THIS STORY IS AWAITING CUSTOMER APPROVAL. CONSIDER IT COMPANY CONFIDENTIAL. DO NOT DISCLOSE.
“Data for good” gets personal
A behavioral science-based app uses sophisticated survey analytics to help users lead more emotionally fulfilling lives
_____________________________________________________________________________________
Challenge: In building Purposeful – an app that prompts users to be more intentional about their emotional wellbeing – Kumanu founder Vic Strecher, PhD, set out to address the root causes of depression, burnout and anxiety in today’s workforce. Strecher and his team mined vast quantities of behavioral survey data for insight into how emotional and mental wellbeing could be improved through a set of targeted interventions that go beyond traditional incentive structure.
Solution: To proactively avert data quality challenges inherent in large surveys, Strecher processes behavioral science data in JMP® statistical discovery software. “One of the things I love about working with JMP is that it resolves the tedious process of cleaning data,” he says. “The kind of checking I can do using JMP is just so much easier than the next best alternative.”
Results: As the Kumanu team continues to refine the algorithms powering Purposeful, Strecher says JMP has proven to be a quick and approachable tool for exploring new ways of extracting insight from both structured and unstructured data. In particular, the JMP platform Text Explorer is his tool of choice for digging into text data. “I’m not a statistician, but I do have a lot of knowledge around behavioral science,” he says. “JMP has freed me from dependence on statisticians because with JMP, I can dig into the data and start analyzing things myself.”
_____________________________________________________________________________________
The profusion of wearable biometric devices and personal data apps in recent years has helped individuals track everything from diet and exercise to mental health. Focusing on data collection alone, however, falls short of empowering users to analyze their personal data, understand its impact, and –critically – change behavior.
At the professional level, human resources (HR) departments have similarly struggled to turn classical approaches to job satisfaction into measurable improvements in employee wellbeing. Traditional HR programs often rely solely on incentives; an approach that is not easily scaled and often falls short of delivering meaningfully on dignity, equity and inclusion (DEI).
With the COVID-19 pandemic and fluctuations in the economy accelerating workplace pressures in tandem with a rapid move toward remote work, behavioral scientist Vic Strecher says organizations’ attention to the wellbeing of their workforce has never been so vital. Strecher’s solution, Purposeful – an app that operationalizes learnings from the field of behavioral science – turns to technology for a more actionable, holistic way of thinking about professional fulfillment.
“Employers should want their employees to be energized and engaged at work, while also wanting to help people get home and be energized in their personal lives as well,” he says. “Employers should want their employees to be aligned with a life purpose.”
Kumanu, the company founded by Strecher and maker of the Purposeful app, is a technology-based, purpose-driven wellbeing company that “helps individuals bring their best self to work – and to life – every single day,” Strecher explains. The company’s approach pairs machine learning, real-time predictive decision support and what he terms the “intrinsic motivation engine” underlying human nature.
The result is a wellbeing reboot with professional applications that go beyond the workplace. “New technology is not going to be the only answer [to problems like burnout, anxiety and depression], but it may help bridge a gap,” Strecher says. “And that’s what we’re doing with Purposeful.”
Lifestyle interventions based on behavior change and habit-forming models
Purposeful prompts users to spend a few minutes every day on self-reflection by journaling or responding to survey questions. The app’s survey instrument, Strecher explains, is structured as an ecological momentary assessment (EMA) with questions asked periodically throughout the day. In aggregate, this data helps generate insights into patterns in the user’s lifestyle and emotions.
The app then draws from a library of over a thousand different suggestions for overcoming or avoiding potential wellbeing detractors. The goal of these interventions is for users to experiment and ultimately develop healthy habits, Strecher explains; “whether it’s flossing or meditating or pausing each morning to think of something they are grateful for.” Sophisticated statistical models running in the background determine which tips to surface to users and when, based on peer-reviewed human behavior research.
“The app will start learning more about that person and establishing a baseline,” Strecher explains. “We want to help them better moderate their emotions by finding purpose in their life – or reminding them of their purpose. We can highlight the big picture as well as the specific things they can do to become less sad or less anxious.” In other words, he adds: “We’re helping people to become better at changing their own weather.”
With blue-sky technology, an investment in figuring out which data to collect
The design of Kumanu’s EMA has been an evolving process. As with any blue-sky technology, Strecher says, deciding which data to collect is a significant part of the challenge, and since its inception, Kumanu has evolved toward more identity-based assessment questions.
Strecher explains: “What are the roles that you have in your life? Are you a partner? Are you a mom or dad? A grandparent? A caregiver? Do you have a dog or cat? What kind of job do you have? Is it busy? Is it boring?” The answers to these questions help users craft a nuanced picture of the roles they have in their lives.
Next is to understand users’ values and how their goals are shaped by those values: Are they a person who is trying to achieve more in their job? Are they trying to be a better caregiver? “Very often, what we value most isn’t a thing at all – it’s people or causes that we care about,” Strecher explains. The EMA also seeks to understand what users are like when at their best: Are they curious? Are they generous to others? Finally, users are asked about their emotions. “How are you feeling? We like to ask that every single day,” Strecher adds.
When users score their emotions against a scale, responses lead to a map of survey questions that dive deeper into those feelings. And, he says, there is the ability to journal throughout the day that, with the use of natural language generation, provides further unstructured insights into a user’s psyche.
“We’re moving from a population model – for example, sleep is good for you – to an individual model – where we can say this much sleep is good for you, but too much sleep may not be good for you. And sleeping at the right time,” Strecher explains. “Purposeful is a very sophisticated tool that’s increasingly getting more sophisticated through data analytics.”
Addressing data quality challenges proactively with JMP
Kumanu’s team of data scientists works to continually refine and improve the algorithms that power Purposeful through a combination of survey analysis and natural language processing. Data sets are pulled from large surveys like the Harris Poll or in collaboration with researchers looking at major behavioral science data sets, Strecher explains. But, as with any survey data, quality can be a problem.
“I’ve learned over the years that if a data set has problems and is not cleaned up, it’s worse than worthless. It can produce incorrect answers,” Strecher says. “Making sure that the data are clean is so important. And being able to see the data visually in JMP really helps.”
JMP, a statistical discovery software and key platform in Strecher’s toolkit, provides unparalleled data preparation and visualization capabilities, he says. “One of the things I love about working with JMP is that it resolves the tedious process of cleaning data. The kind of checking I can do using JMP is just so much easier than the next best alternative.”
Moreover, he explains, JMP enables non statistical experts to work independently on their own data. “I have a lot of ideas and knowledge around behavioral science, but I’m not a statistician,” Strecher says. “JMP has freed me from dependence on statisticians to do the data analysis for me. With JMP, I can dig into the data and start analyzing things myself.”
Text Explorer and natural language processing are essential in further refining Purposeful’s algorithm
One of the key areas of unfettered data exploration at Kumanu is in unlocking insight from unstructured text data. Three Harris-Kumanu polls go out each year surveying roughly two thousand respondents in nationally representative samples. Respondents are asked to write out what their purpose is and then explain how their work helps to fulfill their life purpose.
“We can start using natural language processing to look at how work is fulfilling a purpose,” Strecher explains. “Some people work for money, and that’s something that very clearly comes out through text analysis. Other people work for personal growth, and that also comes out clearly through words like ‘achieve’ or certain clusters about colleagues and teamwork.”
“If advanced natural language processing is used in the right way, we can start using it more therapeutically,” Strecher says. “I like to play around with the concept of natural language processing and text analysis. I think, how would a computer start doing this? What would the problems be?”
“I always start with JMP… And I use JMP because I need to understand the kinds of patterns that emerge from language so that we [Kumanu] know what to develop next. Text Explorer is really fun for me. With personal computers and JMP, you're doing [big data analytics] at no extra cost. I can run a hundred analyses in 30 seconds and very quickly zip through them visually to see what's happening.”
Moreover, he says, “I keep refining that text analysis using JMP. I’m training the algorithm to increasingly reflect a [personal human reading of the comments]. When I see a discrepancy, I then look at my analysis and categorization. Maybe the computer picked up on something I didn’t notice. You can keep training the computer to get better and better. And once you’re done you can start feeding information back into the algorithm to help people understand why they’re working and to become more purposeful in their work.”
Applications of Purposeful go well beyond the professional world. The app has been involved in clinical trials in the healthcare setting, including to support people undergoing treatment for cancer. And in the future, there will be many more public health applications. Fundamentally, Strecher adds, Kumanu’s project is an existential one: “What if we flipped the traditional script and instead think about life? What it is about life that makes it worth living?”
