Latest / Lab to Lives / Dr. Peter Small on Cough Science and the Impact of Acoustic Diagnostics

Transcript

Dr. Peter Small on Cough Science and the Impact of Acoustic Diagnostics · Lab to Lives

• 0:21
  Speaker 1: Welcome to Transformation in Trials.
• 0:23
  Speaker 1: This is a podcast exploring all things transformational in clinical trials.
• 0:27
  Speaker 1: Everything is off limits on the show and we will have guests from the whole spectrum of the clinical trials community, and we're your hosts, ivana and Sam.
• 0:35
  Speaker 1: Welcome to another episode of Transformation in Trials.
• 0:39
  Speaker 1: Today we're going to focus on the topic of COF and today in the studio with me I have Peter Small, who is the Chief Medical Officer of Haif and an acoustic epidemiologist.
• 0:51
  Speaker 1: Welcome, peter.
• 0:53
  Speaker 2: Thank you so much.
• 0:56
  Speaker 1: And Peter, just setting the stage for us.
• 0:57
  Speaker 1: Can you tell us more about how COF has traditionally been measured in clinical studies?
• 1:03
  Speaker 2: Yeah, I would say in a word, badly.
• 1:05
  Speaker 2: You know, cof has been caught in this vicious cycle.
• 1:09
  Speaker 2: It was hard to measure and so companies didn't include it in trials and because it hadn't been brought to the FDA, there was no clear process for establishing COF as an endpoint.
• 1:22
  Speaker 2: I think you know the good news is that there's been a lot of attention to antitussive medications, based on some increased understanding of the pathogenesis, some new neural mechanisms and receptors that can be blocked, and I think you know, for the first time in 60 years we're looking at some exciting drug candidates.
• 1:44
  Speaker 2: And but the disconnect between the COF community and the FDA was really brought into sharp focus just the other day on the FDA pulmonary advisory committee on November 17.
• 2:00
  Speaker 2: The patient reported outcomes which the COF community is quite enamored with, was dismissed out of hand by the FDA as not validated.
• 2:15
  Speaker 2: And the way that COF is currently counted in the FDA trials is by wearing a device on your belt, a microphone attached to your chest, a microphone on your lapel, and it basically records a day in your life and then that sound is listened to by humans who count the COFs.
• 2:35
  Speaker 2: It's a primitive method and it came under a lot of discussion at this FDA meeting as well.
• 2:43
  Speaker 1: Can you tell us more about what the discussions were about?
• 2:47
  Speaker 2: Well, I think there was, you know, some genuine disagreement about whether or not the patient reported outcome was validated.
• 2:54
  Speaker 2: There were some questions about the device, which was FDA cleared as a COF recorder but not counter, and the mechanism by which the recordings were turned into COF counts was discussed in quite a bit of detail.
• 3:14
  Speaker 1: So that sounds like it's actually most of the approach to measuring COF that came under scrutiny.
• 3:18
  Speaker 2: Absolutely.
• 3:20
  Speaker 2: I mean it's clear that the FDA is sort of adrift right now in terms of their alignment with industry on how these exciting new molecules will get validated and approved and that's going to delay significantly getting these really important medicines to the people who need them.
• 3:46
  Speaker 1: So it's a question of how do we measure COF and then also how do we apply medicine to that COF.
• 3:52
  Speaker 1: And if we can measure it, then it's very hard to argue that we can actually make it better.
• 3:56
  Speaker 1: Are there other ways?
• 3:59
  Speaker 1: Are there other schools of thought for how we can measure COF?
• 4:03
  Speaker 2: But the entire focus of HEIFF is continuous, passive, unobtrusive, privacy-preserving COF counting.
• 4:11
  Speaker 2: And COF is complicated and you need continual monitoring because of a couple of different effects.
• 4:22
  Speaker 2: Cof is under volitional control and therefore things like the Hawthorne effect.
• 4:28
  Speaker 2: When you tell somebody I'm gonna put some device on you cough, normally it's sort of like don't think about an elephant.
• 4:35
  Speaker 2: And that issue actually wasn't even addressed in that committee.
• 4:41
  Speaker 2: And the other issue that wasn't addressed is that cough is highly stochastic and people have good days and bad days and in a study if you're just measuring 24 hours, you might catch somebody on a good day at enrollment, A bad day on your follow-up evaluation and entirely miss the impact that that drug has.
• 5:02
  Speaker 2: And, in contrast, with continuous monitoring you get a really rich and robust data.
• 5:09
  Speaker 2: It's amenable to a lot of analysis using simple standard change point analysis to see whether it's statistically changed.
• 5:20
  Speaker 2: You can look at other aspects of cough, like are people coughing in big bouts which may be more problematic for them.
• 5:29
  Speaker 2: So we think that continuous cough monitoring will not only have a really powerful impact in the management of these patients, but more relevant to today's conversation is that it can accelerate the development of new antitussives.
• 5:49
  Speaker 1: Absolutely.
• 5:50
  Speaker 1: How do you measure cough continuously?
• 5:54
  Speaker 2: So we're an acoustic AI company, and if we have a microphone and a chipset which can either be in a phone or a watch or any number of other devices, we monitor sound level and when there's an explosive sound, we grab a half a second of it and then that half second is passed through a second algorithm which recognizes coughs.
• 6:22
  Speaker 2: This is an acoustic AI program that's been trained on 20 million cough, like sounds that have been listened to and annotated as cough or not by humans, and so it's using convolutional neural network, sort of the same technology that Google photos uses to recognize a cat.
• 6:44
  Speaker 2: No one said to that program oh, cats have 20 years, but they tend not to flop over.
• 6:49
  Speaker 2: They just present millions of pictures of cats and they can recognize a cat, and that's essentially what we do, and the algorithm is based on this massive data set and in our FDA enabling trials it's highly accurate.
• 7:07
  Speaker 2: We pick up 90% of coughs, and with one false positive an hour, and that's as people go about their activities of daily living riding on the subway, watching TV Then that's basic, that's the core of our technology.
• 7:24
  Speaker 1: And Peter, why is cough a problem?
• 7:28
  Speaker 2: Well, cough is actually a big problem.
• 7:31
  Speaker 2: For some people it's more than an annoyance.
• 7:35
  Speaker 2: Most of the time when people have a cough it's a transient condition they have a upper respiratory infection.
• 7:41
  Speaker 2: But for a significant number of people probably about as many people have chronic cough as have asthma.
• 7:48
  Speaker 2: They cough constantly.
• 7:51
  Speaker 2: By constantly I mean hundreds of times a day and this is more than annoyance.
• 8:01
  Speaker 2: These bouts can result in vomiting, fractured ribs.
• 8:05
  Speaker 2: Many women have urinary incontinence when they're coughing, so it's quite debilitating, and it has been for quite a while.
• 8:16
  Speaker 1: And during the pre-call we also talked about being burdened with cough during COVID and also post-COVID.
• 8:22
  Speaker 1: How has that contributed to it being a worse condition to have?
• 8:28
  Speaker 2: Yeah, you know, it was always embarrassing to have a cough, but now it's been stigmatized in a way that I mean if someone coughs in a, has a coughing fit in a checkout line at the grocery store, people look at them like they just pulled out an AR-15.
• 8:45
  Speaker 2: I mean, it's, it's it.
• 8:47
  Speaker 2: You know, people recognize the cough is.
• 8:49
  Speaker 2: Is can be a a a a the way that diseases transmit.
• 8:55
  Speaker 2: But for these poor folks who have chronic cough, it's not infectious and yet it's extremely socially isolating.
• 9:02
  Speaker 2: People can't go to the movie, they, they.
• 9:05
  Speaker 2: At the FDA hearing they had 13 patients talk and it was, it was tragic.
• 9:10
  Speaker 2: I mean there are people who at work had been moved into corner offices and told not to come to meetings.
• 9:16
  Speaker 2: There are people who do telemarketing or had to quit their job because when they started talking they would cough.
• 9:20
  Speaker 2: So it's, it's, it's um cough is is a big deal.
• 9:28
  Speaker 1: Is chronic cough, the only reason why people struggle with cough.
• 9:33
  Speaker 2: Um, well, so it's it.
• 9:36
  Speaker 2: There's been an interesting transition in the understanding of cough, and it used to be that people really thought that cough was a symptom of another disease.
• 9:46
  Speaker 2: Things like GERD or asthma, uh, post nasal drip would cause this chronic cough, but what's become known in the last decade is that cough chronic cough itself is a disease and it's a neural hypersensitivity and it's a little bit like, um, like chronic pain.
• 10:05
  Speaker 2: You know, it's, uh, the, the receptors for the coughing get upregulated.
• 10:11
  Speaker 2: The more people cough, the more they want to cough.
• 10:13
  Speaker 2: Um, it's, uh, it's.
• 10:17
  Speaker 2: It's been a very interesting transition, and and and one of the interesting aspects that I didn't know before I got became a cough nerd is that, um, you know, there, if you see a speech pathologist and you have this condition, four hours with them can reduce your cough 40% in a well done trial.
• 10:40
  Speaker 2: So there's, it opens this other avenue for therapy, which are digital therapeutics, which is one of the things that we're working on as well, which is the idea that If, if you have a chronic coffee, have this constant tickle in your throat and you think, oh, coffee will go away, but it's like scratching a mosquito bite when you cough, it actually makes it worse.
• 11:01
  Speaker 2: So by just training people that you know cough is not necessary.
• 11:06
  Speaker 2: It's not helpful giving them some techniques to suppress it and then in in in our Digital therapeutic, giving them real-time feedback, saying like, oh, you've been coughing a lot, are you trying, are you using your techniques?
• 11:19
  Speaker 2: And we think that this is a whole new avenue, either for treatment or Perhaps as a companion app to be used in combination with with molecules.
• 11:32
  Speaker 1: Well, that's I.
• 11:33
  Speaker 1: That was a very large number that you can reduce by by therapy.
• 11:36
  Speaker 1: That's, that's impressive.
• 11:38
  Speaker 1: I'm also wondering so we talked about that these two therapeutics can make a difference, and if you can actually diagnose cough, you can actually also do something about it.
• 11:47
  Speaker 1: Well, what kind of treatments for cough do we have today?
• 11:51
  Speaker 2: Well, right now we don't have much of anything.
• 11:53
  Speaker 2: So the last time that the FDA proved in a cough medicine was it 60 years ago, whoo and Since then most of the treatments are other drugs that are used off-label.
• 12:09
  Speaker 2: So narcotics are a very effective way of suppressing cough but obviously have a lot of side effects.
• 12:16
  Speaker 2: And there are other drugs like gabapentin and others that cough experts use, but none of them actually worked very well, which is why I was so disappointed that that this new drug Didn't fare better in the FDA hearing.
• 12:32
  Speaker 1: But it sounds like we're the way we've been thinking about cough that that is potentially the problem and not just the developing medicines for it.
• 12:41
  Speaker 2: Well, it's a great example where some fundamental biology, biological insights have opened up whole new avenues of alleviating suffering hmm, that's very interesting.
• 12:53
  Speaker 1: I kind of want to broaden the conversation a little bit more because a you you have worked with acoustic diagnostics for a while.
• 13:00
  Speaker 1: What is happening in general in acoustic diagnostics?
• 13:03
  Speaker 1: Where else can can this kind of diagnostics be applied?
• 13:07
  Speaker 2: Well, it's interesting, you know, if you think about Artificial intelligence and these, these deep learning Techniques, they're really dependent on massive amounts of data and you need a very data-rich signal, and and.
• 13:25
  Speaker 2: The richest signal in medicine, and the first target were images, and so AI is already transforming interpretation of Images.
• 13:37
  Speaker 2: You know, if you look at dermatology, radiology, pathology, ophthalmology, I mean, these are all just falling one after another, great technologies to assist in diagnostics.
• 13:51
  Speaker 2: The second richest signal is actually sound, and and and, in addition to being Very information rich, it's quite easy to obtain, given that we're all walking around with high-quality Microphones in our pockets or we have them on our smart speakers and and so I think that really acoustic AI is.
• 14:16
  Speaker 2: It's kind of the breaking wave in medical AI in my mind right now.
• 14:24
  Speaker 1: That's that's really interesting.
• 14:25
  Speaker 1: And if we talk about Hive specifically, I know that we we've talked about like how it can contribute to clinical research, but can it also be used by patients themselves?
• 14:41
  Speaker 2: Yeah.
• 14:41
  Speaker 2: So you know, our core technology are these cough recognition algorithms and we they're available in a few different form factors.
• 14:53
  Speaker 2: We have a freely available Health and Wellness app that anyone can download Just go to your Android or iOS app store and that will monitor your cough and give you a lot of information.
• 15:13
  Speaker 2: Now, it's not a medically clear device, so it's not meant for making diagnoses or managing, but it provides patients with much better understanding of their cough and it's allowing patients to recognize that they actually are coughing 400 times a day and have more informed conversations.
• 15:32
  Speaker 2: So, for example, when I diagnosed myself with COVID, I called my doctor and I said, hey, you know, I have COVID, I'd like some packs of it.
• 15:41
  Speaker 2: And she said, well, you're not at risk.
• 15:43
  Speaker 2: And I said, well, I'm a coffer.
• 15:44
  Speaker 2: And she goes, yeah, yeah, everyone costs.
• 15:46
  Speaker 2: Like no, no, I'm.
• 15:47
  Speaker 2: I caught 457 times yesterday and she's like oh my God, and wrote the prescription.
• 15:53
  Speaker 2: That wouldn't have happened without objective data.
• 15:56
  Speaker 2: Yeah, bringing data through our health and fitness app to people's understanding of their cough, what their triggers are and what the pattern is is is, I think, something which we're feeling there's a lot of traction for.
• 16:13
  Speaker 2: The second instantiation is in a dedicated watch sort of a fit bit for cough, if you will and that is we're we'll be submitting our FDA package at the end of this year and so we anticipate that we will be FDA cleared as a medical device that can be used for diagnosing and managing patients towards the middle of next year.
• 16:41
  Speaker 2: And I think that's going to really improve the approach to cough, which is often empiric trials.
• 16:48
  Speaker 2: If you go to your doctor you say I've got a chronic cough, they'll guess that it's good and give you some pepsid, and but it's an unmeasured empiric trial.
• 16:58
  Speaker 2: So now it allows doctors and patients to work together on, on, on saying well, how much did it get better?
• 17:04
  Speaker 2: And then the final form factor for us is we have a very light SDK that can run on other devices smartwatches, consumer watches, smart speakers and we have a number of partnerships in the works there.
• 17:23
  Speaker 2: And I think what's going to happen is first cough will be a little bit like glucose monitoring.
• 17:31
  Speaker 2: It will be something where, and if you have the condition, you can monitor it yourself.
• 17:36
  Speaker 2: And then I think after that it will become like step counting.
• 17:41
  Speaker 2: It'll be ubiquitous, it'll be on all consumer devices and at that point you'll just pay attention to it when you care.
• 17:50
  Speaker 1: That's super interesting.
• 17:51
  Speaker 1: I know that you're based out of Seattle.
• 17:55
  Speaker 1: Do you think that the whole ecosystem for medical devices that there kind of is around the Seattle area does that contribute positively to the development of a solution like Hype?
• 18:08
  Speaker 2: Well, it's certainly been helpful to us.
• 18:10
  Speaker 2: I should be clear.
• 18:12
  Speaker 2: We're a globally distributed company.
• 18:16
  Speaker 2: We have our data scientists are in Spain, our engineers are in the Ukraine, our user interface person is in Canada.
• 18:28
  Speaker 2: I mean we're scattered statisticians in Tennessee, I mean we're all over.
• 18:31
  Speaker 2: And so when people ask where we're located, I usually say in the country of Slack.
• 18:37
  Speaker 2: But for me personally, living in Seattle has been a huge advantage because it's a very innovative, rich place, and so when I have a question, you know it's very easy to find somebody and, for the price of a beer, get some really good advice.
• 18:57
  Speaker 2: And then also, you know, we participated at the University of Washington's Creative Destruction Lab accelerator, which was a phenomenon.
• 19:08
  Speaker 2: I mean, we've been involved in a number of accelerators, but CDL was really a cut above in terms of the quality of mentorship and their whole approach to helping young companies.
• 19:20
  Speaker 2: So yeah, I would say for me personally, it's been great to be in Seattle.
• 19:27
  Speaker 1: Oh, peter, I would be curious to learn more about yourself, and how did you end up in this particular space?
• 19:37
  Speaker 2: Well, you know, my background is originally in clinical medicine.
• 19:41
  Speaker 2: I learned how to take care of patients at San Francisco General Hospital at the dawn of the AIDS epidemic.
• 19:48
  Speaker 2: I then moved to Stanford where I was on the faculty, and then up to Seattle where I built and ran the TV program for the Gates Foundation, and the common theme and that kind of Aortic career was, um was tuberculosis.
• 20:06
  Speaker 2: And so I spent 35 years of my life focused on finding people with a cough and getting them tested for TV.
• 20:14
  Speaker 2: And then, you know, I am a cough for myself.
• 20:19
  Speaker 2: I've had a chronic cough.
• 20:21
  Speaker 2: I cough 40 to 80 times a day, and I have for three decades, and so I've sort of been on both sides of the bed on this issue and I think it's it's the, with the excitement that we could bring cough into the era of precision health.
• 20:38
  Speaker 2: Yeah, that really motivated me to jump on board with heif and and it's it's been a.
• 20:47
  Speaker 2: It's been an amazing ride.
• 20:48
  Speaker 2: We've gotten so much done so little time.
• 20:50
  Speaker 2: It's been quite fun.
• 20:53
  Speaker 1: And well, that's.
• 20:54
  Speaker 1: That's an amazing experience in general, and this leads me to the question that we always ask our guests on the show If we gave you the transformation trials magic wand that can change one thing in the life sciences industry, what would you wish for?
• 21:10
  Speaker 2: That's a good question.
• 21:11
  Speaker 2: You know I I would better fund the FDA.
• 21:14
  Speaker 2: I mean, the FDA plays such a critical role in enabling innovation and yet and they have great people you know we've been through many FDA meetings because we're in the new route.
• 21:30
  Speaker 2: This art technology is so unique that there was no predicate.
• 21:36
  Speaker 2: And you know, the people at the FDA have been, they've been great, but they're they're just overwhelmed.
• 21:43
  Speaker 2: And you know it results in long cycles.
• 21:47
  Speaker 2: You know it takes three months for them to answer questions.
• 21:52
  Speaker 2: You know when, at the end of that three months, you have only an hour on the phone and and these brief interactions that are separated by long periods of time are fundamentally a consequence of their work load, and I think it's it significantly slows innovation.
• 22:12
  Speaker 1: I love that wish.
• 22:13
  Speaker 1: No one has ever wished for that before, but I think that would actually make a fundamental difference.
• 22:19
  Speaker 2: I it's um.
• 22:20
  Speaker 2: I won't get into the politics of the role of government, but I think this is a case in which, for those of us who are trying to build better tools to improve health, it's a huge issue.
• 22:36
  Speaker 1: I agree, and Peter, if people want to ask you some follow up questions or want to learn more about yourself or hyph, where can they find you?
• 22:45
  Speaker 2: Yeah, so you know, our website is easily discoverable with a Google search on cough monitoring or hyph or whatever.
• 22:52
  Speaker 2: And then people are welcome to email me directly.
• 22:56
  Speaker 2: My email address is peter at hyph.
• 23:04
  Speaker 1: Thank you so much for coming on the show.
• 23:06
  Speaker 2: Thank you, it's been pleasure.
• 23:14
  Speaker 1: You're listening to transformation in trials.
• 23:16
  Speaker 1: If you have a suggestion for a guest for our show, reach out to Sam Parnell or Ivana Rosendale on LinkedIn.
• 23:24
  Speaker 1: You can find more episodes on Apple podcasts, spotify, google podcasts or in any other player.
• 23:30
  Speaker 1: Remember to subscribe and get the episodes hot off the editor.