Can This AI Hear Alzheimer’s on the Phone?

[@SciShowPsych]

Go to ground.news/scishowpsych to download the Ground News app. Get 60% off the Vantage subscription for unlimited access to the most advanced news analysis features. Offer ends 12/31.

[@Bibibosh]

Can someone explain what the hell gravity is. Like its a force that is always on; can't be turned off. It basically affects everything. Even nothing can't stop it.

[@ThisIsFiftyWithLil]

Oof, I hate Alzheimers SO much. Anytime it occurs to me that my grown children are at risk, immense sadness overcomes. Right now, I have not heard of any cousins (26; ages 45 - 75) suffering onset of this horrible disease. At 57, I'm okay, I think. My 94 y/o mother escaped its grasp; many of her siblings did not. But mom did all she could during grandma's aggressive onset of Alzheimers...something we had never heard of. It took a village too because without help from neighbors mom would have lost her beloved mother under other tragic circumstances due to unpredictably of her actions, like leaving the house at any time of the day walking around East Los Angeles... neighbors were on patrol.

[@shmpooch]

Do we need a code? Not seeing the 60% discount by just following the link?

[@tttITA10]

Oh, it is SO important that doctors also know to care for Alzheimer's from the very first symptom. My old father's doctor would resist giving him treatment and the diagnostic when symptoms were very clear both for me and my younger sister, for he feared the diagnosis would be "too definitive" (which is a WILDLY stupid thing to say. WILDLY. I can't even find any logic to it.) So my father only got treatment when my mother could no longer too pretend it was all okay - the problem being she only got to this point in the middle of the pandemic, delaying proper treatment onset for MONTHS. This infuriates me so, so much.

[@instaperil]

Getting doctors to properly diagnose dementia and alzheimers is a huge undertaking. My mom recently died in the through of alzheimers. She died of diabetes but the alzheimers was bad

[@howtowithelizabeth7513]

Try finding a different doctor that can actually help

[@tttITA10]

@@howtowithelizabeth7513 No doubt about it! The only unfortunate part is that a doctor-patient relationship, specially when you grow older, is a heavy-on-trust relationship, which both leads to confidence in poor treatment practices, and difficulty changing doctors. On the bright side, however, continuity on the treatments and long-term medical following are great.

[@ems7623]

But ... What treatments are there? I thought it was untreatable

[@tttITA10]

@@ems7623 It's not curable, nor reversible, but there are a ton of treatments that slow its progression down significantly, maximizing one's quality of life and span of independent life.

[@gSWG3R]

Glad you guys reported on the limitations of the study, thank you :)

[@kira3835]

The fact that SciShow does this consistently shows how much integrity they have, and is a large part of what makes them one of my favourite sources of "Edutainment"!

[@finallyforfeited]

They're usually pretty good about doing that 😊

[@bleep8048]

They usually have the sources available in the description too, if you wanted to follow up or just assure they know what they're talking about.

[@yasmamadoomeh9842]

Heck yeah!

[@mooselove]

I’m a “”””“healthy”””” 31 year old female and I have horrible executive dysfunction and trouble just speaking coherently. They’d think I’m in mid stage Alzheimer’s if they talked to me on a bad day

[@R.M.MacFru]

I've always had that problem. And with hypertension, I wonder how it differs from someone who may have had a minor stroke. Some stroke patients have difficulty with language. Their brain oft times will bring up the wrong word.

[@AndreaCrisp]

I have MS and cognitive difficulties with language. I wonder if it could tell a difference between Alzheimer's, Multiple Sclerosis and a stroke victim. Probably someday.

[@LordRunty]

"A low cost, easily accessible option" And you just know that the US healthcare system as it stands would not be doing that if/when the test becomes available.

[@desu38]

Insurance companies, on the other hand...

[@hazelsleep4264]

Not scared of a lot, but Alzhéimers actually scares me a bit.

[@davidsonmg]

It would be great to see this study duplicated and get the same results.

[@Dolthra]

Part of the issue I'm seeing is that the model was trained on the people who it was then tested on (or the audio files of the people it was then tested on). Recognizing which patients have Alzheimer's and which don't on that sample is a pretty good indication this might be something a computer could theoretically do, but getting no false negatives or positives could easily be due to the testing being on such similar audio files as the ones it was trained on. All that said, while I am skeptical, it is promising. Even a much less effective model, if freely accessible, could go a long way to convincing people in the early stages of Alzheimer's that they should be professionally tested. Hopefully this is replicated, because even an AI that's right 60% of the time that anyone could access without the embarrassment of having to approach the conversation about Alzheimer's that currently exists would certainly go a long way to slowing the effect it has on people's lives.

[@drophy]

@@Dolthra OH, you mean, despite them having separate audios for training (1308) and testing (308), they might have not made sure the ones from the training group were from different people than the audios from the testing group? I really hope they did consider that and chose different people for each group, because yes, that would definitely mess up the experiment

[@gamesturbator]

Several years ago I was showing signs of BAD dementia (hallucinations, lost time, getting lost, the whole list). This year I found a really good neurologist and he said my problem was probably hydrocephalus all this time. My stupid insurance, Humana, wouldn't pay for even testing let alone treatment at the hospital he works at, so in January I'm going back to United Health Care.

[@venmosh4822]

insurance is getting tight, even with the cheapest care costs possible & generic meds.... my sis is going through daily epilepsy seizures yet it takes insurance over a week to "authorize" to fill her literally life saving medications. (it must be "liquid gold"!!🤑) It's sickening

[@gamesturbator]

@@venmosh4822 The ironic thing is that I switched from United Health Care because they did not cover dental or full vision. This coming year they cover everything that Humana did, but I won't have to struggle to find actual treatment. Humana was terrible about authorizing treatment at the very medical centers that employed the specialists they had sent me to in the first place. Oh, and no 24-hour nurse or ANY nurse for that matter. Glad to be rid of them. United Health Care all the way for me.

[@Dexy83]

Anytime an advancement in Alzheimer's looks promising it deeply affects my family. I don't have much hope for my lifetime (I'm almost 47 y/o) but I pray my 10 y/o niece is able to be helped before it hits her like the 4 generations before her. 🙏🏻

[@banjo-beard]

No false positives or false negatives? If anyone showed me a model like this, I would be highly suspicious of the methodology and suspect ovetfitting.

[@sbomorse]

Can we take a moment to appreciate Hank's jacket please.

[@scrunglenut6222]

couldn't they also use this to harm people? an algorithm could make it so insurance companies could know you have alzheimers before you do, and then kick you off their plan for having a pre existing condition

[@michaelmicek]

I think that, in the US, as part of Obamacare, that's illegal now.

[@COKTilYouDrop]

@@michaelmicek lol ppl still use that?

[@michaelmicek]

@@COKTilYouDrop What? What do you think Obamacare is?

[@COKTilYouDrop]

@@michaelmicek ur mom

[@COKTilYouDrop]

@@michaelmicek don't think you got the basics part of the joke there bro 🤣

[@jeffreyschweitzer8289]

Great…Alexa, Cortana, Google and Siri are right now listening to my phone conversations, figuring out how to have me declared legally incompetent so they can divvy up my assets between them

[@MarkWTK]

in my case, i hope they can divvy up my debt among themselves haha

[@ellid7425]

Thank you for making this video! My dad was quite recently diagnosed with alzheimers even though he is only 53 years old. It would be really interesting to see a video about how close scientists are to finding an alzhemers cure!

[@dalpz205]

How was it caught? I'm terrified of it and am curious how it's caught so early.

[@Call-me-Al]

Check out the Alzheimers documentaries with Terry Pratchett, it might be helpful for what to expect. He also was diagnosed at a young age. I wouldn't expect a cure anytime soon, however there will be more and more ways to slow it down hopefully within a decade.

[@dalpz205]

@@Call-me-Al Ty

[@illiengalene2285]

What did you do yesterday? Had a migraine. Slept for 18h. Ate chicken nuggets. Slept again.

[@StarCrusher.]

"hello, who's there?" "This is al" "Al who?" "Alzheimer's" "Wat." "You have Alzheimer's"

[@chillsahoy2640]

Hank (or John, or anyone from Vlogbrothers) I'd be very much interested in hearing one of your unfiltered views on Groundnews back on the original channel. The "jingle" made me very curious. Sincerely, a Nerdfighter who has completed the 2021 census.

[@Scorpio8116]

I'd be really interested in seeing a study that uses this AI with bilingual or trilingual people of different fluency. I'd imagine that would complicate things a bit (and I'm guessing this study was done purely in Japanese if it was developed in Japan).

[@lelaniaorr9884]

I love the subjects you cover, very interesting stuff. Would like to let you know that whoever is editing your videos could leave more natural gaps between sentences, I just realized that it is almost stressful to hear all those thoughts without any gaps at all. Is there some reason that you need to keep the video so tight and short? Who doesn't have time for breathing? Great topics, well thought out, great stuff. Thanks.

[@AnnoyingNewslettersPage6]

Where can we get that sick jacket?

[@Jolfgard]

How would the AI do differential diagnostics? Speech impediments are abundant and can have plenty more reasons than Alzheimers. Take all of the more common form of aphasia for example.

[@codename495]

I would imagine they take samples of the patients speech patterns over a timeframe.

[@david94549]

I was literally thinking last night about how the news, more particularly covid, is portrayed on the TV and how all the different news compares, now I have an advert for it in a video on my homescreen. I was thinking it's strange how the news are so emotive and persuasive with their presenting and do not really present an unbiased analysis of the actual data, and was wondering how different the perspective could be elsewhere

[@mythology2467]

No false positives or false negatives? That's literally just the best test right?

[@karmaarachnid8345]

I assume the issue is that the sample size isn't large enough to be confident that the 100% accuracy wasn't a fluke. If the older approach has been subjected to a lot of testing they can be confident that its true accuracy is within a narrow range of values. The true accuracy of the less tested approach can only be confidently estimated to fall within a wider range of values. It sounds like they're saying there's still too much overlap between these ranges to confidently predict which test will prove to be more accurate in the long run.

[@sohinidutta97]

Yeah but the sample size isn't very large, so we need to wait for bigger studies

[@oldcowbb]

something is fishy when someone claim their ML method has 100% accuracy

[@xarin42]

It seems great for getting yourself to get started on testing without the embarassment of dealing with another person.

[@goreobsessed2308]

Exactly what I was thinking

[@michelleallen2294]

All I can really say as a first line dementia care specialist is that any progress is good progress. We review and change care plans all the time. Diagnosis can be short term. It's great to have a point in the right direction, even if it simply.leads to further investigation. Thank you for sharing the inspiration and creativity.

[@michelleallen2294]

Oh, also... there are so many different forms of dementia that an approach like this is more useful to indicate than diagnose. But that's still progress! Vascular dementia can look a lot like seizures in older adults and alzheimers can look a lot like drug/alcohol abuse or cranial trauma later in life. So there's a lot to cover and it's just wonderful that there is active research.

[@Dryblack1]

I haven't been diagnosed yet, but I think the youtube algorithm figured out I have ADHD. I had never watched videos about anything like that, and one day one was randomly recommended to me. "Should you be assessed for ADHD?" The doctor in the video described my whole life like he'd seen it a million times and I'm now seeking a real diagnosis. If I'm diagnosed, I will tell people as a joke that the youtube algorithm diagnosed me

[@mclijah6472]

very fascinating thank you hank :)

[@PheOfTheFae]

Is it able to distinguish from other causes of delayed speech? They say you talk slower when you're depressed, for example.

[@kkmardigrce]

Hmmmmm... I hope such apps don't end up sending the info to insurance companies...

[@burnyizland]

That GroundNews is a great idea - I currently go over 7 different news sites from 3 different countries (and often more but always at least those) every day to figure out the same thing. Thanks! As for this new AI, it can't get here soon enough. I'm currently fighting both my doctors and my family to recognize something is wrong with my brain - and my language is the most affected after memory. I'm only in my early 40s so they don't think it's likely, and I think because I have a high I.Q to begin with I'm still compensating well, but then my family thinks I'm faking and only 'forgetting' what/when it's convenient and my Doctors (who I've seen only twice in person since Covid hit) aren't seeing me struggle because I take notes and rehearse for weeks before my appointments with them. Easy and accurate diagnosis would be a boon.

[@NewMessage]

Now if only I could remember the number...

[@j.megatron]

That coat.... there's a similar one on eBay for 89

[@RafaelCardoso299]

Awesome video !

[@kafka9627]

My google home Realising i have Alzheimer’s like you forgot to pay your bills to google remember? You gotta pay us again

[@sambednar3351]

I’d be interested to see this study done with participants discussing something other than the activities of the previous day. Seems like Alzheimer’s patients would inherently have more trouble with that topic which could skew the results.

[@nightthought2497]

The implications of this tech is... not great. Cool diagnostic tech, terrifying implications.

[@AuntBibby]

im slightly sure my mom has early signs of alzhiemers (shes a bit forgetful now & slowtalky, and all 7 of her aunts+mom got alzhiemers) how do i get her into treatment? what even is alzhiemers treatment?

[@codename495]

There are meds and cognitive therapy to help slow the progression, but should she have Alzheimer’s progression is inevitable and will come to an end stage eventually. The first step is getting her to discuss her symptoms with her doctor and go from there.

[@Meuszik]

I love the jacket, Hank!

[@Wreckz_Tea]

How can you say it's not significantly better when it was 100% accurate in correctly diagnosing and not misdiagnosing? It literally can't get any better and that seems significant to me

[@thedanelsan]

As I understood it, it is about AS GOOD AS the traditional test (which is likely less accurate than the full diagnosis by a professional that was originally used to determine the subjects that had the condition) in the "not giving false negatives" category, and BETTER at the "not giving false positives" category, but was only fully accurate on that last one. Independently of me being right or wrong about this, the video could have been more clear there, for sure. That aside, "significant" in statistical analysis is basically the likelihood that the result couldn't have occurred by chance. It only being close to significance means that, given their sample size, how much and how consistently better the results were and such, it is still quite possible that the AI just happened to guess right a bit more this time by chance, not that it will consistently perform better than the traditional test. For a really exaggerate example, imagine if this study was fully 100% accurate in the diagnosis of all subjects, but that was 100% of only 3 individuals. Nobody would fault you for not being ready to trust it to be equally successful in millions of people worldwide.

[@socrabe]

It is, but with a sample of only 24 patients with Alzheimer and only in Japanese, they will need more tests to make sure it works for larger sample of people, with a wider range of languages. If it's conclusive, they will have to try to actually diagnose with it.

[@TsjuunTze]

Hello, this is AI speaking. Yes, hello. This is Mr Alzheimer. Can you hear me?

[@Bealzbob]

I might have alzheimers, but at least I don't have alzheimers.

[@RiiFT]

If someone asked me this question without context I'd give them a very strange look.

[@Lady-ER-1967]

I was once on the phone to someone’s sister, I didn’t know the lady however I did know she had diabetes, I know nothing about the condition even to this day however I could tell she needed to do a test on herself, she said she’d just done so, I insisted she do another test right away, she did & found she wasn’t well, maybe humans like other animals are more sensitive to these things.

[@lonewander912]

Didn't I recently hear that there wasn't going to be any more SciShow Psych?

[@socrabe]

I did too. Sad news. :(

[@onalla]

yeah :( although i think they said they'd stop posting around new year's

[@filangimn1746]

This channel will stop posting new videos next year.

[@alien9279]

Man i love science

[@j.megatron]

That coat is making my soul hot 🔥🔥🔥🔥, I need it NOW

[@mysteriousshadow395]

I have no idea what happened yesterday...

[@40megafonzies]

william shatner audio would destroy that algorithm

[@xynor001]

Totaly true ! I often work with elderly people and I assure you its totaly possible by the speech and behavior to tell who as Alzheimer and how early they are in the desease... Well I work in IT maybe because computer need a lot of ourself memory to understand how it work it appear to me more easily.

[@ramonpuntpuig8751]

Does anybody have a link to the Alzheimer's AI? The one that works best, or TICS. Are they uploaded online? I find them interesting

[@p1nkfreud]

How is a 16% differential = p < .05?

[@kenscriven1589]

Should use this on sleepy joe😂

[@jessebob325]

Roses are red Violets are blue I have alzheimer's Cheese on toast

[@noiJadisCailleach]

But... i don't even know where to go to check for Alzheimers. What kind of doctor do I have to go to?

[@nonchalantd]

fly NASA jacket

[@Crootcovitz]

I wonder how the language a person speaks would affect this method.

[@nico-3663]

Oh man I answer phones for an insurance company. Mostly elderly people. I’ve heard the difference between normal old and obviously mental confusion. Paranoia too. They should use those calls as diagnostic tool lol /J

[@rimaruttempest6109]

i was like “who’s Al..?”

[@rimaruttempest6109]

no. no ray it’s AI. not AL.

[@danieldeelite]

Early signs of blunt

[@anirbanpatra3017]

Lol...Feels Great That I know What Boosting... Machine Learning Engineer here

[@thepeff]

This will be a great boon to rich people in twenty years

[@hackall360gaming]

beginning of this sounds jumpy between 20 and 30 seconds... was this AI generated, or AI edited?

[@ababababaababbba]

I cant read

[@curiodyssey3867]

Lmao

[@MichaelValenzuel4]

i wanna do this!

[@CanonessEllinor]

Is there actually any benefit to diagnosing Alzheimer’s early? It’s not like you can cure it, so living in blissful ignorance for as long as possible seems the better outcome. The diagnosis is a death sentence, isn’t it better to postpone it?

[@Zepha21]

That is a great tool to find out if someone deserves medical insurance or a credit card! All hail capitalism!

[@DSAK55]

Ground News, lol must be by EEs. Who else would use the ground symbol

[@cee8mee]

Wouldn't it be more effective to do this over time, as cognitive disease is generally progressive? AI, as an aid to a yearly/bi-yearly interview, could highlight a change over time. I say this, because I stuttered as a child, and part of my therapy is pre-rehearsing common phrases for daily use. I also use verbal tricks for my lisp, which reappeared after some dental work. Would AI data imply my lifelong speech issues were cognitive decline?

[@64fanatic]

Phone scammers 👀👀

[@barbarajeanne8351]

I HAVE THAT SHIRT !!!!

[@seeker296]

people still doubting AI...

[@Viherion]

They fear AI for the wrong reason. Don't be afraid of the machine. Be afraid of the people, who train the machines without understanding what they are doing. Machine Learning is still very much a black box. We often don't know on what exact criteria a learned model makes its decisions. Background noise we humans don't even notice can have a huge effect on the model

[@robertbones326]

Calm down Nasa boy

[@RickyHil]

Joe Biden was giving a speech in the backround when i was on the phone and i got nailed with a false positive!

[@ESL-O.G.]

First! Oh what a glorious day

[@DuAuskenner]

Congratz!

[@xploration1437]

Is your boyfriend impressed Andy?

[@Karishma_Unspecified]

@@xploration1437 what's that supposed to mean?

[@fvckyoutubescensorshipandt2718]

Too bad that won't work when I get Alzheimer's since I'll still be hardly using my phone at all and mostly talking to myself. In my case Alzheimer's sounds like mostly a blessing, most of the crap in my head I'd rather have deleted anyway.

[@lelajoyner9376]

Alexa saving lives. I love it.

[@SaucerJess]

💙🧠💙

[@petrairene]

So, what does this software make of Donald Trump's rambling?

[@BenRitter]

I have a lot of presidential audio to run through this.

[@swimdownx6365]

If slim molds single cells can have internal memory . Doesn't that sudgest have Cells store memory and their for how our brains and bodies store memory

[@SeanLinsley]

I wish everyone would stop using "AI" to mean traditional algorithm with some machine learning bits. an actual AI would be able to find a solution for many problems instead of being programmed to solve that specific problem

[@AshArAis]

That's now referred to as AGI

[@Viherion]

AI is a huge field in computer science. No, it does not refer ro the sci-fi version "AI". All of machine learning is part of AI. What you said is basically like when people say "but evolution is just a theory!". The word theory has a different meaning for scientist than in everyday conversation. Same goes for "AI". But yes, it is a pity, that this difference is rarely clearly communicated to the public.

[@lordsharshabeel]

Can this AI make any sense of Joe Biden’s rambling speech?

[@IrocZIV]

like

[@jilliangellately999]

elmoABC UiI

[@AngryDuck79]

Am I the only one who sees the irony in the Scishow gang being sponsered by a site dedicated to refuting political spin?

[@kellydalstok8900]

Repetitive phrases, tell me about it. Repetitive stories as well. And it’s ALWAYS about herself and her family, because she has been very self centered person all her life.

[@xploration1437]

Better check Brandon.

[@pep206]

They ought to play it a Joe Biden speech and see what it says.

[@AnnoyingNewslettersPage6]

Okay, so you want to go there? Rewatch the part where it talks about using the same turns of phrase or changes the subject when the person becomes confused or can't remember something...Then listen to how often Trump changes topics in wildly incoherent ramblings and uses the same turns of phrase, the best turns of phrase...

[@pyperhaylie3162]

Ooo second!

[@Ava_The_Avatar]

Seveth comment

[@davet319]

Need to use this on Joe Biden and Nancy Pelosi, stat!

[@curiodyssey3867]

Waaaaahhh

[@davet319]

@@curiodyssey3867 Yep, just like you all cried about Trump for 4 years

[@beaudavis3808]

What if the person in question also have Asperger and Cerebral Palsy? What good will this tool be when that is right there?

[@SendFoodz]

Joe Biden