So, well, I do applied math, and this is a peculiar problem for anyone who does applied math, is that we are like management consultants. No one knows what the hell we do. So I am going to give you some -- attempt today to try and explain to you what I do.
Bavim se primijenjenom matematikom, a problem svojstven svima koji se bave primijenjenom matematikom jest da smo poput savjetnika za poslovanje. Nitko ne zna što zapravo radimo. Ja ću vam danas pokušati objasniti što radim.
So, dancing is one of the most human of activities. We delight at ballet virtuosos and tap dancers you will see later on. Now, ballet requires an extraordinary level of expertise and a high level of skill, and probably a level of initial suitability that may well have a genetic component to it. Now, sadly, neurological disorders such as Parkinson's disease gradually destroy this extraordinary ability, as it is doing to my friend Jan Stripling, who was a virtuoso ballet dancer in his time. So great progress and treatment has been made over the years. However, there are 6.3 million people worldwide who have the disease, and they have to live with incurable weakness, tremor, rigidity and the other symptoms that go along with the disease, so what we need are objective tools to detect the disease before it's too late. We need to be able to measure progression objectively, and ultimately, the only way we're going to know when we actually have a cure is when we have an objective measure that can answer that for sure.
Ples je jedna od najljudskijih aktivnosti. Oduševljavaju nas baletni virtuozi i plesači stepa koje ćete vidjeti kasnije. Balet zahtijeva izvanrednu razinu vještine i visoku razinu umijeća i vjerojatno neku razinu početne prikladnosti koja bi mogla imati genetski element. Nažalost, neurološki poremećaji, kao što je Parkinsonova bolest, postupno uništavaju ovu izvranrednu sposobnost. To čine i mom prijatelju Janu Striplingu, koji je u svoje vrijeme bio virtuozni baletan. Velik napredak i tretman postignuti su tijekom godina. Međutim, 6,3 milijuna ljudi diljem svijeta ima ovu bolest i mora živjeti s neizlječivom slabošću, tremorom, ukočenošću i ostalim simptomima te bolesti. Stoga trebamo objektivne instrumente za otkrivanje bolesti dok nije prekasno. Moramo moći objektivno mjeriti razvoj bolesti, i konačno, jedini način na koji ćemo znati kad budemo imali lijek jest kad bude postojala objektivna mjera koja će na to moći sa sigurnošću odgovoriti.
But frustratingly, with Parkinson's disease and other movement disorders, there are no biomarkers, so there's no simple blood test that you can do, and the best that we have is like this 20-minute neurologist test. You have to go to the clinic to do it. It's very, very costly, and that means that, outside the clinical trials, it's just never done. It's never done.
Frustrira to što kod Parkinsonove bolesti i ostalih poremećaja kretanja ne postoje biomarkeri, stoga nema jednostavnog krvnog nalaza koji biste mogli napraviti, i najbolje što imamo jest 20-minutni neurološki test. Da biste ga napravili, morate otići u bolnicu. Veoma je skup, što znači da se osim u kliničkim ispitivanjima nikad ne obavlja. Nikad se ne obavlja.
But what if patients could do this test at home? Now, that would actually save on a difficult trip to the clinic, and what if patients could do that test themselves, right? No expensive staff time required. Takes about $300, by the way, in the neurologist's clinic to do it.
No, što kad bi pacijenti mogli ovaj test obaviti kod kuće? Time bi se poštedjeli teškog puta do bolnice. Što kad bi pacijenti mogli sami napraviti taj test? Bez skupog osoblja. Inače, testiranje u neurološkoj klinici stoji oko 300 dolara.
So what I want to propose to you as an unconventional way in which we can try to achieve this, because, you see, in one sense, at least, we are all virtuosos like my friend Jan Stripling.
Stoga vam želim predložiti jedan nekonvencionalan način na koji to možemo pokušati postići, zato što smo svi mi u jednu ruku, na kraju krajeva, virtuozi poput mog prijatelja Jana Striplinga.
So here we have a video of the vibrating vocal folds. Now, this is healthy and this is somebody making speech sounds, and we can think of ourselves as vocal ballet dancers, because we have to coordinate all of these vocal organs when we make sounds, and we all actually have the genes for it. FoxP2, for example. And like ballet, it takes an extraordinary level of training. I mean, just think how long it takes a child to learn to speak. From the sound, we can actually track the vocal fold position as it vibrates, and just as the limbs are affected in Parkinson's, so too are the vocal organs. So on the bottom trace, you can see an example of irregular vocal fold tremor. We see all the same symptoms. We see vocal tremor, weakness and rigidity. The speech actually becomes quieter and more breathy after a while, and that's one of the example symptoms of it.
Ovo je snimka koja prikazuje glasnice koje vibriraju. Zdrave su i ovdje netko proizvodi govorne zvukove. Možemo zamisliti sebe kao vokalne plesače baleta zato što moramo uskladiti sve ove govorne organe kad prozvodimo zvukove. Svi imamo gene za to. Na primjer, gen FoxP2. Kao i kod baleta, potrebna je izvanredna razina uvježbanosti. Samo se sjetite koliko treba djetetu da nauči govoriti. Pomoću zvuka zapravo možemo pratiti položaj glasnica dok vibriraju, Parkinson zahvaća glasnice, baš kao i udove. Na donjem ispisu možete vidjeti primjer nepravilnog tremora glasnica. Vidimo iste simptome. Vidimo glasovni tremor, slabost i ukočenost. Govor zapravo postaje tiši i zadihaniji nakon nekog vremena i to je jedan primjer simptoma.
So these vocal effects can actually be quite subtle, in some cases, but with any digital microphone, and using precision voice analysis software in combination with the latest in machine learning, which is very advanced by now, we can now quantify exactly where somebody lies on a continuum between health and disease using voice signals alone.
Ovi govorni učinci mogu biti vrlo suptilni, u nekim slučajevima, ali s bilo kojim digitalnim mikrofonom te koristeći precizni program za glasovnu analizu u kombinaciji s najnovijim dostignućima u strojnom učenju, koje je veoma napredovalo, sad možemo točno izmjeriti gdje se netko nalazi na pravcu između zdravlja i bolesti, i to koristeći samo glasovne signale.
So these voice-based tests, how do they stack up against expert clinical tests? We'll, they're both non-invasive. The neurologist's test is non-invasive. They both use existing infrastructure. You don't have to design a whole new set of hospitals to do it. And they're both accurate. Okay, but in addition, voice-based tests are non-expert. That means they can be self-administered. They're high-speed, take about 30 seconds at most. They're ultra-low cost, and we all know what happens. When something becomes ultra-low cost, it becomes massively scalable. So here are some amazing goals that I think we can deal with now. We can reduce logistical difficulties with patients. No need to go to the clinic for a routine checkup. We can do high-frequency monitoring to get objective data. We can perform low-cost mass recruitment for clinical trials, and we can make population-scale screening feasible for the first time. We have the opportunity to start to search for the early biomarkers of the disease before it's too late.
Kakvi su ovi testovi temeljeni na glasu u usporedbi sa stručnim kliničkim testovima? Pa, i jedni i drugi su neinvazivni. Neurološki test je neinvazivan. Oba koriste postojeću infrastrukturu. Ne morate graditi nove bolnice da biste to napravili. I oba su točna. No, uz to, testovi utemeljeni na glasu nestručni su. To znači da ih možete izvesti sami. Vrlo su brzi, potrebno je najviše 30-ak sekundi. Veoma su jeftini i svi znamo što se događa. Kad nešto postane vrlo jeftino, također postane masovno mjerljivo. Ovo su neki čudesni ciljevi s kojima se sad možemo nositi. Možemo smanjiti logističke teškoće s pacijentima. Nema potrebe za odlaskom u bolnicu na rutinski pregled. Možemo provoditi česte nadzore kako bismo dobili objektivne podatke. Možemo jeftino i masovno pronalaziti subjekte za kliničke pokuse i po prvi put je izvedivo napraviti procjenu na razini cijelog stanovništva. Imamo mogućnost početi tražiti rane biomarkere bolesti prije nego bude prekasno.
So, taking the first steps towards this today, we're launching the Parkinson's Voice Initiative. With Aculab and PatientsLikeMe, we're aiming to record a very large number of voices worldwide to collect enough data to start to tackle these four goals. We have local numbers accessible to three quarters of a billion people on the planet. Anyone healthy or with Parkinson's can call in, cheaply, and leave recordings, a few cents each, and I'm really happy to announce that we've already hit six percent of our target just in eight hours. Thank you. (Applause) (Applause)
Kako bismo učinili prvi korak prema tome, danas lansiramo Parkinsonovu glasovnu inicijativu. Zajedno s tvrtkom Aculab i stranicom PatientsLikeMe ciljamo na snimanje jako velikog broja glasova diljem svijeta kako bismo skupili dovoljno podataka da se počnemo baviti s ova četiri cilja. Dostupni su nam lokalni telefonski brojevi kojima pristup ima 750 milijuna ljudi u svijetu. Svi zdravi, ili koji imaju Parkinsona, mogu jeftino nazvati i ostaviti zapise, koji stoje po nekoliko centa, i vrlo sam sretan što mogu reći da smo već dostigli 6% od našeg cilja, za samo osam sati. Hvala. (Pljesak) (Pljesak)
Tom Rielly: So Max, by taking all these samples of,
Tom Rielly: Dakle, Maxe, uzimajući sve ove uzorke od,
let's say, 10,000 people, you'll be able to tell who's healthy and who's not? What are you going to get out of those samples?
recimo, 10 000 ljudi, moći ćete reći tko je zdrav, a tko nije? Što ćete dobiti iz tih uzoraka?
Max Little: Yeah. Yeah. So what will happen is that, during the call you have to indicate whether or not you have the disease or not, you see. TR: Right. ML: You see, some people may not do it. They may not get through it. But we'll get a very large sample of data that is collected from all different circumstances, and it's getting it in different circumstances that matter because then we are looking at ironing out the confounding factors, and looking for the actual markers of the disease.
Max Litlle: Da. Dakle, kako to ide. Tijekom poziva morate reći imate li bolest ili ne. TR: Dobro. ML: Vidite, neki ljudi možda to ne učine. Možda ne prođu kroz to. Ali dobit ćemo jako velik uzorak podataka prikupljen iz mnogo različitih okolnosti, a upravo je to važno, zato što želimo riješiti zbunjujuće faktore i tražiti stvarne znakove bolesti.
TR: So you're 86 percent accurate right now?
TR: Dakle, trenutna točnost iznosi 86%?
ML: It's much better than that. Actually, my student Thanasis, I have to plug him, because he's done some fantastic work, and now he has proved that it works over the mobile telephone network as well, which enables this project, and we're getting 99 percent accuracy.
ML: Mnogo je bolje od toga. Zapravo, moj učenik Thanasis, moram ga spomenuti zato što je napravio fantastičan posao i dokazao da funkcionira i preko mobilnih mreža, što omogućuje ovaj projekt. Dobili smo točnost od 99%.
TR: Ninety-nine. Well, that's an improvement. So what that means is that people will be able to — ML: (Laughs) TR: People will be able to call in from their mobile phones and do this test, and people with Parkinson's could call in, record their voice, and then their doctor can check up on their progress, see where they're doing in this course of the disease.
TR: Devedeset devet. To se zove poboljšanje. Dakle, to znači da će ljudi moći – ML: (Smijeh) TR: Ljudi će moći nazvati sa svojih mobitela i napraviti ovaj test. I ljudi sa Parkinsonom će moći nazvati, snimiti svoj glas i onda će njihov liječnik moći provjeriti njihov napredak, vidjeti gdje su u tijeku bolesti.
ML: Absolutely.
ML: Apsolutno.
TR: Thanks so much. Max Little, everybody.
TR: Hvala puno. Maxe Little, ljudi.
ML: Thanks, Tom. (Applause)
ML: Hvala, Tome. (Pljesak)