Health Futures: Variations on a Theme

Today the majority of our health insights come from doctors and fitness instructors. We explain our goals and they make recommendations on how to get there: exercise more, eat more vegetables, and get more sleep. To track our progress, we manually record our measurements, diet, and workout routines. As we understand how our behavior affects our health, we can get more specific prescriptions and start seeing better results.

Mobile and wearable technology are the next iteration of this system. Our phones now have sensors that track physical activity. Wristbands and watches measure heartbeat and sleep habits. When paired with fitness apps, the devices we carry every day plot colorful graphs of our data and send it to the cloud for processing. We can track more behavior than ever before and, while there is still manual record keeping, this additional data helps get at specific insights faster.

Instead of looking outward on health issues, mobile technology is allowing users to take more control for themselves. This trend will continue as we collect more data and are able to derive more actionable insights. Projekt202 and The Livestrong Foundation put together a panel explore this kind of future. They discussed how fitness apps are used today, strategies for making them better, and what to expect once this technology becomes more commonplace.

Fitbit's personalized fitness dashboard. Source: Fitbit Inc.

Fitbit's personalized fitness dashboard. Source: Fitbit Inc.

In an event about data, it was a little shocking to hear the panelists say that the numbers are not important, at least not right now. It’s more important that users get insight into habits that might otherwise be invisible. Take number of steps, for example. At the end of the day, we can only say if we walked a lot or a little. We don’t think of our activity in precise measurements; any number is going to sound a bit arbitrary. The value in today's steps is comparing it with yesterday’s steps. As long as the measurement is consistent, we can observe trends over time, set goals, and track progress. While it’s important to walk often, there is no one size fits all. The numbers are specific to the user. Some people will realize they walk more on some days than others. Some people don’t walk enough at all. Access to this data will help them to be more proactive in the future.

The goal is to improve poor behavior. One of the panelists, Robin Krieglstein, says behavior can be broken down into motivation, a trigger, and an action. Developers can influence motivation by understanding the emotionality of the data. People feel more connected to activity data because it’s a kind of representation of themselves. With real-time processing, developers should find the moments when the data is most meaningful and present it as a trigger to inspire action. A simple example is to send a notification when the user has been more active today than they were yesterday at the same time. Encourage them to keep going. Or maybe send a notification when the user has been sitting for too long. You can get the user to go for a walk, moving them closer to their activity goals.

Kyle Samani, CEO of Pristine, thinks the second use case might come up more often. Sometimes, the data is going to be depressing. Few people are as active as they should be and it’s a bad user experience to open an app to be reminded of that. For apps to be successful, he suggests, they have to account for this failure and fold it into a narrative. Lark does this especially well. The secret is that it functions more like a messaging app than anything else. Users can explore their fitness habits by starting a conversation with a “personal trainer” robot, complete with typing indicators. Lark has a great way of saying, “Yeah, you’re not walking as much as you should, but you did better than yesterday.” The personal approach is motivating and reassuring, the perfect cushion for missed goals.

No matter how friendly, it feels a little weird to entrust a computer with health advice. Luckily, no one suggests people stop seeing a primary physician once they buy a Fitbit. It’s a supplemental device. The panelists were quick to note that while doctors have a deep knowledge of medical conditions, they are much less familiar with individual habits and lifestyles. Prescriptions are based off of general symptoms and general case studies. The data that comes out of an activity tracker is unique to the individual and can be used to give a more personalized prescription.

Some people call this precision medicine. It’s the idea that doctors will be able to pinpoint a patient's exact needs for their exact circumstance. However, Maninder “Mini” Kahlon, Vice Dean for Partnerships and Strategy at The University of Texas Dell Medical school, prefers the term precision health. She argues that doctors don’t just prescribe medicine, they prescribe lifestyle changes. Diabetes was used as an example where frequent exercise is important. But not every diabetes patient has the same lifestyle. More advanced data can give doctors the information they need to make the right kind of recommendation. If the patient is largely sedentary, it’s easier to confidently prescribe less medication and more exercise because the reality is that people rarely self-report as harshly as they should.

While in its infancy today, precision health will become more established with more data. Apple wants to help accelerate this process. Two weeks ago they launched a platform that helps medical institutions create survey-style apps. The platform, ResearchKit, utilizes the built in sensors and data collected in the iPhone and sends it to researchers. Medical institutions, long plagued by the slow pace of these studies, can now enter the world of continuous development. Data is available almost immediately and, as insights present themselves, they can iterate on the study to learn even more.

The greatest innovation of ResearchKit is that it makes these studies available in the app store. Rather than exhausting medical resources to find people, participants now come to them. In one day Stanford signed up 11,000 people for their cardiovascular study. Before ResearchKit, it would have taken a year and relied upon over 50 medical centers across the country. In the words of Alan Yeung, medical director at Stanford Cardiovascular Health, “that’s the power of the phone."

Stanford's My Heart Counts cardiovascular app, Source: Apple Inc.

Stanford's My Heart Counts cardiovascular app, Source: Apple Inc.

Precision health will also become more present in the fitness apps we use today. With more than our individual data at our disposal, we can understand our activity in a much greater context and make much healthier decisions. Christian Hernandez presents a great example of this in his post, The Age of Context:

In the future my Jawbone won’t simply count my steps, it will also be able to integrate with other data sets to generate personal health insights. It will have tracked over time that my blood pressure rises every morning at 9:20 after I have consumed the third coffee of the day. Comparing my blood rate to thousands of others of my age range and demographic background it will know that the levels are unhealthy and it will help me take a conscious decision not to consume that extra coffee through a notification. Data will derive insight and that insight will, hopefully, drive action.

None of this is a new innovation for healthcare. We are simply layering technology on top of our current way of doing things. As sensors improve and algorithms are refined, all the data we need will become available. The challenge will be how we use that data. Whether that means leveraging smarter notifications or building a narrative, we need to focus on  maximizing insights and encouraging healthier lifestyles. Our health is more than just a bunch of numbers. Hopefully by abstracting more clinical data analysis, prescriptions and interactions will become more personal than ever before.

Notifications at Meetup

Note: This post was originally published as part of my portfolio in December of 2014. I moved it to the blog because it seemed more like a blog post.

The internet has always been good at connecting people and creating meaningful interactions. Meetup has been a part of that narrative since 2002. In fact, while working on the new messaging feature, we found code for a email messaging tool that pre-dated gmail. Thank goodness we weren't using that code, but it was time we thought beyond email.

There is a big push at the company to reevaluate these aspects of the product in light of a more modern approach. Email isn’t always the best way to communicate anymore. With mobile we have a new set of tools in front of us. Messaging apps are a great way to communicate, does that fit into the context at Meetup? We decided yes. What about notifications?

Recently, there has been a lot of talk about how notifications are the future of mobile. Apps can run in the background and surface information exactly when it’s needed. Users can act on notifications right from the notification center. There's no need to open the app. It’s quick, it’s easy, and it compliments my mobile lifestyle.

Companies have been using this to their advantage and making their apps more useful. Foursquare’s app recognizes when I’m at a restaurant and notifies me of the most popular things to order. Later, as I’m leaving, they might recommend a place for drinks nearby. It's catered specifically to me and where I am right now. The way that foursquare uses notifications, it acts as my guide for all the best places to check out in the city.

That’s a good notification. I would just as eagerly open a text from a friend offering a similar recommendation. These notifications are non-intrusive and relevant to me. However, they also illustrate the system that is foursquare. Foursquare takes an input, a location mapped to a restaurant, and produces an output, recommendations. I can open the app for more information, but the value of the app is the real-time service they provide, not the static destination on my home screen.

So, where do notifications fit in at Meetup? Well if foursquare helps you find your place, then Meetup helps you find your people. For one of our internal hackathons, I decided to play with this idea. Maybe we could send a notification when like-minded people were nearby. It sounds great, but there are some complications here.

Many agreed that this was interesting and relevant information, but there isn’t an obvious time to send this notification to users. The input to our system is not as obvious as someone walking into a restaurant.

The solution was to send a notification when a certain threshold was crossed. For example, maybe there aren’t a whole lot of astronomy enthusiasts in Park Slope. However, once 25 people have indicated interest in astronomy, suddenly there may be a much greater desire for a fellow enthusiast to create a group. Maybe not perfect, but a good theory.

The notification would say something along the lines of, “Park Slope now has more than 25 people interested in astronomy.” That sounds compelling, especially for someone who lives in Park Slope and is interested in astronomy. They want to know more. With this kind of input, what kind of output can users expect? They’re curious so they tap through for more information.

Naturally, it would make sense to see and connect with these people. However, in order to meet these people, users would have to create a Meetup group first. (For this thought experiment I decided not to pivot the company’s business model. Meetup is in the business of creating community, not one-off, spur of the moment connections.)

It’s a lot to ask someone to create a Meetup group. Even with the newly launched price points, it’s a big responsibility. I wanted to be sensitive to that. Besides, acting on a notification is supposed to be quick and easy. If a user is going to create a group, I want them to take a little more time and put some thought into it.

But throw all that away. Suppose it’s a highly motivated segment and they’re ready to create a group, name it, write a description, enter their credit card information, and pay for it. Right now. On their phone. It’s mighty presumptuous of me to think that they want to create an astronomy group. Even if they’re interested in astronomy, maybe they would rather start a Double Dutch group. It seems unlikely that I could get that right the first time.

While thinking about this, I stumbled upon Janel Torkington’s article Small Data: Why Tinder-like apps are the way of the future. He argues that cards are a great container for content and the interface Tinder made famous is ideal for making quick decisions. More importantly, these decisions build up over time and that data gives you better insight into what your users actually care about.

I thought I’d give it a shot. When users tap on my notification, they will be taken to the Meetup app where they can see the people in their neighborhood laid out on a card. Rather than asking them if they want to create a group, I ask if they would be interested in bringing these people together. If they say no, we now know a lot more about them. Maybe astronomy isn’t one of their primary interests. If they say no to more than one topic, maybe they don’t want to organize any kind of group. That’s good to know, we’ll stop bothering them. This notification should be relevant.

If they say yes, it still doesn't mean they want to create a group, but they have demonstrated more interest in meeting up. We can track that and follow up later.

They have to make a decision before they can consume any other content in the app. Either way we get a helpful data point. The cost is one tap, or even a swipe. This is the kind of quick and easy decision making that compliments the qualities of a good notification.


You may not see this as part of any future product launch, but I believe it addresses the constraints of the platform well. It sure was a fun hackathon project.

The team at Meetup is working hard to create great notifications. In the next few years an app will only be as good as the underlying system and notifications will allow users to react on any shape screen imaginable. If you want to help craft the future and work for a fantastic company that brings people together, Meetup is hiring!