This is fascinating (and sad), but it makes me wonder, could police use your fitness tracker as a sort of poor-mans polygraph? i.e. interview a subject normally, and then get a warrant (or not get a warrant) to view the online data and correlate them after the fact? Wikipedia mentions "blood pressure, pulse, respiration, and skin conductivity" as parameters for a polygraph [1], and while fitbit only supports 2/4 right now, the goal of fitness tracking is to have all of those. Whether or not polygraphs even work is another question, but being able to do them secretly and without consent is scary none the less.
I can sort of see that as a concern, but... why on earth would you wear your Fitbit to a police interview? If I ever got hauled in for questioning by the cops, then guilty or innocent I would definitely not be wearing something that tracks my heart rate!
In any case, I don't think this would be admissible evidence. As far as admissible evidence, polygraphs are only just scraping by on the skin of their teeth, and wearable devices don't come close to their level of accuracy/depth. The main reason these devices can be sold with none of the regulation that accompanies "real" medical equipment is that the companies state up front that "to be used for general fitness only, not intended for use to treat any particular disease", etc., and their accuracy reflects that. The heart rate measurements of Fitbits and other wearables aren't awful, but they're sufficiently inaccurate that any decent lawyer should be able to get that thrown out without question. Edit: Not to mention JTsummers' point about there being no baseline/control measurement on this data, if it existed. Definitely not admissible.
If you want something to be worried about vis-à-vis wearables and the privacy of health data, worry about your health insurer getting a hold of it and jacking up your premiums for not taking enough steps every day. That, to me, is the real worry here, not the cops.
But perhaps you were wearing it while murdering your neighbour. By itself it should be circumstantial evidence at best, but it could possibly help pointing the police in the right direction.
It'd almost certainly require consent or a warrant to get the data (in the US). But it would have a problem that polygraphs (presumably) attempt to handle. You have no baseline questions and timeline. A stressed individual (being interviewed by the police would be stressful) will look different from an unstressed individual. Without the control questions at the start, you have no basis to draw conclusions from.
I don't think that this will happen, but one can imagine solving this problem by subpoenaing your past FitBit data and social-media record, and correlating the two (by time stamp, which, if maybe not now, then I am sure eventually will be sufficiently fine grained to allow this). I'd imagine that it would be far more accurate than whatever calibration they do now.
Yeah. Protomyth made the point about aggregating biometric data with other data (shopping, akin to the Target pregnancy notice letter, browsing, etc.). That could be effective, if the sensors and social media, shopping, other data were available.
I know everyone (or most people) get hunches about what's going to happen by observing other people's behavior. Whether it was predicting a break-up, guessing that someone was pregnant, or in a new relationship. The glut of data available now would likely reveal a lot if we allowed it to be aggregated.
I guess I shouldn't have thought so narrowly about just biometrics (and heart rates, in particular).
My understanding is that polygraphs absolutely DO work, but they're not that reliable, and they don't work correctly for everyone which is why they're inadmissible as evidence in any decent jurisdiction. Worse, people who are good liars and people who have trained for the test can beat it, and those are frequently the people you want to catch with the test, so all you're doing is scaring and harassing the honest people, because they get the opposite problem, that they fail the test because their nervousness shows up as lying.
So, the fundamental premise of the machine, that these physiological factors frequently change in response to peoples' thought processes brought on by particular lines of questioning, is true, but it's nowhere near reliable enough to be used as any kind of test. It'd be like a police radar gun that gives inaccurate readings for 20% of the cars, leading to both lots of unearned tickets and lots of speeders evading a ticket.
I think it's pretty widely accepted that they don't work as a "is this person telling the truth about _____"
They're fine if you don't mind false positives (true statements flagged as dishonesty). This is largely why they're still in use e.g. for hiring with the CIA. It's much better to filter out many otherwise fine candidates than it is to let someone in who would lie at that point in the interview.
The polygraph is shit for chirst's sake. Clench your anus and it will mess up it up. Polygraphs have time and time again have been shows to be ineffective, and yet people still keep advocating for them.
They are still useless things built on pseudoscience considering what question they're attempting to answer, and I find their use in court cases a blight on an otherwise advanced society.
[1] https://en.wikipedia.org/wiki/Polygraph