Slashdot Mirror
Technology To Detect Alzheimer's Takes SXSW Prize
An anonymous reader writes "Being able to diagnose people with Alzheimer's disease years before debilitating symptoms appear is now a step closer to reality. Researchers behind Neurotrack, the technology startup that took the first place health prize at this year's South by Southwest (SXSW) startup accelerator in Austin. The company says their new technology can diagnose Alzheimer's disease up to six years before symptoms appear with 100% accuracy."
A good screening test is one that identifies a treatable disease.
Only on Austin would a music festival give out "startup accelerator" awards...
You morons just posted this article last week. Here's the link:
Uh oh...
This device can detect Alzheimer's, and it sure as hell can detect Alzheimer's!
Not quite in line with the data. FTFA "Kaplan said 100 percent of subjects who scored below 50 percent on the test have gone to receive an Alzheimer's diagnosis within six years, while none of those who scored above 67 have developed Alzheimer's." This doesn't equate to 100% accuracy. What happens between 50 and 67%? Plus it doesn't say what the sample size is. Is it 1, 10, 100, 1000? Some more robust statistics would have been nice. They were probably trying to keep it simple instead of confusing people with 99/99, but they could have said "approaching 100%".
GENERATION 27: The first time you see this, copy it into your sig on any forum and add 1 to the generation.
Claim of 100% accuracy.
A Twitter full of "launch" and "pitch" announcements and not much else.
A website that is nothing more than a placeholder.
Yeah, they're going straight into the history books, they are.
You want me to believe you, publish, and let people rip it apart. If the public-facing part of your whole organisation is talking of nothing more than startup awards and pitches, I don't see how you can be doing proper research, or how you can be selling it to medical establishments. And without bothering to provide evidence of either, I can only assume it's snake-oil.
Within 6 years is a pretty easy prediction if you ask me.
My prediction: "100% of those who scored below 100 percent on the test will be stone cold within 100 years at the most". I guarantee it's 100% accurate too.
"A door is what a dog is perpetually on the wrong side of" - Ogden Nash
They've done studies, you know. Sixty percent of the time, it works every time.
It is very easy to make a test that detects 100% of patients who will eventually get a disease. Just make it always say "positive" and you're done. The hard thing is balancing the ability to detect a disease and avoid false negatives (sensitivity) with the ability to detect absence of disease and avoid false positives (specificity). Related to this are the positive predictive negative predictive values. Since Alzheimer's is very difficult to diagnose clinically and the only definitive proof is a biopsy/autopsy, I very much doubt a screening test would exist with a 100 % sensitivity and/or specificity.
Asking "What time is it" every 3 minutes is a pretty good indication of Alzheimer's
This will be useful to patients when there's something that can really be done for them once the early diagnosis is made. Right now all of the actions that can be taken are in the "we hope this will slow down the symptoms" category, and the sad fact is that it's hard to even prove that (unless you ask a Big Pharma marketing agent). The big concern is how organizations like medical insurance companies will use such information to the detriment of the patient, as in resulting in sky-high premiums if they can get insurance at all. My wife has MS, and she can't get long-term care insurance at any price. I have a family history of Alzheimer's, but my brother and I have both decided that we won't take any tests of this nature until there's a gain for us to balance out the risk of the information being used against us.
Especially bad if it turns out that 99% of the people tested scored between 50 and 67...
"By monitoring the way a person moves their eyes, and watching how they view novel images versus familiar images, we're able to detect perturbations that exist on the hippocampus". That's an unsupported leap. Each of us makes about 250,000 saccades every day and their targeting is controlled by a variety of brain modules, including the amygdala. Saccades are made to targets that hold significance. Alzheimers patients are reported to make far fewer saccades than healthy people, and some studies have shown improvements in mental functioning with exercises to increase the frequency of saccades. To some extent dementia sufferers lose contact with the world because they lose all interest. Memories that are not felt to be important are either not laid down properly in the hippocampus or are quickly deleted. All this experiment proves is that those who did poorly had little interest in the set task.
My grandmother had Alzheimer's. My mother is now at the stage where it may show up and has been under the stress of that knowledge for at least 6 years now. She may have liked to know 6 years ago what she should try to plan for the future.
Me? I'm not sure if I would want to know or not -- might be able to decide depending on whether mom starts showing symptoms or not -- right now the risk seems a little removed yet.
The only peer review study I can find searching for "Neurotrack" and "Alzheimer’s" is "A Behavioral Task Predicts Conversion to Mild Cognitive Impairment and Alzheimer’s Disease" (Zola et al, 2012; doi: 10.1177/1533317512470484). They had 32 mild cognitive impairment and 60 controls, and followed for 3 years.
From the abstract:
Scores on the VPC task predicted, up to 3 years prior to a change in clinical diagnosis, those patients with MCI who would and who would not progress to AD and CON participants who would and would not progress to MCI.
So it's hard to know what data is substantiating the claims we see in TFA. Certainly nothing at a clinical level, but it also seems quite promising. Probably a mix of genuinely encouraging early results and typical start-up exaggerations. I hope it lives up to its promise!
Hate to self-reply, but I've read the paper now – and it's definitely the source paper for the data in the posting (50% and 67% figures are right in the text). Given the paper was published december 2012 and reported a 3-year follow-up, and that the current report claims a 6 year followup, I have to wonder why it took them basically 3 years to publish the original study. In any event, a few more details:
9 subjects fell into the sub-50% range, 8 of these had further impairment at the time of publication (3 years), so presumably the holdout also ticked over in the subsequent 3 years. I've got to say that 9 is an awfully small set to draw strong conclusions from, with respect to positive identification.
43 subjects fell into the >= 67% range, and none of these worsened in three years. If we believe the current report, they're still ok.
40 fell between 50 and 67, and 9 of these got worse.
To summarize, about 43% of the sample fell in the grey area. Within that area, doing some hacky fiddling with the ROC curve they provide, the area under the curve is something like .56 – which ain't very good. So basically what the data says is that people who perform very poorly are likely to show cognitive decline. Of the remaining folk, 50% won't get worse, and 50% may or may not....
So we're not looking at a slam-dunk here. But it's got some promise.