Yeah, voice-to-state mapping is basically the whole bet. I've been trying to figure out how to actually test it — plan right now is to run the voice signals against PHQ-9 and NASA-TLX in the beta and see if they line up. Same day: take a voice snapshot, have them fill out the survey, check the correlation. I'll report back if it goes anywhere interesting.
The interpretation question is the part I'd spend time on before becoming too confident in what the correlation is actually saying.
About
It started with my back. I have chronic back pain, and I wanted to find the patterns behind it — how activity, sleep, stress, and whatever else lined up with the bad days. So I built a daily tracker.
5 Comments
What stuck with me wasn't the on-device engineering story.
It was the decision to move from tracking the outcome you cared about to tracking a signal you believed could stand in for it.
That's one of those decisions that can look validated much earlier than it's actually understood.
The tricky part is that several very different explanations can produce the same early feedback.
That's why I'd spend as much time pressure-testing the interpretation as the implementation.
Yeah, voice-to-state mapping is basically the whole bet. I've been trying to figure out how to actually test it — plan right now is to run the voice signals against PHQ-9 and NASA-TLX in the beta and see if they line up. Same day: take a voice snapshot, have them fill out the survey, check the correlation. I'll report back if it goes anywhere interesting.
What makes this tricky is that a strong correlation can reduce uncertainty while still leaving the underlying interpretation unresolved.
That's one of those situations where the next decision can look validated long before it is actually understood.
I wouldn't try to unpack that properly in a thread.
If you're curious, drop your email and I'll put together the tighter version.
Appreciate the depth — sabber at healthos dot live, would love to read it
Perfect.
I sent you a note by email.
The interpretation question is the part I'd spend time on before becoming too confident in what the correlation is actually saying.