Voice algorithms spot Parkinson's disease
 

From the BBC:

Voice algorithms spot Parkinson's disease

" ... mathematician Max Little has come up with a non-invasive, cheap test which he hopes will offer a quick new way to identify [Parkinson's] disease."

"Mr Little has discovered that Parkinson's symptoms can be detected by computer algorithms that analyse voice recordings."

"He is aiming to record up to 10,000 voices and has set up local numbers in 10 countries around the world. In the UK the number is 01865 521168."

http://www.bbc.co.uk/news/technology-18427851

An excellent idea.

If it can also measure progression accurately, this will give a very useful tool for clinical trials: ring up every day to get a "score". With enough people, you'd know within days whether the treatment was working.

John

Web site with national telephone numbers (US, Canada etc.):

http://www.parkinsonsvoice.org/index.php

John

Thanks, John! This sounds really promising. I will share on social media and in a few other forums.

Laura

Making the call is easy. You get an ID number which you can use at their website to request feedback.

I've just received an email from them:

The good news: >3800 calls.

The bad news: feedback is delayed.

John

The PVI was featured on Fox News :

Could speech test help detect Parkinson's?

The Parkinson's Voice project has now had over 5,000 replies.

A paper by Tsanas et al. [1] gives more details. For instance, with their system "UPDRS can be estimated to within about 3.5 points difference from the clinicians’ assessment, which is clinically useful given that the inter-rater variability for UPDRS can be as high as 4-5 UPDRS points."

I think, however, it will be better than what that implies. This is because what we really need is a measure of the severity of our Parkinson's, per se, not an estimate of our UPDRS score. My hunch is that the quality of one's voice reflects deeper, more visceral processes better.

[1] "Using the cellular mobile telephone network to remotely monitor Parkinson's disease symptom severity"
Athanasios Tsanas*, Max A. Little, Patrick E. McSharry, Senior Member, IEEE, Lorraine O.Ramig
http://www.maxlittle.net/publication...ular_updrs.pdf

John

twins
 

I have an identical twin who does not have PD. We both submitted our voices for comparison and they were pretty excited about it!

As of 16th August PVI had received 11,217 calls (US, 6601; UK, 2097; France 1337; Canada, 754; ...).

In my view this is an exciting initiative. It's getting huge numbers of people taking part. With these numbers its benefits should go way beyond diagnosis. As I understand it, when fully developed, the technique should be able to give a Parkinson's score. This should allow very rapid clinical trials, e.g. ring in to give a baseline, take curcumin for a week, ring in an end-point call, a day later have the results.

So, if you haven't already done so, please call in.

John

Here's a transcript of "THE NEXT LIST Detecting Parkinson's Disease by Voice Aired December 2, 2012 - 14:00 ET " on CNN.

http://transcripts.cnn.com/TRANSCRIP.../02/nl.01.html

The most interesting bit is:

"LITTLE: This sort of high-frequency data about how people's symptoms change over time has not readily been available for this. This is a new frontier in exploring how individuals respond to treatment to different drugs. And the value of this, this has enormous value in clinical practice"

His approach offers so much more than just diagnosis.

John

Huffington Post article by Max Little:
http://www.huffingtonpost.com/max-li...b_2545128.html

John

i can see an immediate benefit if it can only help detect pd at stages where the first physical signs are enough to make 1 see a doctor. i imagine millions are wasted in tests for pinched nerves and other problems on diagnosing which to the trained eye is pd.
seems every yopd'er i have talked to had a few or more tests/doctor visits before seeing a neuro.
i imagine in a few years diagnostic tests for a lot of diseases will be a smart phone app.
there already are some.

so this math guy sets up a web site to record 10,000 voices, big deal, there's been research in using voice to diagnose and measure pd remotely for years. billions have likely been spent on voice recognition software by securitty/law enforcement/cia. seems like just an attention getter to me. this is going to take a tremendous amount of $$$$ and research before it might become useful for diagnosis and drug companies/researchers would use it.
and there are companies much farther along in this area. TED imho can be more hype than substance,
http://www.imedicalapps.com/2011/07/...sons-patients/

i imagine if you add video motion analysis, face recognition, voice recognition, movement measuring, cognitive testing, vital signs, all which could be cheaply captured on a smart phone, that would be far more useful and will likely be what going to materialize rather than just voice. there are likely hundreds of diagnostic/measurement apps on iphones either tied to your healthcare provider directly or for example you can take a picture of a mole, etc and send it to a company to be diagnosed.

http://www.idt.mdh.se/kurser/ct3340/...bmission_1.pdf

sorry if i seem a little negative, but this guy is getting way too much attn. imho.

soccertese,

Smell the coffee!

Little gets the attention he does because he has a game changing technology, the support of Andy Grove, one of Intel's founders, and is a natural communicator. You're right about other people having similar technology: it's an idea whose time has come. But it is Little who in just a few months got over 10,000 people to submit a voice sample.

Little's approach enables rapid, free diagnoses. But it offers other advantages as well. I suspect that used in addition to the normal doctor-led diagnostic process it will reduce the number of misdiagnoses. It enables symptom measurement which, in turn, offers the possibility of mass trials, reporting quickly.

If I were to select those initiatives in the last 5 years which were most likely to translate into an improvement in the lives of medium term PwPs, I would choose (in no order):
Little, Parkinson's voice initiative;
Brin, 23andMe, mass genetic testing;
Alberts, forced exercise;
Jannetta, peduncle compression.

John

i smell the coffee all the time, i'm in seattle.
he got 10,000 anonymous calls, no way to verify the honesty of the caller, iseems like a publicity stunt. he still has to do some real studies with real patients over time to see how well his algorithms can measure pd.. other researchers are taking a more comprehensive approach.

i understand the value of early diagnosis or better diagnosis or easily accessible diagnosis. but until he actually publishes something for peer review that has gives an error rate rather than just blow his own horn i'm not going to take him seriously.

if he thinks he can detect a higher risk of developing pd before there are visual symptoms, unless he can correlate that with a person's genome or other biomarkers, he has to wait until pd can be proven. either way, he has to do careful research over time. and if there are too many false positives/negatives it's worse than useless.
where does tt say andy grove is supporting him? i think the article stated intel was developing it's own pd detection methods using voice/etc and shared some recordings with little.

ji wouldn't attempt to give an opinion on the "top5" but i'm not sure even 23ANDME would be included. imho, the jury is out on forced exercise until the results from larger clinical trials come out.

He deserves his 15 minutes
 

Where has the comprehensive approach approach gotten us? In over 40 years of research, Sinemet despite its downfalls is still our only really proven treatment. We still have no common measuring point to develop blood or cerebro spinal fluid tests. When Andy Grove got involved in funding PD research, he made the key observation that never before had he seen people work so hard and achieve so little.

Andy Grove's Last Stand.

What Little is envisioning is revolutionary. He, along with Brin and Grove, are taking the medical research industry to the next level. They recognize we have a model that excels in generating a lot of info and data but scarcely makes anything of it. When does anyone and step back and say why no real progress? Andy Grove is doing just that at UCSF; they designed the virus vector after studying the Amgen records within five years of their aborted trials. That is being used in the Ceregene trials and will be used in the NIH GDNF trials. These guys, along with MJFF, are the only reason we seem to be making any progress.

I think Little's research is huge; if it pans out, we have one element that links us all, so we finally have that much desired point of measure to use in both diagnosis and tracking progression. Soccertese, you are right. Much literature already exists on vocal and speech changes in PD. From what i gather, Little is using that info and data, then using mathematical algorithms to make it useful. He essentially using what is already proven in many other studies and synthesizing it. How do you put that through phase trials? It either works, or it doesn't.

I do think the app development stuff a bit premature, but I don't think Little is being hyped.

Soccertese,

You write that Little "seems like just an attention getter to me" and that his collection of 10,000 calls "seems like a publicity stunt". What evidence do you have for these assertions?

It seems to me that he's a good guy doing a good job with good technology.

John

LAURA,
i gave an opinion on LITTLE, nothing in his background makes me think he can pull this off, and imho TED = hype, it's more about presenting IDEAS, not actual results. If you want the maximum attention you present at TED
and some of his TED presentation made no sense, for example:
"Current symptom tests are done in a clinic. They are expensive, time-consuming, and logistically difficult. So mostly, these tests are not done outside clinical trials".

THESE TESTS ARE EXPENSIVE AND TIME CONSUMING? WHAT TESTS? granted, many pd'ers get tested to rule out other problems, often given an MRI but even if this voice test detected PD you still have to see a doctor to verify it.

" In collaboration with one of my students, I worked for several years and now have algorithms that, in the lab, reach around 99 percent accuracy in detecting the disease. We also know how to predict the severity of symptoms to within a few percentage points of clinical judgment."
I'D LIKE TO SEE THE RESEARCH ON THIS. THIS WAS LIKELY DONE IN A CONTROLLED LAB SETTING.
http://www.maxlittle.net/publication...hort_07-11.pdf


I am not attacking the concept. I also realize just how difficult it will be to pull this off, imagine if the software has a bug in it and pd'ers are told they don't have pd and vice versa, so this is going to require A LOT OF TESTING. if you were a healthy individual and were told this test had a 90% chance you might develop pd, would you even take it? And even though the FDA doesn't regulate software apps as medical devices yet, they will and that's not an easy approval.
imagine the can of worms that will open up if you called in, took the test, were told you may have pd or a high probability of developing PD and your're 25. that shouldn't stop anyone from trying to do this but i kind of see this voice test as better done in the doctor's office for initial diagnosis and then done by the patient to track their progression. i don't know about you but when my pd diagnosis was traumatic.

think about tracking progression. i can sound perfectly normal during the day and have the typical pd voice when my med's wear off. so even that isn't a simple proposition.


http://www.maxlittle.net/publication...00887-2011.pdf
This paper had 43 patients and didn't provide a table of each patient's results.
so unless they have done a tremendous amount of work since then, i'd say they are in the VERY EARLY STAGES of development and god help us if every pd researcher hyped their work as much.

Sorry if i'm being overly argumentative and/or critical. I certainly don't want to turn this into a debate about pd research, and again, just criticizing the self-promotion of this guy, he ain't working on a cure.

AAV vector
 

e
Laura
I am interested in this information you posted . If you have a reference or source where you got this info, could you please post it? As I am looking into gene therapy-based therapeutics, it seems like AAV vector is the most often used one and has been shown to be safe in several studies. Though NIH's gdnf trial says it is with AAV, it was not clear if it is the same/similar to the one used in ceregne trials. I was just about to send letters to NIH and PCORI about this matter. My specific questions are:

Is NIH using the same AAV vector as the one used in Ceregene trials?
If so, all the relevant protocols, assays for testing antibodies etc., are rather should have been approved by the FDA. Ceregenes studies already established this vectors safety.

If so: Why is NIH taking so long to start GDNF clinical trials? I wanted to see if I can get a straight forward response from the NIH.

If not, is there anything PCORI (since it is the patient centered arm of Human Health Services ) can do about this issue.

Your comment caught my attention and it is just what I was looking for. Thanks

Girija

Voice initiative started with data from Intel
 

[QUOTE=soccertese;951341]i

where does tt say andy grove is supporting him? i think the article stated intel was developing it's own pd detection methods using voice/etc and shared some recordings with little.
/QUOTE]

Try googling their names together. Max Little did not have PD on his radar until he ran across a colleague researching voice in PD for an Intel at home PD program of some sort that never saw the light of day. Intel had 50 recorded voices and all kinds of data, and their guy thought of Max's algorithms...a team for the voice initiative was formed from this post meeting Grove.

http://www.medgadget.com/2012/08/int...nitiative.html

These original recordings were part of a 6 month research project. Some of these tests are performed routinely by speech pathologists. Have you ever been assessed by one? They can tell you more about your condition than your neuro from your voice including severity and if only one or both sides of your brain are affected. Little is synthesizing a lot of what is already known.

John, this is very interesting.

In my own studies I discovered that the pulses from your brain stem and spinal chord that control every part of your body are rich with information about who you are. These pulses show up in the tremor in your hands, in your voice, in the pulses in the iris of your eye, in your breathing, heart rhythms, facial expressions, handwriting, etc. So, theoretically it is possible to capture and mathematically evaluate every movement you make for a host of diseases and physical problems.

I've already found data that indicates that your ancestry can be tracked by those signals as if you are looking at genetic DNA.

It sounds amazing, I know, but I sincerely feel that we are on the cusp of breakthroughs in mathematics and science that are going to transform the way we look at the world. Chaos Theory has already had some significant impacts and it's a brand new science.

It's an exciting time to be alive. At 62 years old I might be too old to benefit from many of the breakthroughs that are yet to come in PD but generations younger than I have much to be hopeful about. (Well, I guess that DBS sounds exciting and it's available now.)

Isn't it a pity that the social, religious and political sectors always have such a hard time keeping up with science? (i.e. stem cell research).

Thank you for your posts.

Steve

My error
 

Girija,

My post on the AAV is a hot mess. My apologies. I partially included some of my notes which had lots of assumption built in and mixed up research groups. I need to start proofing evrything. There is so much misinformation out there on PD already, and I just added to it. Ugh.

Anyway, the connections are not as clear as I wishfully made them out to be in my notes though there are some very close ties. It turns out that Ceregene has its own patent on the AAV being used and they are working with Genzyme. Genzyme in turn holds license to the convection AAV patent holder at UCSF (Krys Bankeweicz), but are studying delivery of an enzyme AADC that helps produce dopamine. To further muddy the waters Ceregene's founders and owners comprise two scientists from UCal SanFran. Andy Grove funds research at both UCSF and MJFF with a strong interest in neurotrophics. In two articles, it appeared that all of researchers I just mentioned answered directly to Andy Grove, so it seems this is not forté.

As for the convection AAV; my assumption was that NIH must be licensing as I could only find UCSF as being sole patent holder on the convection enhancement, but these were my notes. I intended to have a qualifier in the post. Now it is too late to edit.

So sorry about this; I am embarrassed. Though in researching this anew to be sure I got it right this time, I ran across the following updates see the last entry on GDNF...is the NIH study their Phase I?

http://neurosurgery.ucsf.edu/bankiewicz/parkinsons.html

Laura

[QUOTE=Conductor71;951954]

i'm certainly not criticizing speech analysis nor speech pathologists. nor science.
i used a speech pathologist for my son, loved him. i've never used one but might need to get some speech training as my pd progresses.

i'm just incredulous at the attention this guy got by speaking at TED yet he' was virtually unknown before TED. and how the news media just gushed all over him without questioning one thing he said or sought other opinions.

Little's latest (2013) paper:

"Objective: To assess the clinical value of online, self-reported PD symptom data recorded by users of the
health-focused Internet social research platform PatientsLikeMe (PLM), in which patients quantify their
symptoms on a regular basis on a subset of the Unified Parkinson‟s Disease Ratings Scale (UPDRS)."

Data
6074 PD patients on PLM
100 included in this study, qualification to be included 15 UPDRS scores.

Results
"Validation demonstrated the high-frequency self-reported data is consistent with a low-frequency clinical data
set in common use in clinical PD studies. The distributions of PLM to PD-DOC “off” scores are essentially the
same".

Obviously it can be perfected, but as I see it, the technology is in place. The question now is:
Is there the political will to run simple clinical trials in this way?

Reference

[1] M.A. Little, P. Wicks, T.E. Vaughan, A. Pentland (2013)
Quantifying short term dynamics of Parkinson’s disease using self-reported symptom data from an Internet social network
Journal of Medical Internet Research, 2013;15(1):e20, doi:10.2196/jmir.2112
http://www.maxlittle.net/publication..._pd_web_v2.pdf

John

Laura,
Thanks.
No worries, I just got excited thinking that there is some sort of coordination among researchers! Thanks for all the info.
Girija

Girija,

There very well may be but would not know for sure without contacting them. Do you have any guess as to what the hold up may be? Do you why convection enhanced is better?

Thanks!

Laura

Wider data collection.

"PatientsLikeMe and the nonprofit research organization Sage Bionetworks announced today the launch of a new crowd sourced study to develop voice analysis tools that both researchers and people with Parkinson’s disease (PD) can use to track PD disease progression

In collaboration with TED Fellow Dr. Max Little, the Patient Voice Analysis (PVA) project will for the first time combine data from two sources: phone-based voice recordings that Dr. Little’s software collects and analyzes to detect markers of PD; and information reported by patients using PatientsLikeMe’s Parkinson’s Disease Rating Scale (PDRS), a tool that documents patients’ answers to questions that measure treatment effectiveness and disease progression."

http://www.epda.eu.com/en/news/02-05-patientslikeme/

I like this news:

DATA + ANALYSIS = INTERVENTION IDEAS

I wish the collaboration included mention of trials of minor interventions, such as circumin.

TRIALS + ANALYSIS = THERAPIES

John