[MUSIC] Hello, I'm Nicholas Genes, and this lecture will focus on m-health apps, mobile health apps, and really how they're going to make the transition from consumer grade fitness trackers and wellness devices to true medical grade apps that we can safely prescribe to patients. This issue first came to mind years ago, I was watching Eric Topol who is kind of a visionary, he's the head of the Scripps Institute, he's a cardiologist and a big evangelist for m-health and wearables. And he was giving an interview where he said these days I'm prescribing a lot more apps than I am medications, and that really caught me by surprise. This was a few years ago and it made me think what will it really take for apps to have the same degree of safety and evidence behind them that drugs have today. And that's what we'll be focusing on in this lecture. We'll talk a little bit about the potential of apps for patients, barriers to adoption, why they're not in widespread use yet. Problems with apps and different approaches to curating apps and regulating apps. And then finally, I'll talk about an approach that's close to home because it involves some of my work here, using research kits to kind of validate the efficacy of apps. Throughout this lecture we'll be giving anecdotes, examples, and some evidence. When we talk about the potential of medical apps in improving patient care, we're thinking about a few domains. First apps can promote mindfulness of disease that can help patients correlate their symptoms with activities and other factors over time. Apps can promote adherence through pop ups and reminders so that patients can stay on drug regimens or exercise regimens. Apps can help with monitoring, they can passively and actively collect data on patients' activity levels or functions, capabilities. And finally let's not forget that apps are on a phone that can do text messaging and make phone calls, so apps can promote communications and help patients stay in touch with providers. From a hospital perspective, in terms of managing chronic illness, hospitals are gradually taking on the role of insurers. They're becoming more interested in detecting risk and assessing risk of patients. So they're interested in figuring out which patients are at risk of being hospitalized again or just having a rockier course. And so this passive and active monitoring that happens, if it feeds back to a hospital or an insurance company, they're both very interested in knowing this. And the temptation is also there to intervene before a patient decompensates, before a patient has an exacerbation of a chronic disease state, and hopefully to be able to prevent a hospitalization. Why aren't medical apps being prescribed regularly today? Why aren't patients using them and feeding data back to their doctors to inform care decisions and help monitor patients. Well I think first of all, there's not enough evidence to support this practice. There's a lot of reluctance about the safety and efficacy of these apps, and there's not enough financial incentive yet to disrupt practice. A lot of doctors and healthcare systems tend to be set in their ways and unless change is trust upon them, sometimes we don't see that change. I think that change is coming, but we're not there yet. However, from the patient's perspective, I think there's a lot of frustration and a lot of confusion. And when we look at app stores today, we see there are over 100,000 apps divided between the Apple Store and the Google Store. They're categorize broadly based on wellness or a few disease states, but there's no curated section of the stores where they can find trusted apps that had been validated. And it's really very much a Wild West out there where the quality and evidence behind apps is just not immediately available to patients. They can sometimes see ratings and reviews but sometimes those reviews are really not. They focus on the usability, or different features, or whether a new version has disabled some features or changed the menus around, they haven't focused on real medical utility for patients. In different reviews, when people tackle the ten of thousands of apps that are out there, they see most of them are devoted to education and instruction. A few of them, a smaller fraction are devoted to recording symptoms over time, tracking fitness and medication adherence. A smaller fraction than that actually intend to guide patients to making medical decisions and have reminders and alerts to prompt behavioral change. And a very, very small fraction actually facilitate communication with care providers. I'm going to take a moment to go into an example of one domain of apps which is mole checking. Something benign, kind of interesting I think, we all have moles and we worry about them sometimes. It turns out there are a variety of mole checking apps out there and mole tracking apps. You have two on the screen here, one that's from University of Michigan and the other is called Mole Detective. They both make use of a magnifying glass and they both have similar kind of names. One is actually very safe, and responsible, and good for patients to use, and the other is downright dangerous, and there's no easy way to check. The University of Michigan app is very safe and effective, that was the good app. This app prompted users on a regular interval every few months to take a picture of a mole, and to hold their phone in a standardized fashion so that they can capture the mole the same way each time. And they could actually plot the development and progress of the mole. So if it starts to grow, or if it starts to take on dangerous features, a patient could actually make an animation of that mole and show it to their doctor. And their doctor could then refer to a dermatologist, or get a biopsy, etc. The Mole Detective app on the other hand, actually took matters one step further and attempted to actually use computer vision to diagnose a mole, and to actually say whether a mole was malignant or not. Which is pretty cool if it worked, but there was really no evidence that it worked, there were no trials, there was no publications about this. It was just an attempt to use some familiar algorithms to kind of classify moles. And it turned out when it was studied and this was a study from a few years ago. Found that even the best of the Mole Detective apps were routinely missing malignant melanomas, they were misclassifying them as benign. So you can imagine that a patient might be falsely reassured that their dangerous mole was actually Classified as benign by an app, and they maybe didn't seek care when they should have. So, the federal government got involved, the FTC actually stepped in and said that these apps were advertising claims that weren't true, or at least there was no evidence that they were true. So, the apps shut down and had to pay a fine, but who knows what kind of damage was actually done? In addition to some of these health risks, very real health risks, there are other risks to patients beyond just misdiagnosis. For instance, there's breach of privacy. Patient's health records could be hacked or intercepted on the way to a server. There's the risk of just a failure to notify. If you become dependent on an app to program all your medication dosing and schedules and the app fails to properly notify you, you could miss medications. There are symptom-checkers online that are somewhat popular. You download an app that you enter what symptoms you have. And you trust the app to steer you to the right healthcare setting or to tough it out at home and seek care later when things get worse. But when these symptom checkers were reviewed, it turned out a lot of them had arrangements with local healthcare providers. And so they would sometimes inappropriately steer you to healthcare settings that you might not actually need to go to. And, finally, there's just the risk of false positives, and we'll come back to this later in this talk and another talk. Sometimes an app can detect a condition that you might not have, leading to an unnecessary work up, unnecessary doctors visits and unnecessary testing. And all of those tests also have some risk and downsides to them. So, for patients, there's really a lot of concern out there as they're browsing the app store. And too often what happens is that an app is released and then years later the study comes out that talks about that app's effectiveness. When you think about it, this is the exact opposite of how drugs are regulated. Usually a drug is trialed in a very controlled setting. It has to demonstrate its effectiveness and only then is it released to the broad public. So, keep in the back of your mind the possibility that maybe we could use this model of studying apps just like we do for studying drugs. But in the meantime, I'm going to talk about a few other approaches to curation and what an ideal mHealth app could look like. Some of my colleagues in Boston published a nice article a few years ago, In Search of a Few Good Apps. The authors wrote that the potential for apps will only be realized if patients and clinicians trust the apps. If apps are known to be effective. And if apps can communicate securely and meaningfully with electronic health records and personal health records. And establishing an unbiased review and certification process is a key step in helping mHealth apps achieve their potential. We're really not there yet, even now this was written years ago. We're taking steps in that direction but it's still a ways to go. There have been some early efforts in having this independent review, this independent authority for judging apps. The HON code, the HON code, it stands for Health On the Net Foundation. This is a non-profit, non-governmental organization that operates worldwide. It first rose to prominence in the 1990s because it was there to validate websites. And then later, bloggers and social media figures that were spouting health information. And this was a process where you would submit your website or submit your social media stream, and an independent panel would review you, review your qualification, review the postings. And then if you passed there was a certification process which involved disclosing conflicts of interest and giving your real name and giving your academic credentials and a host of other things. Then you would get to display the honor code badge, and you could be a trusted resource. The process was free. But it was totally dependent on the good will of its independent reviewers. And so it takes time to get certified. It's a long waitlist. And for apps the waitlist is even longer. If you submit an app to the Honor Code, it might take over a year, up to 18 months to get certified. And that's kind of a lifetime in the apps stores. So, it's really not a viable solution for a lot of new developers. Happtique promised a different approach, where if a developer paid a fee, then they would get reviewed within a 30-day time. And, again, they had a list of criteria that they would use for evaluating medical apps. And then they would have a little boutique, a Happtique boutique, that would showcase the best of the apps that they reviewed. However, there was a bit of a scandal when their first class of apps that they proudly displayed actually had a few apps in there that shouldn't have passed. It seems like maybe they were rushed through the review process, and when this was exposed, this lead to the downfall of Happtique. Although, some people are trying to resurrect it in a different form. In the meantime though, the federal government in the US and other governments have been attempting to determine really what makes an app a consumer device that is really not prone to regulations. Versus what is something that should be regulated and should be scrutinized closely. First, European Medical Device Directive came out, and the FDA really borrowed upon that and adapted it for the US audience. I encourage folks to read the FDA guidance because it's actually very readable, very clear and easy to understand device, document I should say. And the FDA, what they were saying is that they intend to apply their regulatory oversight only to mHealth apps. Mobile apps that are medical devices. Or apps that transform your smartphone or your tablet into a medical device. And whose functionality could pose a risk to a patient's safety if the mobile app were not to function as intended. So thinking back to our mole detector, I think the University of Michigan app, the safe, responsible app that takes pictures of moles, that doesn't function as a medical device. It's functioning as a camera with a reminder attached to it. And it can prompt good medical outcomes. And it can prompt you to see your doctor and look into issues. But it's not functioning as a medical devices. Whereas the Mole Detective app, the one that tried to diagnose your melanoma, that really is trying to turn your smartphone into a medical device. And so that would fall under FDA scrutiny. The FDA rules have been in place for some time, there have been some skirmishes and some issues and some open letters from the FDA to different companies. But for the most part, I think it's well understood in the development community what will be under FDA scrutiny and what will not. However, there are also new approaches to app curation that are being studied and developed, and some of them involve colleagues of ours here at Mount Sinai. Our friend Ashish Atreja has started an app lab and also something NoDEhealth. This is the Network of Digital Evidence for Health. It's a pure discussion of the evidence behind the apps. So a lot of scientist, developers, doctors and patients are able to review apps and comment on their usability. But more importantly, review the literature, what's been published about this apps and the evidence behind them. And so, when the doctor browses these reviews, or patient browses these reviews, they can feel confident that they're using an app that has been vetted by the community. And can learn more about this at nodehealth.org or visit the sinaiapplab.org. Others have tackled the issue of just what is, what makes an [INAUDIBLE] app usable and well reviewed? And some people have taken very academic approach and breaking down an app's various features in studying them and seeing, and doing a lot of surveys of patients. Turns out that patients seem to find a few features of most importance, that is the ability to export the data and the ability to upload a medical plan, like a decision tree. So that when they're faced with new situations, they can council that document and use it. In the US, we don't have a federal repository of apps that have been reviewed by the FDA or approved by FDA. However, in the UK, they seem to be heading towards that. The National Health Service, the NHS, said to be developing an app store where users can browse, see reviewed apps, and actually download them directly from the NHS. We'll be watching that closely for further developments. But as I mentioned before, instead of releasing apps into the public first and then studying their efficacy, or seeking FDA approval. What if things were more done like they are with drugs and medications. What if we instead, consented potential app users first? Told them that they were participating in a trial, learn about them, the demographics there, the background information there, the medication history. And then through analytics, studied how they were using apps. And what interventions, what parts of the apps they found most helpful in managing their care. Then we could actually make claims as to an app's capabilities and effectiveness, and know which populations would be best served by certain apps. And that potential was realized last year when Apple introduced ResearchKit in March of 2015. And I'm proud to say I was very involved in this system from the beginning. ResearchKit is a framework where app developers can easily make modules so that patients can consent to participating in a clinical trial. And their data can be securely collected and stored. So the patient doesn't have to worry about risk of data breaches. We entered into a partnership with some local developers, LifeMap and with Apple in developing our ResearchKit app which is called Asthma Health. And it's a research study. The PI is, the principal investigator is Yvonne Chan. And when you download the app, you're allowed to watch a video, you learn about the study, and if you choose, you can then consent and participate in the app. And the app then collects a lot of data about you. You can even link your health kit data. So your step counts, your activity levels and so forth. You answer a lot of questions about your history, your medications, your hospitalizations. And there's a daily survey about how you're feeling, how you're coping with your asthma. Whether you need your medications, whether something you feel is triggering your asthma. And they can view a dashboard of key metrics. They can look at local air quality. They can look at their step counts, their peak flow data, how their symptoms have been changing over time. And the app has proven to be pretty popular in part because it debuted in a very public form. But thousands of people downloaded the app and consented. In fact, even though our app development was relatively inexpensive, we were able to recruit as many patients as some of larger federal trials on asthma. And it turns out that the data that we are getting mirrors a lot of data that has been collected in the traditional sense. So this was very validating for us that we were able to collect data via smart phone, and have it match up well with what's been seen in prior large scale studies. And we are finding data from all 50 states about asthma rates. And it does seem to match up well with what has been collected before. Plus the testimonials have been very gratifying. It turns out, users really do appreciate being able to plot their symptoms over time. It gives them new insights into their healthcare and lets them see trends, and triggers that they might have missed. It is true that over time, we do tend to lose users. They get a little disengaged over time. It's true we're not competing really with other clinical trials that sometimes financially reward patients. Instead, we're competing with other apps that are much more engaging, like Facebook or Candy Crush. However, we have found that sicker patients, patients with objectively worse asthma, tend to stick with the app longer. Which also is gratifying and is also the opposite of what happens in a lot of clinical trials where the sicker patients tend to drop out earlier, because it's difficult to check in with a principal investigator, or drive across town to submit your data. So we think that, by making it really easy to participate in this trial, we're capturing a cohort of patients that would otherwise be lost. And the number one concern of our patients has been that asthma limits their activity. And it turns out that those that are sticking with the app over time actually have higher activity levels. So, it seems that by visualizing their trends over time, by being more mindful of disease, our app users are overcoming their biggest obstacle and becoming more active, which is great. So if you wanted to use our app as, like one who could built it as something that could improve the activity level of asthma sufferers, we'd actually be ready to do that. We have a broad base of data, we have a lot of characteristic on our users, and we've gone through a process, say for an effective process to demonstrate the app's utility. This is how I think apps are going to be released and studied in the future. There'll be consenting users in advance, they'll be collecting data and then only after that data is analyzed and published, will we be able to say that this is an app that's effective. When apps, whether we're talking about asthma, Parkinson's disease, breast cancer, any other research kit apps that have been approved. They'll go through this process and then eventually, they'll be added to a hospital formulary or a special section of the app store. And we'll be able to confidently and safely prescribe these apps to the patients that need them. Thank you. [MUSIC]
