AI and wearables helping track dementia risk in India
Posted by data_optimizer in Cognitive & Nootropic - 1 points, 4 comments.
https://today.usc.edu/how-ai-and-advanced-computing-are-accelerating-alzheimers-research
There's a piece out of USC about how AI plus wearable devices are being used in India to study speech patterns in older adults, as part of the LASI-DAD project looking at population-level dementia risk. Basically they're using small wearables to capture speech and other signals, then running machine learning on it to flag early cognitive decline before someone walks into a clinic with obvious symptoms.
What caught my eye is the scale of it. Most cognitive studies I've read about are tiny, 50 people here, 100 there. This is trying to do something at population level in a country where dementia is massively underdiagnosed, which honestly makes sense to me. Early detection is the part nobody talks about enough. By the time someone gets a formal diagnosis, a lot of damage is already done.
It's still research, not a product you can use tomorrow. But if speech biomarkers actually hold up, it could change how screening works. Does anyone here actually track their own cognition over time with anything objective, or is it all subjective notes?
Comments
- grinder265: Speech biomarkers is actually a smart move ngl, way easier to scale than blood draws or imaging for something like this. Catching stuff pre-clinic is the whole game. I don't really track my own cognition objectively tbh. I log sleep and HRV pretty consistently but cognition stuff feels like it needs a baseline first which i dont have from 10 years ago to compare against. Did anyone here actually start doing any of the cognitive test apps or anything before a problem showed up, or is it all kind
- data_optimizer: Yeah the baseline problem is real, i struggled with the same thing when i first got into this stuff. I started doing a couple of the apps (brainhq and a dual n back thing) mostly just so i'd have something to compare against later if anything felt off. Not saying its super scientific but at least its something vs flying blind đŸ˜…
- reads_labrat: Yeah speech markers are low‑hanging fruit for big studies. I tried a few brain‑game apps yearly just to see trends, but without a solid baseline it’s hard to know if a dip is real or just a bad day. đŸ”¬
- data_optimizer: That's exactly the issue I keep running into with my own tracking too. I log reaction time and some cognitive benchmarks monthly but yeah, one bad day and you spiral yourself into thinking something is wrong when really I just slept poorly or had a stressful week at work. The baseline problem is real. What's helped me is using multiple metrics together instead of relying on one number. Like pairing reaction time with mood scores and sleep quality. If they all dip together, that's a signal worth
Community discussion - research and educational context only. Not medical advice.