See that’s the issue I am hoping to solve, how to know if something is wrong, there would in theory be no click on an action from an actionable notification, or a reply to a text message/twilio/telegram etc. It’s something that I feel could be a benefit to someone out there, that hasn’t thought of this as a way to keep someone safe.
idk, call me crazy, but I think it’s a neat idea.
Glad I’m not the only one!
But I don’t discount the idea of using HA to make to check for “signs of life”.
I’m just saying that the part where you want HA to take some action when someone doesn’t reply to a text or phone call is not really needed and I wouldn’t know what it would do anyway.
If you are concerned that you might not know that the text or call was made then you could have HA also send you a notification in the same automation that there was a phone call/text made and you would then know that you should be expecting a response. If that response doesn’t come then you need to be the one to take the appropriate action.
Waiting for the response to an actionable notification is done with a wait_for_trigger which does have an optional timeout.
Not sure how familiar you are with actionable notifications, so here’s a blueprint that I worked on that could be modified to suit your needs with the above-mentioned timeout.
Would I use Home Assistant to monitor a critical task like a person’s well-being?
Only after implementing a boatload of fail-safe, fault-tolerant measures to mitigate outages and minimize false-positives/negatives. Otherwise, no.
If it was only used as a “just in case because no one ever knows might happen to even a healthy person” situation then it would be fine.
But if you knew that someone was actually at risk because of known risk factors then I agree.
I think things got a bit confused here.
the OP was about sending a notification to someone outside of your own household which is monitored by HA. presumably that person wouldn’t have a way for them to interact with your HA and respond to a notification within HA itself.
Like in the apartment complex situation someone had to “notice” that someone else wasn’t in a “normal” condition (hasn’t been seen, etc) and someone outside the household would check on them.
In this case HA would do the “noticing” and contact someone to check on you.
When I added that last bit above about notifying you if a call was made so you could expect a response I changed the assumption that the person being contacted is outside of the monitored household and not linked in to the monitored persons HA environment.
in one situation you are being monitored by HA. In the other you are the one HA is calling about the monitored person.
…And me reading the above again I’m not sure that I have done anything to minimize the confusion tho. 
So, yes, if both parties are within the same HA environment and can both send and receive notifications, etc from HA then a wait_for_trigger from a response to an actionable notification would work.
But if the contacted party is outside the HA environment then there’s not much that can be done if the monitored person doesn’t reply to a call back check-in aside from someone going to physically check on that person.
I use a similar system right now.
My wife stays home alone quite a bit when I’m at work and she was concerned about what she could do if something happened and she couldn’t get to her phone to call someone (911, one of the family, etc).
So I set up my Amazon Echo speakers to listen for a key phrase (Alexa, I need help) which sets off a routine that will have HA call and text everyone in our family (me, my wife, our kids) using Twilio to say that someone at our house needs help and then they can take appropriate action (come over, call 911 on our behalf, etc).
Is it like calling 911 ourselves? No. But it is a backup in case the situation is such that a real 911 call can’t be made. I have recently learned that Twilio can actually make direct 911 calls but I haven’t figured out how to do it yet or if the local emergency services would actually respond to a robo-call. 
But if no one answers the call or sees the text then there isn’t much HA can do about it.
Fair point. I think we were just answering different levels of the question. Is it in the realm of possibility? Yes. Is it applicable to something like an apartment complex? No. At least not without getting creative, like setting up a REST API, sending a link with a POST request to the person, and checking if they clicked it with a REST sensor after some amount of time. Again, within the realm of possibility, sure. Realistic and reliable, probably not…
I think I’d prefer a buffer between an automation and a 911 call
This is probably another instance of an interesting thought experiment, but not necessarily something that anyone would want to implement at the moment:
I’m torn on the subject but I think I agree. Which is why I haven’t really pursued it yet.
So far tho I’ve had my call system in place for several months and I haven’t had any false positives yet. So I’m fairly confident it would be OK. But you never know.
OTOH, what would be worse - a false positive 911 call (as long as it only happened once) or an actual emergency with no way to call 911?
Like I said…I’m torn.
To each their own, but for a healthy individual I think the probability and consequences of a false “911” call probably outweigh the benefit of the small advantage you’d get in a real emergency.
I’d end up setting the delay/refresh interval long enough that if I hadn’t managed to get help through some other means, I’m probably dead anyway, in which case what’s the rush.
yup…still torn… 
If in doubt take the action that requires less effort. 
exactly. which is why I haven’t done any more with it. 
I understand 100%
This is a budding/brewing idea, I don’t know who all could use this, maybe someone that is depressed and wants a back up life line in case something happens, maybe you live alone, and hell anything can happen when you live alone, this acts as a safety net.
Not in its current brainstorming system would I even think to put this into implementation as a single life line.
This is why I want ideas and chat about it, I want to know if this is something doable, and then branch out from there. Maybe there are new integrations that could be added. Only way to make something new is to brainstorm, and we have some amazing minds here.
Exactly! This is the boss that hasn’t seen you in a few days and not answering your phone calling to have you checked on (and that legit happened once, it was over a week, thankfully the guy just went on a bender, when he opened the door, you heard the pile of cans shifting lol)
Agreed, in it’s current state I wouldn’t want it calling 911, if AI in Alexa for your example could learn sounds, example I hope it never happens, glass breaking, mixed with time/motion sensors, then maybe. That is just an example, maybe even Alexa hears an unknown voice. Either way, what you have now is what I am thinking to start, till we can branch this out (if there is enough traction, I am not a programming wiz anymore, I used to dabble, but I am sure not in the position to write a whole integration for something lol)
I’ll check it out, thanks!
So this brings us back to the idea of how can this be used, and can it be done. Can the single person living alone with depression, set up a back up plan/system to keep a check on themselves? I mean hell, they don’t even have to be depressed, if you live alone, anything can happen, its a part of life. Can they set something up as a back up life line?
Apartment complexes have so much red tape and legal stuff to deal with, it would be tough for them to set it up, but hey, if you are clear and up front about the tech that is in your apartment, I am sure there are legal ways to set this up, and you never know, this could be the next million dollar idea!
This is why I wanted to start the discussion though, since I am sure there is someone out there, or many people out there that may want something like this, if they read/see.
Some of the best ideas in the world started out on a napkin, maybe this is the napkin, and we are all just starting the brainstorm?
For the implementation side, I’d go with something like a “haven’t seen you in a while” entity. It would be set to “yes” or “true” when HA detects (more on that later) that you haven’t followed your normal pattern in a while. At the same time, it would send you a text notification: “Hey, you OK?”
If you are, you’d go into HA and turn the flag off. If not, after a set time HA would make the notification. I’d go with text here, too. I’m not sure a voice call adds much value. You could text one person or a dozen.
As for the detection piece, that depends. In my case, if I haven’t tweaked something in HA for 24 hours, something is very much not normal. If you have presence detection, not coming home for an extended period might work, if you’re the type who keeps to a schedule. Or maybe you haven’t logged on to HA (or some other system) for a certain period. Or your location isn’t where it’s supposed to be (like not at work on a work day.) Even something as simple as the lack of light switches and door sensors triggering could indicate a problem. The possibilities are endless. But in all cases, you would want to keep it as simple as possible.
I don’t think the OP ever suggested this be used commercially for an apartment building, only that the idea came from having to do welfare checks at one. Obviously the solution would have to be finely tuned to each individual using it.
I did think it could be used in an apartment community that has a special need for something like this, did the resident not leave their home for 36 hours? Could be a sign of trouble, maybe not for the average apartment community, but let’s say a drug rehab clinic? Nursing home?
I do agree it would have to be tailored for each person/situation, but can something general be made, that allows enough tweaking what I am thinking. Could an integration, or an HACS addon be made that can allow enough customization to be made that we can make something for the general public? Since that was the point of this whole thing, trying to make something for the public that can be customized. I love the dead man switch, but that should/could be just 1 option in the greater addon.
I personally would be using motion detection, I routinely go to the kitchen for more water, and or move around on my couch while watching TV (it’s right above me, covers a huge chunk of my house) so even if I fall asleep, the bedroom one would catch me moving when I went to bed (as an example) so there would be almost no chance of a false alarm. Using 2-3 points of motion detection could be an option in the addon.
I know there is a way to create a php script (there used to be) with a home server, that you could send email and I believe text someone, so HA could in theory call that script with the emergency contact information and notify them that way. That could be another part of the add on.
Oh here’s a better one, is there a way to make a google home/alexa call someone’s phone and speak? I know they be asked to call, but is there a way to call directly? Maybe that can be created?
I just want people to be safe.
I have the idea, but not the skills.
I have google, but good to know Alexa has that skill, is there a false alarm though if you drop a glass? Does it get confused?
That wouldn’t be good if it did. So I would be curious about it.