I write columns on news related to bots, specially in the categories of Artificial Intelligence, bot startup, bot funding.I am also interested in recent developments in the fields of data science, machine learning and natural language processing ...

Full Bio 
Follow on

I write columns on news related to bots, specially in the categories of Artificial Intelligence, bot startup, bot funding.I am also interested in recent developments in the fields of data science, machine learning and natural language processing

This asset class turned Rs 1 lakh into Rs 625 crore in 7 years; make a wild guess!
910 days ago

Artificial intelligence is not our friend: Hillary Clinton is worried about the future of technology
914 days ago

More than 1 lakh scholarship on offer by Google, Know how to apply
915 days ago

Humans have some learning to do in an A.I. led world
915 days ago

Human Pilot Beats Artificial Intelligence In NASA's Drone Race
916 days ago

Google AI can create better machine-learning code than the researchers who made it

More than 1 lakh scholarship on offer by Google, Know how to apply

13-year-old Indian AI developer vows to train 100,000 coders

Pornhub is using machine learning to automatically tag its 5 million videos

Rise of the sex robots: Life-like doll goes on sale for 15,000 pound

How Clue is using data science to revolutionise female health

By Rajendra |Email | Oct 25, 2017 | 8052 Views

Data science is a vast, fascinating field. Analysis of datasets can be applied to multitudes of areas to improve people's quality of life. It would be difficult to find a better application of data science changing people's lives than Berlin-based menstrual and ovulation tracker Clue.

Launched in 2013 by CEO Ida Tin, the app's empowering tracking features and absence of menstrual euphemisms (think butterflies and flowers) has made it a huge hit with users. Clue gives users the option to track not just their menstrual cycles, but a whole host of other patterns, from pain experienced to food cravings. It can even predict when users will feel PMS symptoms.

The app has grown into a multimillion-user phenomenon across 190 countries, and it aims to be the No 1 female health app in the world. It's been steadily gaining plaudits and was named the top free menstrual-tracking app in a study published in the journal Obstetrics and Gynaecology in May 2016.

From astrophysics to data science

Data science and analysis of usage data is crucial to the success of Clue, something Marija Vlajic Wheeler knows all about. A former astrophysicist, Vlajic Wheeler has been a senior data scientist at the Berlin-based app's headquarters since 2015.

Siliconrepublic.com spoke to Vlajic Wheeler about the role of data science in driving Clue forward and making a difference to users lives.

She explained that she left academia just as the data science wave began to crest, and was also working in a major start-up hub in Berlin at the time, so it was a case of "very fortunate timing" that saw her enter the field.

She said that a lot of her existing processes have carried over to her work at Clue.

"Whether it's galaxies or people's menstrual cycles, the way you think about the data - when you ask, 'What can we do with this dataset?' or 'What are the limitations of the data?' - a lot of that stuff actually remains back from when I did astrophysics. The domain knowledge changed but a lot of the processes around it actually remained similar."

Collaboration at Clue

Clue has been collaborating on projects with major universities such as Oxford, Columbia and Stanford since 2015 to advance female health research.

It is sharing anonymised and aggregated data with the institutions to examine the evolution of menstrual cycle patterns since the 1960s; the link between menstrual patterns and the onset of disease; and connections between IUD use and bleeding patterns. As well as this, the team also has its own in-house research group.

As a senior data scientist, Vlajic Wheeler spends some of her time working on the data side of these various research collaborations. In terms of specifics, "that can be anything from menstrual health to machine learning".

She also said that in recent times, the data team has been working more closely with colleagues at product development. Between the product and pure data work, it can be a different set of tasks each day.

"These things aren't set in stone and are often sort of in flux. In general, there is a lot of overlap and collaboration, and that makes it interesting as you get to work on multiple aspects of the app."

As it is a health-tracking app, Vlajic Wheeler does undertake some research around new medical findings, but ensuring the clinical accuracy of the information on the app falls within the role of the research specialist.

"I do some [research] - not as much as I would like to do, as it is a fascinating field. Once you get deep into it, you realise the more you know, the more there is to know."

How Clue analyses usage data

Clue truly is a collaboration between the user and the development team. Without usage data, the team cannot make advances to improve the app and female health at large, and it is the beating heart of the operation.

Vlajic Wheeler explained that in terms of analysis, the team does do some A/B testing for certain app features, but it also tries to dig deeper into the data.

"[With A/B testing], we can see how certain features perform over certain metrics like retention or number of sessions, but we also like to see if people use the feature in the way we imagined."

She gave the example of a feature called 'Insights', which gives each person user-specific health updates. Usually, the team avoids pop-ups, so Insights didn't originally appear in a pop-up on the app. However, in its analysis, the data science team found that not enough users were engaging with the feature.

"Informed by this info, we changed the delivery format into a pop-up so people would see it."

Clue data scientists use a combination of information from account-holders that share their data with Clue, with other data provided by the company's analytics partner, to "get deep down and see if people use a feature in the way that we planned for it out in the wild".

In terms of tools, there are a multitude of different features for different purposes. "When we do ad hoc and preliminary data analysis, we use Python and Jupyter Notebook."

Analytics-wise, the team works with Amplitude to spot user behaviour patterns that could be used to improve the app. For secure access to different data pools in one place, Clue uses Amazon Web Services. "That brings data from all different sources into one place that we can analyse."

Keeping user data safe

Obviously, the team at Clue deals with sensitive information about the health of its users, and Vlajic Wheeler is passionate about continuous improvement on best practices. "We are always looking to see how we can be more careful with users' data", she noted. "When the data team works in the warehouse with all of the information from different sources, we have no identifying information about the user - that lives only in one place."

She also explained why the collaborative philosophy at Clue is so important when it comes to user health. "We have a research person who is basically on the product team, and they make sure everything that we do is in line with the most recent science and the stuff that we do; that it's actually scientifically backed."

Data with a mission

When asked about what she enjoys most about working at Clue, Vlajic Wheeler simply said two words: the mission.

"I was a huge fan before I joined, but just realising what you do really affects people's lives.

"The type of feedback we get is just really what makes me excited to come to work every day."

That sense of purpose seems to be what keeps the team excited to consistently improve the app.

User feedback can be especially powerful, according to Vlajic Wheeler. Not only does Clue track the menstrual cycle of the user, it can also be instrumental in helping people receive early diagnoses of potentially dangerous conditions. "The most powerful stuff is when people say that tracking with Clue has helped them discover, say, an ectopic pregnancy, and saved their life."

"We had a pilot project where we notified people who we thought were at a higher risk of developing a certain reproductive health condition, so we sent them an email telling them they should potentially see a doctor. Not everybody who we sent the email to did develop a serious condition - some did not - but there were people who were able to treat cancer early. People write stuff in through support and it then gets shared within the company, and it often makes me cry to read these things."

Vlajic Wheeler stated what Clue's data mission is at present: "I think a lot of stuff we are planning now has more of that short-term goal of providing peace of mind, in the sense [that] you know what your body is going through. First you start noticing patterns, and, even if people don't notice them off their own bat, we should have enough data to point out their patterns for them."

That empowerment Clue provides can be the difference in blaming a bad day at work on your incompetence as a person (which is unlikely) or on the fact that your upcoming PMS might be having an effect on your mood, or even the reason you go to the doctor when you notice an unfamiliar pattern forming.

In terms of the future of Clue, Vlajic Wheeler hopes to one day be able to have users tracking enough data that the app can reliably warn them if there's something they need to investigate around their health.

"There is this 'Aha! moment' when you track these things that you've been feeling your whole life. You're feeling shitty and you're suddenly like, 'Oh, it's my hormones'. It's not that you're a failure as a person or anything."

This self-knowledge ethos is powering the future of Clue, and allowing users to feel connected and informed about the idiosyncrasies in their bodies, while advancing medical research milestones in innovative ways.

Source: Silicon Republic