The team at Clue on how to integrate scientific research into product development cycles.
“We’re currently doing research with Oxford University showing how the female menstruation cycle is actually a health cycle, where the body responds very differently to inflections throughout the cycle. One working hypothesis is that PMS is related to infection in the body. We’re specifically looking at how undiagnosed sexually transmitted infections might express themselves as PMS. The idea that severe period cramps could be a sign that she has a high infection in the body, would be revolutionary, given how much people suffer.”
Talking to Ida Tin, CEO of Clue, you wouldn’t know that she has no medical background and used to run motorcycle tours in Cuba, Mongolia, the US, Chile and Vietnam. Frustrated with the side effects of hormonal birth control, she launched an app that tracks women’s menstrual cycles, with the ambitious goal of improving reproductive health. The app is now used by over five million users globally and was rated number 1 by the American College of Obstetricians and Gynaecologists last year. Scientific reliability seems to go hand in hand with user acquisition. How did she do it?
There has been a lot of interesting research on the relative importance of R&D within large corporations. Equally, there has been a lot of information on how start-ups need to “fail fast,” treating their entire business as a sort of pseudo-science lab to test business models and try again. But there has been much less on how to incorporate scientific research into the product development.Yet the effectiveness of new “digital health” startups will rely heavily on whether this can be done well. Facebook’s famous motto, “Move fast and break things” is potentially unhelpful for a company that aims to help women understand their bodies. On the other hand, they are faced with a young, impatient customer base who want to move fast.
When startups speak of how they have like “innovation labs” and that they “test everything”, you’d be forgiven for thinking they are describing a scientific process. But science’s goal is to explore everything and anything, at all costs, a business’ goal is to acquire customers. They have fundamentally different goals. Yet they can, and in the case of a health product, must work together. This presents both challenge and opportunity.
How, then, do you build the team that can create a commercial product which relies on not just current scientific knowledge, but its developments? How do you tread the line between academic accuracy and product usability?
Hiring a gynaecologist: structuring the team for science
The first thing Tin is focused so much on is building the team and the structures to ensure that the product leverages expertise. They began first using a network of doctors, gynaecologists and academics, who could read their texts and make sure they are scientifically accurate. However, this was time intensive; they needed to do this across 15 languages, for a global user base. Just a few weeks ago, they hired their own gynaecologist and have been bringing that expertise in-house to speed up the process, to check, inform and validate their product decisions.
Tin explained: “I think if we can connect our user base with our ability to understand the questions they have into our product, with the universities who know how to crunch the data and the clinicians who have the expertise and deep knowledge, then we have something which is incredibly powerful. We’re still early in that process, but that’s what we’re trying to instigate.”
Part of this team is Anna Druet, one of their in-house scientists, who works directly with the product team and, along with her colleague, Marija Vlajic Wheeler, with the academic communities. Druet talked me through the process in a bit more detail.
She explains that there are three branches of research at Clue:
- Desk based research which enables them to stay on top of all the latest research coming out in the medical and scientific community. We don’t know so much about the female reproductive system so it’s more critical than ever to stay on top of the latest findings.
- Collaborations with academic partners. They have ongoing relationships with the likes of Oxford, Stanford and Columbia. This enables them to leverage the expertise and robust methodology used by the academics with a data set which is both one of the most accurate and the largest in this field from Clue.
- Internal, smaller scale studies. These are not intended for publication in academic articles but are large scale analyses of big data. For example, one of their data scientists (who also happens to have a PhD in Astrophysics) did a study on whether menstrual cycles correlate with the moon. They got thousands of emails from their users about this. She analysed around 7 million cycles and found no correlation!
Collecting the data: seeing the menstrual cycle as a vital sign
Data is key to this research. Not only do they collect millions of menstrual cycles each year, they have also done surveys with their very active user base. Last year, they conducted a surveywhich 90,000 people responded to from over 200 countries on the social taboo of saying the word “period”, collecting the diverse range of euphemisms that are used across the globe. Tin laughed as she told me of the incredible linguistic gymnastics that people go through in order to not have to say the word period. “Amazing, when you think that 50% of the world will have one!”
They have also reached out to their user base for a survey on disease detection, in which users’ cycles were analysed to see if certain irregularities matched diagnostic patterning.
“PCOS (Polycystic Ovary Syndrome) is very common. It’s amazing the ratio of how common it is to how little we know about it. We tried to test a hypothesis that having data on someone’s cycle coupled with information on certain symptoms they experienced would enable us to say something on potential medical condition that went undiagnosed. Early results were promising. Before the app, collecting data on this scale would be almost impossible. When we speak to academics or medical researchers, they generally feel like it’s Christmas when we show them how much data we have.”
Users didn’t have access to their own data previously in quite so much detail, and neither, for that matter, did the medical profession. As Anna Druet explained,
“Even if you go to the doctor, it’s not like checking your elbow for a symptom; they need to have your whole record. Your menstrual cycle is like an extra vital sign, a pulse or your blood pressure, if you use it. Knowing the normal ranges of what your menstrual cycle should be falling within, is so important and can help spot anything from hormonal disorders, to whether you’re exercising too much and not getting enough calories. We can spot these things pretty quickly, especially in adolescence.”
Her digital body: user centred research
Every woman receives personalised data and analysis on their cycle
Tin was keen to emphasise that the research cycle at Clue begins with the user. This is key even in how they structure the team itself, “We begin by finding the questions the users are most interested in answering. This feeds through to the researchers, product team and data scientists, who all work together to ensure that when we find something of value, we can get it to our users as soon as possible. That’s not just so we can get the data from the customers to validate our findings further, but so they can get the benefit of such insights faster.”
This user centric approach also applies to how they think about data privacy, an issue which the medical and healthcare communities have recently been wrestling with. Efficiently collecting data is one thing, maintaining trust is another. This is something Tin feels very passionately about:
“The starting point is to have a really clear contract with the user: this is what you give and this is what you get, this is what we will and won’t do with the data. Most companies try to hide that, and they try to hide it because what they do is questionable. It’s also important to educate people around it. Do people truly understand data? That’s a massive topic, but we want to do our part to make sure people understand. We want to make money on selling what people understand and not by just selling their data in the dark.”
Non linear collaboration: research, product and the customer
For a medical product, where research is continually updated all the time, they needed to develop a process which was far more fluid and collaborative; a circular rather than a linear process. Druet sits between the product team and the research team, often writing the notes that sit on the app itself.
“The back and forth between product and scientists is really important. We normally start by writing a summary of scientific literature for a product we are building. The product team then pair things down, to make it more simple. Normally, they then start building tools, and create the code. At this stage, they will start asking whether it would work to put this piece of information here but the other over there. And sometimes the answer is no. These two pieces of scientific information must sit together to be complete, you can’t always break down an idea and have it keep its validity. It’s this constant back and forth between conceptualising, building and optimising for the product team and then coming back to us and asking us if it’s OK, that makes it work.”
Working with academics is not always easy, but Druet was also keen to point out that they are nevertheless indispensable, “Their pace of work is very different to a tech startup. They have much more red tape to go through: whether that’s getting something approved, winning funding or just the general bureaucracy of academia. They are also fundamentally trying to write papers. We’re building an app. But equally, the relationships are invaluable. One of our collaborators is Paula Hillard, a well-known researcher who sits on our medical board, she is always on hand to advise. Endocrinology [the study of hormones] can be a complex field to navigate and we need information that has been verified by those who understand its subtleties. There’s still a lot we don’t know in this area, and that’s important to appreciate too.”
The app has to be simple for users to input data, without avoiding the nuances and differences of every woman’s experience
Balancing the need to be scientific and the needs of the user when using the app is a tough challenge. “From an academic perspective, it’s often better to have more information, more citations, but that doesn’t always flow well with the context of an app with the space and flow functions it has. In the end, we need a accurate, rigorous product people will use as we then get more data back. This in turns enables us to do much more to move women’s health research forwards.”
And what about when there’s grey areas, when the science itself is not conclusive? Or even what happens when someone is an outlier, who doesn’t fit normal patterns? How do you design an app which appreciates them?
“There are always caveats. It’s about striking a balance between what a norm is and what typically happens, and then empowering people to look at their own patterns. You might have the same hormonal surge as others at the same time, but experience it in a different way. There are trends, but just saying things in a black and white way is unhelpful and we don’t want people to think “oh I’m not normal, because I don’t have this exact pattern or symptom”. In some cases, it’s a time to go and chat to a doctor, but in so many cases, it’s just difference. That’s why we provide references and tend to give people more resources. We also use language like “may” or “likely to” or “tend to”, or have sentences like “people experience things differently” just to empower people to listen to their own bodies. We also want people to look for their own patterns.
The team at Clue is, ultimately, on a mission which they hope will change both the way research in female health is done, and the way their users relate to their own bodies.
“It’s incredible how little people are taught about their bodies. Globally sex education is done very poorly, most women don’t know the very basics. Things like “what’s a relatively normal level of pain I should be experiencing?” or “at which point of my cycle am I more likely to get pregnant?” or “is this a healthy bleeding pattern?”. These are common questions we help our users find answers to. When it comes to the menstrual cycle, reproductive health, and health in general, knowledge is power.”
At Draper Esprit we are investors in Clue, Push Doctor, Lifesum, Fluidic Analytics, Horizon Discovery and Graze. If you’re building something exciting in the digital health and wellness space, get in touch via firstname.lastname@example.org