A study used data from phones to predict depression and personalized treatment | Hacker News Reader

A study used data from phones to predict depression and personalized treatment

ucsdnews.ucsd.edu

23 pointsrustoo14 comments

Not entirely sure how personalized it is to say a person should exercise and go to therapy when they are depressed.

I'm not sure how you got that from reading the article. "As an example, exercise and daily caffeine intake emerged as strong predictors of mood for one participant, but for another, it was sleep and stress that were more predictive, while in a third subject, the top predictors were brain function and cognitive responses to rewards."

Basically I agree: most advice for most people is going to be very similar. However I can see added value in enhancing engagement with that advice by, for example, linking it to that person's specific behaviour.

Abstract example: If I told you to exercise optimally, you'd snore. If I said that having analysed your behaviour over rather last month, you'd benefit from two additional HIT sessions each week... I guess you'd care more. Even more so, if it told you it'd help you fit into your bridesmaid dress for the upcoming wedding... you see where I'm going with this.

Most advice is generic, but compliance is poor, and that gap is where more of the untapped benefit is to be found.

I remember an article from Lifehacker I think back in the day where the author wrote that they could arguably infer what mood their friends were in from their Last.FM scrobbles. For the uninitiated, it's just a way of tracking music you've listened to.

While I don't think that's quite accurate in hindsight, I do often wonder just how many indicators there are that could be picked up and reflected back to the user, in order to see if their otherwise passive behaviour is something they might want to be aware of.

A common example is usually number of screen unlocks. I even find myself opening Twitter multiple times without realising or identifying it as "Oh, this is me procrastinating"

Tinfoil hat take: such variables already exist for all users in the backends of most major tech companies, particularly ones in the advertising business.

The ultimate targeted ad is delivering it at the right time to the right person. If they can infer from your usage patterns the way you feel at any time, or better yet make you feel a certain way, then you're more likely to click on that ad and make a purchase.

I'm not sure if the targeting is this aggressive yet, but in this race to extract as much value from attention it wouldn't shock me if it were.

The Social Dilemma shows a sensationalized version of this, but it is concerning to think of the implications of all this data collection.

"The one-month study collected data from 14 participants"

This is a preliminary study. A much wider group of participants is necessary, from a range of backgrounds and mental health status, before it could be said to 'predict' or 'personalisze' anything.

It's good to look at this as a topic.

I am sure that my behavior changes when I am depressed and that I would express it first via my phone - after all it is at my fingertips.

While I would say that this is very limited in its findings, I hope that there is some further study that expands on this.

Checking HN more than 20x per day? Oh, you must be depressed!

What if the cure was to not stare at a screen all the time?

Pretty obvious, despite the gains in other areas, technology is having a significant and largely detrimental effect to the mental health of all generations. Not even to mention that the cognitive algorithms on most web platforms can be manipulated for greater interaction (addiction?) or worse weaponised for political ends.

Teen suicides are up, self esteem for millions of people are through the floor because of ego battles on Instagram. Porn addictions on the rise, with a new generations thinking pornhub is normal.

The best part of this is that they registered an invention to fix the problem.

Here’s my invention: turn it off when you don’t need it.

After reading The Age of Surveillance Capitalism and its many examples of how optimistic psychological studies have been misused by the ad industry to make better ads, I can’t help but think about how this may fall into the same category.

The models used in this study probably measure your mental health along one or more axes that start on “perfectly fine” and ends up at “problematic”. Knowing where someone falls (within the range of what’s ethically acceptable to push ads to) could be very profitable to Facebook and Google. Scanning for mental health problems could also provide them with an excuse to get users to hand over their data.

I thought the same. Happy people just don't buy that much stuff. Depressed people trying to self-medicate through shopping, on the other hand, that's an advertisement gold mine.

The anime Psycho Pass has mapped all this out in full, better watch that.

In short mood detection is ultra refined and remote, via cctv type installations. Probable criminal activity is intercepted before action is taken and pre-criminals treated for their mental health.

How reliable is a one month study of 14 people when depression is a complex disease/set of symptoms that seem to typically last much longer?

uh, this is fucking horrifying.

Another steps towards human-machine symbiosis.

It is inevitable.

Reply on news.ycombinator.com