Search CDOQ

Friday, August 30, 2013

Fitness on Your Wrist: The Genesis of a Wearable Device

Four years ago, I took a job where I had a one line job description: create a new $100M business in three years for Personal Mobile Devices. 

We developed an innovation methodology that worked really well in deciding in which new technologies to invest. Using that methodology, we selected four technologies to commercialize. One of the four was called SmartSense which was a personal wellness platform that included a bracelet, a mobile app, and cloud computing.

SmartSense bracelet


The SmartSense bracelet measured: 

1. Physical activity (steps, intensity, miles walked, calories burned) 

2. Stress 

3. Skin temperature and ambient temperature 

4. Heart-rate

5. Sleep pattern 


The magic lay in intelligently gathering the data from the body, analyzing it with history, understanding the correlation and causation among various data sets, and making sense of the user's behavior; then representing the output to the user in a way that changed the user's behavior. The "cloud" continuously learned from the user and became adapted to the user. It was a very complex undertaking. The cloud software, the smartphone app, and the sensors on the bracelet had to work in sync to break old habits and shape new habits that lead to a healthy lifestyle.
Interconnectedness 

We had a holistic idea of fitness which included both physical and mental fitness. Understanding mental fitness is a relatively new phenomenon in the western world. However, more and more people are getting into yoga and meditation. Furthermore, we noticed that fitness is not fun for most people and we were able to make fitness a fun and social activity. People like to have fun and people like social activities. 

On the smartphone app, the user created an avatar that effectively communicated to the user the impact of the behavior. For example, if a user did not exercise, the avatar would get fat and if the user got stressed, the avatar's face would get red. 

Additionally, for the average person, fitness is a solo activity. We learned that if we could create a social contract amongst users' friends to complete a certain task every day (like walk 8 miles) then users were more likely to complete that task. Our app brought that into being beautifully. Friends could set up a joint goal of, let's say, walking 8 miles a day. They could track each other's progress, encourage each other, etc. Most people spend their lives doing things based on what others think of them, so social pressure proved to be a good incentive.

Avatar on smartphone app

We created an end-to-end, user-centric working product to get real feedback from the market. We did that jointly with a mobile operator. The idea was that a mobile operator can distribute the product, create differentiation for itself, and create a new, monthly recurring revenue stream. The main hypothesis we tested in a field trial was: would consumers would wear a personal wellness device, pay for it upfront and and pay fees on monthly recurring basis just like they do for smartphones? The monthly recurring charge would be for the personal fitness coach service that SmartSense provided. The response was overwhelmingly positive. 

We started before there was a defined Quantified Self movement and we were in the market before there were any wearable devices. I am glad to see that new wellness bracelets like Nike+ Fuelband, Jawbone Up, and Fitbit Flex have become available in the past 18 months. However, we are far from realizing my vision. 


Why did we decide to invest in a personal wellness platform? The developed world, for the most part, has moved from the industrial age to the information age. Until we got to the information age, most people did physical labor and there were not many choices for unhealthy food. If you go further back in history, most people were poor with not enough to eat. So, it follows that obesity is a relatively new phenomenon; mainly the result of new lifestyle - sit eight hours at work, sit almost five hours a day at home and watch TV, lie down and sleep eight hours a day, sit and drive anytime you step outside the house. And on top of that, eat over-sized portions of meals; which are made with who knows what, every day. In addition, we spend our lives worrying about how we get more stuff and do more of the same. This worrying causes a lot of stress. Forget about focusing on the mind and mental health. There are ways to make your brain obese too. You don't become obese in a day. It is a slow process. The best analogy to describe how one may become obese is by my hero, Charlie Munger, "A frog tossed into very hot water would jump out, but that the same frog would end up dying if placed in room-temperature water that was later heated at very slow rate." Obesity happens the same way. A big issue is awareness i.e. people don't realize that they are not doing any physical activity. 

The healthcare system we have is designed for cure and treatment, not for prevention. Vaccinations provide prevention against some known dangerous diseases. However, the responsibility of staying fit and taking care of oneself has always lain with the individual. The individual is lazy by nature and likes to eat large amounts of fast food. You don't go to the doctor because you are not working out. You go to the doctor when that lack of physical activity causes a bigger problem like heart disease. And we can't really quantify what the healthcare costs are for a high-stress lifestyle. The current spend on healthcare in the US is more than 17% of GDP and it is expected to rise to ~33% of GDP by 2040. Isn't that unbelievable? Unfortunately, the solutions to the healthcare problem are again related to cure and not prevention. No one wants to die. If we make people aware that their lack of physical activity is getting them closer to death and provide social support around doing physical activity and reducing stress in their lives then people may change their habits. Needless to say, the opportunity for personal wellness devices is huge. 


There are a lot of challenges when you are creating something that has elements of identity, fashion, changing habits, hardware, software, design, big-data, wireless connectivity, battery life, etc. After making the decision to apply mobile technologies to healthcare, the driving force for the product development was an awesome User-Experience (UX) that changes user-behavior. I faced the following challenges in bringing the product to market: 

1. Wellness Device vs Medical Device: A simple distinction is that a device requires FDA approval (in the US) is a medical device and a device that does not require FDA approval to sell is a wellness or fitness device. It is a actually a lower risk investment proposition to make a medical device because they are usually single-purpose like a blood-pressure monitor, diabetes monitor, etc. and there is an already-built ecosystem of doctors and pharmacies. Furthermore, people who want to use medical devices have a known disease/problem for which they are actively looking for a solution. The downside: FDA approval can take years. Wellness devices, on the other hand, require people to change behavior which is incredibly difficult, particularly when most people don't acknowledge that they have a problem. 

We chose wellness devices because of the market potential and fast time to market. 

2. Form Factor: We could have made almost anything we wanted since a 3-axis accelerometer (sensor) can detect physical activity from anywhere on the body. And, there are trade-offs with other sensors. Heart-rate is easier to detect closer to the heart. Outside temperature is better measured with exposure to the environment. Actually, an accelerometer will give better step count closer to the foot. So, with any form factor you chose, you are making a trade-off. I leveraged the Stanford Design School in need-finding. 

We went with the bracelet because: 

i) The idea was to make SmartSense a social product so it had to be visible to others. The user had to like wearing it and sharing it with others. Innovation spreads faster if it is observable

ii) Most younger people don't wear a watch and are happy to wear a bracelet as a fashion accessory. Even if you do wear a watch, the other wrist is available for a bracelet. 

iii) The device had to be visible to the user as a constant reminder to be fit and as a visual communication of the achievement of daily fitness goals. 

3. Design: Now we know that it is a bracelet. What does it look like? Is it a fashion accessory or is it a sports accessory or is it a watch? Do we create separate ones for men and women? What is it made of? Is it easy to wear? How does it feel on the skin? Does it have a display? How big is it? How much does it weigh? How tightly does it have to be worn on the wrist?

Z axis is the height 

I learned a lot about fashion and human behavior during the design research. People are very particular about what they wear on their bodies. I met with Dior, D&G, and other fashion houses just to learn how they make these decisions. A big x-y axis size for a bracelet does not bother wearers but a long z-axis does. SmartSense was simple, visible, cool. At the time the idea was to make the bracelet like the LIVESTRONG wristband which was bought by 80M people before Lance Armstrong's image was tarnished by doping charges. The unit we took to the field trial was eventually going to be reduced to one-fourth of its size in mass production. 

As an aside, another big learning was how to work with design agencies. If you thought that creative people in ad agencies are sensitive about their work, try working with the designers! 

4. Wellness Bracelet or Smartphone Accessory: Is the bracelet solely dedicated to fitness or can I also use it as a second screen to my Smartphone? Can I get notification on the bracelet when I get an email/SMS? Can I read a text message on the bracelet? Can it tell me who is calling so that I don't have to take the phone out of my pocket? Why can't it be both a wellness bracelet and a Smartphone accessory? 

All decisions are trade-offs. I decided that it was complex enough to make a wellness device with multiple sensors. Adding the additional complexity of making SmartSense a second screen would make it a watch, and that may limit the market to technophiles. We wanted SmartSense to have global appeal. Due to that, all electronics had to be hidden. In merely looking at the bracelet, the casual observer should not be able tell what the bracelet is for. 

5. Architecture: This required a synthetic view of how a combination of cloud, smartphone, and sensors would work with UX being the primary focus. How is sensor data transferred from SmartSense, to the smartphone, to the cloud? How do you prolong the battery life of SmartSense and not drain the battery of the smartphone? How do you lay out the PCB so that the SNR is maximized? What is the memory size for SmartSense? Does SmartSense sync automatically with the smartphone or do you press a button? Who initiates the sync? What happens in the cloud? How is the data secured? Do you open APIs for developers so that they can develop new applications? Do you let people use Facebook as their social network for SmartSense? 

We made the cloud, the smartphone, and SmartSense work in harmony. Most of the data processing and storage was in the cloud. On SmartSense, we included an ARM Cortex M0 for some basic processing. SmartSense synched automatically with the smartphone every few minutes i.e. without ever touching a button, a user had feedback on his physical and mental activity at any given time. Of course, the APIs were going to be open. Very basic information like low battery, reset, full battery, sync in progress was communicated through LEDs on SmartSense. Everything else was on the smartphone app. People look at their Smartphones 100+ times a day anyway. We also had to think about the new habits we wanted the user to adopt related to device usage. Given the state of Bluetooth at the time, we wanted people to charge their SmartSense every night like they do their Smartphone. However, the battery life was more than enough for a weekend camping trip. 

6. Features: The goal of the device was to keep people fit by making them aware of their physical activity and stress levels and prompting action with commitment, visible to the social group. How does the device communicate with the Smartphone? Or should it directly communicate with the "cloud"? Does it have a display? What kind of display? How do you measure stress? What sensors are included? How long of a battery life is needed? Is the bracelet waterproof? 

The architecture and features are highly interdependent. The sensors we chose to be included in SmartSense were based on function, cost, size, and power consumption. We stayed away from the common mistake of jamming in as many features as possible. Our focus was UX. 

It did not really make sense to make SmartSense a "standalone" device i.e. it talks directly to the "cloud" given that most people have a Smartphone in their pocket. So, we chose wireless sync between the smartphone and SmartSense using Bluetooth. A large display on the device is an expensive proposition and it conflicts with the perception of SmartSense being a cool thing to wear. SmartSense had an LED display and ~72 hours of battery life (you could spend the weekend in the wild with SmartSense without charging it). The main drain on the battery was automatic syncing every few minutes. We could have made the battery-life exponentially longer by varying the time between automatic sync. By the way, at that time there was no BLE on any smartphone in the market. The stress level was measured using GSR. There is no scale for GSR, so we created a scale that showed deviation from "normal" levels. We could have made SmartSense waterproof but at that stage water-resistant was enough. 

7. App: Data visualization is a big challenge on a small screen. However, the user should have access to all of the user's mental and physical fitness history in the palm of the user's hand. How do you present the data in a way that creates awareness, understanding and prompts action? What is the sequence of screens? How many clicks before the user gets to what the user wants to see? How do you show correlation between different data points? How do you interact with friends? Do you see all your friends' data points? Is there such a thing as overall fitness level (physical + mental)? How do you create an inspiring UX?

We wanted to make fitness fun and social. On our app you picked an avatar that represented you and it would get fat if you didn't have enough physical activity. Its face would get red if you got stressed. This was a very effective way of projecting the results of current behavior into the future and prompting change. People are more likely to break habits if they make a commitment to themselves and to their social group. So gaming, competitive, and collaborative components were an integral part of the app. Since we had open APIs, the plan was to collaborate with the developer community for evolution of the device and the app. 

A big achievement was that we created a personal relationship between the user and the avatar. 

8. Calories: How do you measure the calories someone consumes? The wellness device can only measure the calories burned. Do you carve out a section in the app that allows people to input every calorie they consume? We are always burning calories. Even breathing burns calories. This "automatic" calorie burn is commonly referred as passive calorie burn. What calorie output do you display - the calories burned by physical activity (active) or a sum of active and passive or show both active and passive separately? 


Who is counting the calories?

A lot of the fitness enthusiasts and Quantified Self participants were already tracking their calories in some way. And, given the constraints, we did not develop anything that would help them import historical calorie consumption data to our app. And, our app did not have a calorie input ability. It would have taken the fun out of the app. We displayed total calories burned (active + passive) to keep things simple. 

9. Business Model: How do you create awareness about the device? What distribution channel do you use? How do you charge people for the device? Is it a one time sale like buying a TV? Or can you create a recurring revenue model? Where do you sell it? Health insurance companies? Health Clubs? Enterprises? 

When we partnered with a global wireless operator to develop the device, the idea was not only to reduce investment risk but also to have a distribution model for the device. The wireless operators have been going through a tough time for the last few years. They are looking for new revenue opportunities and sustained differentiation. On the positive side, they have the marketing muscle and retail distribution network. SmartSense was to be distributed just like smartphones are distributed through the wireless operators - with a subsidy and with a monthly recurring fee for the service (albeit the subsidies and the monthly recurring fees are much smaller than those for smartphones). This not only creates a new revenue stream for the wireless operator but also reduces churn (increases the switching cost for the people to leave the operator). 

10. Coaching: Should there be a coaching element in a user's interaction with the SmartSense app? Isn't making people aware of their fitness level enough? And, then there is the pressure from the social group to take action to keep fit. Would people still want to be told what to do? Will people pay for such a service?

A personal Coach helps...

This was a key learning from the trial. People want a personal coach who tells them what to do every day in terms of keeping fit mentally and physically. And, they are willing to pay a monthly recurring fee for this coaching. The personal coach idea is very similar to a personal trainer, which people pay for at health clubs. 

I have captured the main decisions and challenges we had to face during the development process. Of course, there were many more challenges. And, for the trial, the heart rate and sleep monitoring features were disabled. The open APIs were not ready. The biggest learning for me was how to align people with very diverse backgrounds - firmware development, app development, algorithm development, cloud software, industrial design, manufacturing, system architecture, semiconductor design, market research, product development, marketing, legal, business development, etc. - in two very different organizations, to work together toward a shared vision which was not to be compromised with aggressive deadlines and budget constraints. There were scores of people, in five countries and eight locations, who worked together to bring SmartSense to market. This exceptional team* and was the most critical success factor. 


Where do go from here?

The hardware costs will continue to drop with scale. The accuracy for readings will continue to increase with more usage. Data visualization will continue to get more effective. The number of sensors on the device will continue to increase. Not all the current players will survive. If Apple enters the game, it may exponentially increase the adoption of wellness/wearable devices. We are still at the early stages of this new market category. Following are the things I would like to see happen: 

1. Community and New Habits: To date no one has figured out how to leverage a user's social group to prompt action towards changing behavior. The interaction amongst friends on most apps for fitness devices is close to pathetic. The current focus from major players like Nike seems to be selling more and more devices without much attention to creating a community that prompts action. I am afraid that without the element of community, commitment, and fun, people will lose interest in the devices. Wearing a fitness device has to become a new habit just like wearing clothes is. 

2. Integration with the Healthcare System: When I go and see a doctor, she should be able to see years of my historical fitness data and make a more informed decision about diagnosing the problem I may have. Furthermore, she should be able to make suggestions for changing my fitness routine based on the diagnostics. This integration will be most beneficial to society as a whole. However, alignment of the personal, medical, and insurance spheres is extremely complex and may take a long time. 

3. Adaptive Algorithms and Context-Awareness: Can you adapt without being context-aware? The algorithms should continue to learn from my behavior and adapt to me. For example, if I walk every morning with coffee in my hand, the steps should not be shown as zero because my hand was not moving (sadly, this is how many devices record activity today. The devices don't measure spinning either). The device has to adapt to me and not the other way around. Given the history, along with the specific motion of my hand while I am drinking coffee, the movement should be detected as walking. In the early stages, the app can even ask me for feedback, that is, was I walking between 7AM and 8AM? The algorithms should also correlate outside data to provide coaching. For example, if I am in my car every morning between 8AM and 9AM and the stress reading goes up then the algorithms should suggest that I listen to calming music. The algorithms have to become context-aware and the learning should be continuous. After a while the device should be an integral part of me. It should understand me better than anyone else. 

4. Standards: Let's say I spend two years with the Nike+ Fuelband and then Apple comes out with a much better product. Should I let go of my two year history and start afresh? There should be standards around how the data is captured and stored and how it is migrated from one system to another. Without standards, one system cannot read the data from another system. I am not sure if this will happen because companies tend to do things to lock you in their platform/ecosystem. There are companies like Myfitnesspal which let you get data from multiple devices into one app but I am not sure of their standards and migration policies. 


I wish the smartphone told me the calories too...

5. Calorie Intake: The way optical recognition and cloud computing are going, it maybe be possible that one day our phone camera can tell the number of calories in the food we are about to eat and what the ingredients are. This will enable a complete solution to the fitness problem. The current devices focus on calorie burn and capturing intake is a very tedious process. 

6. Low-income people: Ironically, fitness is a bigger issue for people with restricted resources who cannot afford healthy food or wearable fitness devices. They may not be educated about the importance of good mental and physical health.  With Obamacare, they maybe able to get healthcare insurance but how are they going to buy these new cool gadgets? Does it make sense for healthcare insurance companies to provide them with these devices? Do you create a low-cost version of the device?

Note: I have used fitness, wellness, and wearable devices interchangeably in the article. Products like Google Glass are often referred to as wearable devices as well. It is true that both Google Glass and wellness bracelets are worn on the body but lumping Google Glass and wellness bracelets together is not a good idea. They serve two completely different purposes and the design, architecture, electronics are very different. Calling these devices wearable computers is a misnomer because most of the computing occurs in the cloud or on the smartphone. 

I am delighted that I had the opportunity to work with a great team to bring mobility to preventative healthcare. We created something new. It was a thrilling experience! Every time I see someone wearing a wellness bracelet, I feel proud. 


*A few extraordinary members of the core-team were: Satya Mallya (who led the project at the global operator we partnered with), Rajiv Jain, Brani Dubocanin, Shmuel Ungerfeld, Jian Zhang, Bengt EdlundSteve Matson, John Benko, VĂ©ronique Henry, and Jim Schuessler. I am eternally grateful for their contributions. And, nothing would have been possible without the support of my bosses - Mike Polacek and Don Macleod.