4 ms·
this is wonderful, thanks for publishing this. I have a plan to launch something somewhat similar to this one day and will test it on myself (as I deal with sev
by dchuk 14y ago
this is wonderful, thanks for publishing this. I have a plan to launch something somewhat similar to this one day and will test it on myself (as I deal with severe depression) to see if it really does help things in the long run.
Here's a few suggestions from my list of things I want to add to my idea:
-Allow me to add tags to my mood checkin. Over time, patterns will emerge for both positive and negative mindsets (hopefully)
-make the checkins randomly spaced out throughout the day so people don't just start ignoring the app after a while
-try and extrapolate some compound meaning from the data, something like "it seems when you don't get a lot of sleep then see this person, your mood drops"...can be tricky to do, but could really add to the long term value.
Good luck!
- Harj 14y agothanks for the suggestion, really appreciate it. - What kind of tags were you thinking? I've noticed myself starting to go in the direction of tags via the notes I add when I'm tracking a mood. - Yep that's how notifications will work. - This is the hardest and most valuable part of working on something like this. I'd love to make meaningful suggestions based on the data, am thinking of the best way to do that.
- dchuk 14y agoIn terms of tags, my idea was to just allow people to drop any tag they want in (similar to tagging a post in wordpress) and have the system autocomplete their tags when they're typing. That way, if someone is constantly tagging "mom" and their mood is low, after a while, you'll be able to positively associate the tag and the moods. In regards to your last bullet, yeah, compound insights are tricky. I still don't really have a solid system of doing these, but essentially you want some sort of rule-based system to pull it off. Please let me know if you find any resources on it, I'd love to help you with what you're working on.