2014 in inspiring and driving positive change for an individual user \u2014 that\u2019s the important aspect here, since the data itself (step counts, sleep hours etc) is unlikely to be entirely accurate either. (To wit: A 2020 study<\/a> published in the journal Nature and the Science of Sleep<\/em> which looked at eight consumer sleep trackers, including the Apple Watch, Fitbit Ionic, Oura smart ring and Whoop band, found what the authors called \u201ca remarkably high degree of variability\u201d in the accuracy of commercial sleep technologies \u2014 so even seemingly simple metrics should be taken with a pinch of salt.)<\/p>\n Different individuals may prefer different wearable form factors and\/or app approaches to support them in being more active and getting adequately rested. So assessing product value is necessarily personal and subjective. But, again, the rule of thumb is so long as a tracker helps you get your steps and sleep trending in the right direction that\u2019s really the point (back to Rule Three)<\/p>\n Most wearables do also include heart rate monitoring as a basic feature these days. One researcher we spoke to questioned the value to the average consumer of tracking this metric, i.e. outside a specific pathology-based reason for doing so (and the use of a chest strap which is typically a more accurate way to measure it). But most saw value in having a read of it \u2014 including because a longitudinal view of resting heart rate can be a springboard to quantifying the healthiness of your lifestyle.<\/p>\n \u201cThe heart rate, the resting heart rate, gives you a measure of your fitness,\u201d says the University of Edinburgh\u2019s Shah. \u201cIf you\u2019re stressed, if you don\u2019t sleep well, your resting heart rate [may increase]\u2026 There are resting heart rate ranges for if you\u2019re physically very fit \u2014 like athlete-level \u2014 or average or below average. And they make sense.\u201d<\/p>\n But what about tracking more nuanced biological metrics? A growing number of fitness devices will now calculate HRV (aka, heart rate variability) \u2014 a measure of the time between successive heartbeats that\u2019s supposed to provide information on the interplay between different branches of the nervous system<\/span> (the sympathetic and parasympathetic), and the functioning of the system as a whole. Which, in turn, may provide indicators of bodily stress. (<\/span>Interestingly, the Apple Watch tracks HRV but doesn\u2019t surface the metric in the Health app overview; you have to go digging into the heart folder to find it \u2014 which suggests the company isn\u2019t convinced of its general utility quite yet.)<\/span><\/p>\n For a general consumer, it\u2019s fair to say the utility of tracking HRV is less straightforward than monitoring heart rate as it\u2019s a sensitive, highly individualized metric that\u2019s harder to interpret \u2014 whereas there are established ranges for resting heart rate that are considered healthy, as Shah notes. (And also target heart rate zones you may want to aim for for different exercises and\/or exercise intensities.)\u00a0<\/span><\/p>\n HRV scores are complicated by there being different methods<\/span> of calculating this metric, too. Some devices may continue tracking HRV during sleep, while others only track during waking hours. So different devices can produce distinctly different scores\/ratings for it \u2014 further confusing the picture for consumers when it comes to knowing what the metric might be telling them.<\/span><\/p>\n If you\u2019re a pro athlete focused on pushing your edge performance there may be some value in tracking HRV as a way<\/span> to measure stress and recovery. But \u2014 equally \u2014 such a user may well need more structured (and manual) HRV tracking, with measurements taken at specific points in their training schedule, rather than the continual background tracking that\u2019s typically offered by consumer devices that offer an HRV rating\/score.\u00a0<\/span><\/p>\n Given all the uncertainty, apps that score\/rank HRV for a general consumer risk being misleading, in our view. Again, trends may help \u2014 so<\/span> letting the tech establish a baseline for the metric and tracking any movements up or down \u2014 but th<\/span>is is still a data-point the average consumer probably shouldn\u2019t lose too much sleep over.\u00a0<\/span><\/p>\n \u201cNobody can say with authority that if your heart rate variability goes below this number or that number then something wrong is happening,\u201d agrees Shah. \u201cThese are things people are still exploring. It\u2019s not completely clear.\u201d<\/p>\n \u201cThe bottom line is that, today, [some biomarkers tracked by wearables] might not be useful for an average consumer,\u201d he continues. \u201cBut I think they could become useful once we establish which exact metrics are valuable \u2014 and, also, how do you present the information? I mean, all these are open questions at the moment. Companies like Fitbit, Apple \u2014 these are the big players and they\u2019re still innovating.\u201d<\/p>\n \u201cAll these things have potential value,\u201d he adds, pointing to the link between lifestyle factors and people\u2019s risk of developing chronic diseases. \u201cThe reason I\u2019m saying potential is there are indications from research studies that some of these [biomarkers], like heart rate variability\u2026 might have value. For example, if it decreases a lot then it might be a sign of something like your body is stressed. But the challenge we have is that conducting these type of studies is incredibly hard.\u201d<\/p>\n The category promise is that by tracking more and more bodily signals we\u2019ll generate the data that helps make the correlations that drive preventative medicine \u2014 so the hope is the tech will get better and better at nudging users towards healthier lifestyles, including by being able to detect possible health problems earlier than our reactive healthcare systems currently do.<\/p>\n But while it\u2019s now incredibly easy for consumers to get a whole host of data about themselves, if they buy into the tracking trend and strap on a wearable or two, the rock solid proof that there\u2019s value in collecting and processing all these signals isn\u2019t there yet \u2014 all we have for now are \u201cearly indications\u201d, as Shah puts it.<\/p>\n Further along the complexity axis, the value of the tracking tends to be even less clear. To wit: There\u2019s a growing number of startups offering to sell \u201cpersonalized\u201d advice based on consumers testing\/tracking their bodily fluids \u2014 to measure things like blood glucose swings or the make-up of your gut microbiome. <\/span>At this highly experimental end of the market, unless you\u2019ve got a particular, personal motivation to dig deeper \u2014 say you suffer from recurrent UTIs or have concerns about fertility and haven\u2019t found traditional healthcare routes pleasant or helpful; or you\u2019ve got a problem with weight management and conventional approaches to diet and exercise haven\u2019t worked for you \u2014 there\u2019s probably a lot more uncertainty than utility to be had from this sort of tracking.\u00a0<\/span><\/p>\n Certainly, it\u2019s vital to keep in mind that any value to such experimental tracking is, at best, speculative. So while the sampling and testing processes involved may lend some of these products an aura of scientific credibility, it\u2019s important to keep a cool head. Because when the science is so open there may not even be a confirmed understanding of the bit of biology they\u2019re offering to quantify \u2014 making any interpretations of your results, at best, informed guesswork. (While novel techniques to speedily analyze your test result remotely, rather than requiring you to mail your sample off for a lab test, could introduce inaccuracies at the source.)<\/span><\/p>\n The main beneficiary of such complex and \u2014 as yet \u2014 unproven trackers is thus likely to be the company that\u2019s getting your money and\/or data<\/span> to build a business. <\/span><\/p>\n By<\/span> serving up<\/span> \u201cpersonalized\u201d test results they\u2019re positioning themselves to flog their users quasi treatments, too \u2014 whether it be diet advice or vitamin supplements, or even a consultation with a qualified medical professional (for a fee) \u2014 cross-selling other products and services to address user-specific needs their proprietary tech has apparently picked up in your data\/bodily fluids (but without any requirement to show evidence that would convince a regulator). <\/span>I<\/span>t\u2019s a<\/span> dynamic that looks extremely convenient for drumming up customer demand. So there\u2019s an obvious risk of conflicts of interest.\u00a0<\/span><\/p>\n Anything this experimental and unproven generally falls into the \u2018caveat emptor<\/em>\u2018 category. Unless you have a specific concern \u2014 and are willing to take a punt on exploratory self investigation \u2014 you\u2019re likely better off not wasting your money.<\/p>\n If you do find yourself worrying about something your tracker is flagging up then don\u2019t be afraid to book an appointment with your a doctor and ask for an expert opinion. It\u2019s a golden rule.<\/p>\n While you might think healthcare professions get a sinking feeling when patients arrive armed with a PDF of their wearable data, the opposite is true for Dr<\/span> Hemal Shah, an NHS GP who is also medical director of the digital healthcare platform, Dr<\/span> Dropin<\/span><\/a>. \u201cMy personal thing is I encourage patients to tell me \u2014 because it does make a difference. It just gives you more information on someone,\u201d he tells TechCrunch. \u201cCertainly if you\u2019ve got symptoms, and you have whatever condition and you\u2019ve noticed the trends changing [in your wearable data], it is useful to tell the doctor because it can indicate different things going on. So more information is always better.\u201d<\/p>\n He also points to the value of doctors being able to see longitudinal data they otherwise wouldn\u2019t \u2014 at least not outside a long term hospital admission scenario.<\/p>\n Trackers can fill in gaps in patients\u2019 memories, too \u2014 offering an available record of whether you\u2019ve had decent or not so decent sleep over the last several months, for instance, whereas your own memory of how much sleep you got might be more hazy.<\/p>\n Additionally, consumer trackers have the advantage of just being there, on your person\/in your proximity, where they\u2019re in a position to record some data \u2014 say in the middle of the night when you woke up with heart palpitations and could reach for the ECG feature on your smart watch \u2014 information which, for all its potential fuzziness, is going to be better than your doctor having no intel at all, per Dr Dropin\u2019s Shah.<\/p>\n \u201cThere are moments \u2014 especially with the Apple Watch and some of the other devices \u2014 I definitely do encourage patients to take more readings,\u201d he says. \u201cIts ECG monitoring, it\u2019s not a medical device, but it gives you an indication of what your heart rhythm is.<\/p>\n \u201cI recall patients who would have episodes of feeling palpitations and their heart going really fast. And one of the things we really want to know is the electrical activity in the heart. And [trackers like the Apple Watch with an ECG feature are] able to give a snapshot of what it was like at that moment in time. And so we definitely do encourage it.\u201d<\/p>\n \u201cI certainly refer patients to cardiology specialists who do look at it and will actually make a diagnosis based on that information if it all fits the picture of what they expected to see,\u201d he adds.\u00a0\u201cIf someone has palpitations and they have them very infrequently they may last a few minutes. What you can\u2019t do is you can\u2019t get to a proper ECG machine or a medical grade machine at that time, and it\u2019s impossible to wear that machine all the time. So it allows you to recognise abnormalities on your watch and take a bit more control over things.\u201d<\/p>\n There may be a risk that doctors\u2019 time could be wasted unnecessarily if lots of people start booking appointments because of stuff their wearables got them worried about. But Shah says fewer patients than he\u2019d expected actually bring in tracker data. So far, in his experience, the rise of wearables and the quantified self movement hasn\u2019t added to the workload burden on traditional healthcare services.<\/p>\n Indeed, he flags lower usage of wearables among the elderly \u2014 as he reckons there could be greater benefit to this sort of passive monitoring for frail populations \u2014 suggesting more adoption of the tech among seniors could drive bigger public health benefits.<\/p>\n What about the risk of all this tracking triggering unnecessary health anxiety for some users, such as people who may be prone to hypochondria?<\/p>\n \u201cI think it does sometimes add a bit of anxiety when they see something abnormal [in their data],\u201d he responds on the anxiety point. \u201cBut I also believe that when they speak to a clinician that anxiety dissipates, or they get taken seriously, or they have further tests and then it\u2019s all absolutely fine.\u201d<\/p>\n So, again, the rule here is simple: If something in your data is making you worried, grab a download and take the concern to your doctor.<\/p>\n In the rush for consumers to take a personal interest in their health and shell out for tech to track their bodies and activities, it can be easy for people to forget that the data being captured, stored and processed \u2014 and potentially shared with others \u2014 is highly sensitive personal information.<\/p>\n \u201cThe signal is very, very private,\u201d warns Cambridge University\u2019s professor Mascolo. \u201cWe think imaging is private. But what about audio? What about your heart? Your heart signal is a unique fingerprint. So if we start sending everything back to central server, that\u2019s a bad thing. It could lead to unwanted exploitations.\u201d<\/p>\n Consumers should carefully consider who they may be handing their private health data over to before they strap any device on. Some companies may be considerably more credible than others when it comes to claims of respecting privacy. Look for clear and prominent statements about the personal information they intend to collect and what they will do with it. Companies whose privacy policies aren\u2019t clear, or leave you confused, are best avoided.<\/p>\n Where your information is being stored and processed may also be important, given there can be big differences in legal protections for personal information depending on where in the world the data is being handled.<\/p>\n Also consider business model. How is the company behind the tracker\/service intending to make money? Does their approach look sustainable? Some free-to-use period tracking apps, for instance, have been found trying to monetize their software by plugging user data into the online advertising ecosystem<\/a> \u2014 which is obviously horrible for user privacy. (For example, a few years ago, period tracker app, Flo<\/a>, settled with the FTC after allegations it had shared sensitive user data with ad platforms despite promising users it would keep their information private.)<\/p>\n Even if a tracking product maker\u2019s business model looks legit, given the sensitive nature of the data routinely being collected and processed you should proactively consider the risk that your information could be breached and what an accidental leak might mean for you. (The breach last year <\/a>of ancestry data held on millions of users of genetic testing service 23andMe<\/a> is instructive of the high stakes for getting involved with quantified self tech.)<\/p>\n Ultimately, consumers wanting to tap in to tech and services in this health-adjacent fitness & wellness category will need to weigh up the utility they believe they may gain from whatever tracking is offered with potential risks to their privacy if their information gets misused or is not kept secure.<\/p>\n Business models that are selling self quantification\/tracking itself, either in the form of hardware and\/or a subscription service, may generally look more credible than those which rely on offering a free product to scale usage and amass data. But lots of companies in this space are also using customer data for product development and wider research and, given the sensitivity of the information involved, privacy questions and considerations come with the territory. So it\u2019s important to stop and proactively consider the risks.<\/p>\n \u201cThe problem is that in order to develop algorithms, a lot of these companies are actually collecting personal data,\u201d warns Imperial\u2019s Rodriguez-Villegas, raising concerns about the extent of data collection by consumer devices who may be hoping to use the info to develop a medical device. \u201cIn order to use the device, the first thing that is requested from people \u2014 I mean, this is even after having paid for the device \u2014 is to click on the box saying they \u2018agree\u2019 for this data reuse\u2026 [that their] data could be shared with partners or with collaborators.\u201d<\/p>\n Even if you\u2019re the sort of person who\u2019s happy to donate even sensitive personal data for speculative commercial research \u2014 say, for the possibility of future upside for humanity if your data can support research into certain medical conditions and diseases \u2014 at least look for product makers who make it clear how they want to use your information; and, ideally, ask for consent to use your data for research.<\/p>\n Companies operating in the health\/wellness space that make an upfront pitch for data for research, and provide specific details of what and how they want to study, are more likely to be doing credible research, too.<\/p>\n In recent years, a lot of femtech startups have popped up, touting novel wellness products which rely on women\u2019s data to power predictive algorithms. Many make a point of pitching potential users on helping them close the female health data gap, given how medical research has historically focused on male subjects. You might feel your values align with such a mission \u2014 so great, it could be a win-win. But, even so, always read the small print and check you\u2019re happy to support the kind of research they say they plan to do.<\/p>\n An interesting confluence of factors we\u2019ve touched on in our Six Rules could end up shaping the next big evolutionary leap in wearables \u2014 namely this trio: Accuracy issues; privacy concerns; and a push for greater efficiency of biomarker data processing, including to allow for more powerful software to be housed in smaller physical devices we may carry on our person. (And \u2014 yep \u2014 this is where the promise of health-monitoring \u201chearables\u201d (or \u201cearables\u201d) comes in.)<\/p>\n Imperial College\u2019s Mandic, who says he was the first to outline an \u201cin-the-ear recording concept\u201d (in a 2012 paper<\/a> on \u201cuser centered and wearable brain monitoring\u201d), highlights deep learning work he\u2019s undertaking to extract \u201cclean\u201d biomarkers from \u201cvery noisy\u201d environments using models that are \u201ccomputationally cheap to run\u201d, as one of his papers<\/a> puts it. So the hope is for greater processing efficiency will allow for smaller types of devices to become trackers.<\/p>\n \u201cThe \u2018corr-encoder\u2019 and other models we do and are proud of don\u2019t require much power,\u201d he tells TechCrunch. \u201cThey can be implemented on microcontrollers, on edge devices, or we can stream data using Bluetooth and process them on smartphones \u2014 we don\u2019t require huge servers.\u201d<\/p>\n \u201cWe need to move away from those brute force approaches [with AI] \u2014 \u2018just let me add more layers in my neural networks\u2019 \u2014 to basically thinking more, including domain knowledge and working towards smaller, much smaller, models which can work on microcontrollers \u2014 even on the earbud,\u201d he suggests, adding: \u201cMy current model basically works directly on the earbud.\u201d<\/p>\n Another interesting possibility here is that privacy concerns and (AI-aided) efforts that are gunning for more efficient signals processing could conspire to drive each other. The University of Cambridge\u2019s Mascolo believes privacy concerns could encourage development of commercial AI models that are designed to live and work on the user\u2019s device, avoiding the need for sensitive health data to be uploaded to the cloud.<\/p>\n \u201cWe can do things on device. But we need to find the right business model for this,\u201d she suggests. \u201cIt might be that the business model is privacy \u2014 and new devices allow you to to do this. Companies are possibly still exploring if there is a privacy-based business model.\u201d<\/p>\n \u201cThe first step would possibly be the generation of [AI] models from data that is collected in large scale and then perhaps the pushing of new products that have these models scaled down to on-device apps on your phone and they use your own data but the data is not sent any further. I think that is very achievable,\u201d she adds.<\/p>\n The privacy advancement in this scenario would mean users don\u2019t need to bare their raw biological signals to any third parties; the biomarker processing could just take place on their device. (And Mascolo also flags the potential of machine learning techniques like federated learning to further support privacy-preserving processing of wearables\u2019 data.)<\/p>\n There would still need to be a pool of users willing to share data for models to be developed in the first place \u2014 but she likens this to how, during drug development, tests of novel pharmaceuticals may cause side effects in test subjects that can be avoided in the final product. (So once a fine-tuned AI model is put on device the wider user population wouldn\u2019t have to submit to the \u201cside effect\u201d of losing their privacy.)<\/p>\n \u201cI think we\u2019re getting to a stage where privacy and efficiency are driving the solutions that we want to see,\u201d she adds.<\/p>\n Returning to hearables, Imperial\u2019s Mandic reckons health-monitoring in-ear devices could be a commercial reality in as little as \u201ctwo to three\u201d years\u2019 time.<\/p>\n \u201cIt\u2019s been 10 years since [my paper]. So for the first five years I was struggling to convince people that you can record from these canals. The next five years\u2026 many companies [were] set up and failed\u2026 The time has come now that things are bit more mature so I expect myself and maybe some bigger players to come up with something,\u201d he suggests.<\/p>\n \u201cClearly, if the current wearables were that good then we wouldn\u2019t be looking for anything else but they\u2019re not,\u201d he goes on, adding: \u201cWith the emerging e-health [movement] we need reliable devices which can be used to monitor people at home.\u201d<\/p>\n In-ear devices have an edge over wrist- or finger-based wearables for accurately measuring certain bodily measurements, per Mandic, as the ear canal doesn\u2019t suffer the same \u201cvasoconstriction\u201d effect as outer skin \u2014 a phenomenon that can cause accuracy problems and bias for optical measurements performed on the wrist or finger (so bad news for smart watches and smart rings).<\/p>\n The head also offers a relatively stable location to perform measurements of biomarkers vs the arm\/hand, which are more likely to be moving around a lot. Plus, as Mandic points out, you can wear earbuds for an extended amount of time. (Indeed, many consumers already do.) So in-ear-based tracking looks exciting for getting better signals consistency from wearables.<\/p>\n The University of Cambridge\u2019s Mascolo is also working in this area. Her research includes looking at \u201cearable\u201d tech for fitness and vitals signs monitoring \u2014 using an in-the-ear microphone to monitor activity and heart rate in order to proxy VO2 Max. (<\/span>VO2 Max is a measure of cardio or aerobic fitness which some existing consumer wearables, like the Apple Watch, already offer to estimate, based on tracking users\u2019 heart rate and movement. Although, as with other more nuanced metrics (like HRV), the accuracy of current-gen trackers\u2019 VO2 Max features is questionable.)\u00a0\u00a0<\/span><\/p>\n The gold standard VO2 Max test requires a person to attend a specialist center and undergo intense physical exercise while wearing a mask hooked up to a machine that measures how much oxygen they\u2019re breathing out vs in. So the test is usually only undertaken by athletes. Whereas wearables offer the chance for many more people to track their cardio fitness. But it\u2019s only going to be really meaningful if the accuracy of these proxy measurements steps up.<\/p>\n Beyond interesting-looking potential for in-ear monitoring, and the possibility of developing more privacy-preserving tracking, there\u2019s a notable trend already for consumers to get involved with more intimate\/invasive types of tracking. The University of Birmingham\u2019s Cummins, for one, predicts further growth here \u2014 suggesting we\u2019ll see more activity around these sorts of \u201cchemical sensors\u201d in the coming years.<\/p>\n He posits that the adoption of CGMs for fitness\/wellness (and\/or metabolic health) tracking is \u201cthe first sign of a shift\u201d that\u2019s taking consumers beyond wearables with \u201cphysical sensors\u201d to products that feature \u201c\u2018chemical sensors\u2019 in the devices\u201d. The goal is \u201chaving a richer data set on what\u2019s actually happening within the body\u201d, he says. \u201cNot just looking at heart rate, lung capacity, activity \u2014 it may be looking at glucose spikes, or cortisol for detecting stress, things like that.\u201d<\/p>\n So, on some level, the demand for chemical trackers may reflect a sense of frustration with the limits of what sensing wearables have been able to read through our skin.<\/p>\n His own research spans so called \u201cingestibles\u201d \u2014 novel devices that aren\u2019t worn on the body but swallowed; allowing for built-in sensors to get a read on what\u2019s going on internally, in a user\u2019s digestive system. It\u2019s a concept that could also push tracking to new depths. (And even, potentially, be used to deliver drugs in a more targeted way \u2014 moving from tracking to treating.)<\/p>\n \u201cThe capsule endoscope would really be the first ingestible but there are limitations to those devices, in terms of the quality of the diagnosis, because you\u2019re just using an optical camera. So what we\u2019re doing in my lab is we\u2019re looking at either trying to improve the accuracy of these devices by integrating additional sensors,\u201d says Cummins. \u201cWe\u2019re looking at different form factors that you can swallow that would give you data about your gastrointestinal health \u2014 or they could potentially be used for targeted delivery of drugs as well.\u201d<\/p>\n More generally, he suggests wearables and fitness trackers are at an \u201cinflection point\u201d \u2014 with some devices starting to be used in hospitals and traditional healthcare settings, working towards the big vision of preventative health. He also flags research that aims to see if longitudinal monitoring can help detect the point at which a healthy person might be developing an illness. Plus he emphasizes the foundational trend of consumers proactively tracking themselves in a bid to make positive behavioral shifts and shrink their risk of bad habits leading to health issues. But while the direction of travel for tracking looks clear, further research is needed to stand up the tech\u2019s preventative potential.<\/p>\n \u201cThere\u2019s an awareness amongst the clinical community that wearables are here to stay,\u201d Cummins suggests. \u201cThey\u2019re going to be useful down the line \u2014 and it\u2019s a case of just how to integrate them into the current clinical pathways or adapt clinical pathways to use wearable data. So I think there will be a change coming in how they\u2019re used in clinical decision making. But at the moment, you wouldn\u2019t use them, by themselves, for that.\u201d<\/p>\n \u201cLet\u2019s put it in an optimistic way,\u201d adds Imperial\u2019s Mandic. \u201cIt\u2019s good to have an awareness about the possibilities of good wearables \u2014 and for the public to get used to that idea \u2014 until we have a class of variables which are going to be \u2018bulletproof\u2019 and rock solid, including being granted clearances as medical devices.\u201d<\/p>\n<\/p><\/div>\nRule One: Mind the hype, read the small print<\/h2>\n
Rule Two: Pay attention to the instructions<\/h2>\n
Rule Three: Focus on trends, not data points<\/h2>\n
Rule Four: Tracking the basics can be powerful<\/h2>\n
Rule Five: Worried about your data? Take it to your doctor<\/h2>\n
Rule Six: Don\u2019t forget about privacy<\/h2>\n
Looking ahead: Where next for wearables?<\/h2>\n