After the dazzling innovation on display at our inaugural Data Science Showcase—AI-powered tools with the accumulated medical knowledge of humankind, for example—you might think we'd have a hard time topping it just one year later. But our recent Data Science and Intelligent Automation Showcases did exactly that. Hosted by Najat Khan and Pallaw Sharma (co-chairs of our Data Science Council) together with Ajay Anand and Steven Sorensen (co-chairs of our Intelligent Automation Council), the three-day all-virtual event proved to be no less dynamic, vibrant or awe-inspiring than its in-person predecessor.
This time, the spotlight was on two areas: data science and intelligent automation—and on how our global community of data science practitioners are applying them across sectors to drive real-world impact right now. Check out four standouts from the event to get a sense of what’s possible when you apply your unique skills, passion and energy to our collective purpose today.
AI for Happier, Healthier Babies
Our goal is nothing short of changing the trajectory of health for all of humanity, and we believe more targeted and effective treatments and interventions early in life will be a key part of how we accomplish it. The Healthy Baby Initiative (HBI), a venture of the World Without Disease Accelerator (WWDA), is a case in point.
“By identifying infants at risk of childhood disease and intervening to confer their healthy development, HBI aspires to prevent childhood-onset allergic and autoimmune disease,” explained Dick Insel, who leads the HBI.
Right away, however, Gabe Al-Ghalith and others on Dick’s team realized that making an impact of that magnitude would require developing altogether novel data science tools.
Gabe framed the challenge, and the opportunity, this way: “During early childhood, there's a unique window to intervene and intercept all kinds of non-communicable diseases at the root. The developing infant gut microbiome"—a term that refers to all of the microorganisms in the given part of the body—"actually trains the developing immune system to be able to recognize 'friend' from 'foe,' including friendly microbes from pathogens."
In the eyes of Gabriel and his colleagues, enhancing our understanding of that microbiome just might open the door to transformative new treatment options.
But one problem stood in their way: namely, the lack of a comprehensive, publicly available source of data on infant microbiomes that included essential data around genetics, health and disease. Without that, it would be hard to identify the key features of healthy and unhealthy infant microbiomes—and develop the right interceptions from there.
This was "a clear gap," according to Gabriel.
And it was equally clear to Gabriel and his team that they would have to change that.
Enter "The Baby Cloud": the largest database of its kind in the world by orders of magnitude, according to Gabriel, which has tremendous potential for real-world impact. Think: Novel ways of identifying infants at risk of developing certain diseases. More precise approaches to identifying and testing targets for health-promoting interventions. And prediction frameworks that may have broader applications, as well.
We’re hopeful that it will help us finally move the needle on some of the most pressing healthcare challenges of our time.
AI and ML to Eliminate Diagnostic Gaps for Rare, Fatal Diseases
Pulmonary hypertension (PH) is a type of high blood pressure affecting specific arteries in your lungs and the right side of your heart. It's a progressive disease, one that's still poorly understood. The symptoms can be nonspecific, and non-invasive diagnostic tools aren't currently available; patients are often misdiagnosed for years.
Yet, unchecked and untreated, PH can be fatal.
Put all of those factors together and you can see the need to shorten the timeline between when patients first experience symptoms and when the disease is diagnosed. Might it be possible for us to reduce that timeline from years to months, weeks or even just days?
That was the question confronting an interdisciplinary team that featured data scientists and clinicians from Janssen R&D together with external partners nference, the Mayo Clinic and the University of California, San Francisco.
Najat Khan, Chief Data Science Officer, Janssen R&D, explained how we approached the problem: “We looked at various data types—electronic health records, claims data, electrocardiograph (ECG) data—in order to find a better way of diagnosing PH earlier, and as a team, we ultimately landed on the ECG data. The structured and ubiquitous nature of this data improves the potential for early detection in primary care settings and holds the potential to reduce the diagnostic gap as much as possible.”
Eventually, after mining more than 15 million ECG data points, the team developed a novel machine learning algorithm to analyze electrocardiograph data based on real-world health records.
And the results of that work have been very promising so far, potentially allowing for the early diagnosis of PH patients many months in advance. As Debbie Quinn, a Senior Clinical Lead specializing in PH, put it, "The algorithm is working better in the early stages than we ever imagined!”
We wanted to fully leverage recent advances in machine learning and explore novel techniques to capture the potentially complex trends across the subtypes of PH and allow for early detection.
Share
The team’s mission was a particularly urgent one, too, given the worrying increases in the incidence of PH that have been documented in recent years. Going forward, hopefully, innovations like these will help point the way to better outcomes for everyone affected by PH.
For now, the next step will be to deploy the team’s solution in a larger number of health systems to benefit PH patients. Down the line, what’s more, they intent to expand their approach to take on other rare diseases and better address the unmet needs of patients.
Safer, Lower-Cost Joint-Replacement Surgeries
More than one million joint-replacement surgeries are performed in the U.S. each year—it's a procedure that has been described in peer-reviewed academic literature as "among the most cost-effective and successful interventions in medicine," and Johnson & Johnson has long been a leader in the field.
Of course, we're always looking for ways to deliver more value and promote healthier outcomes for our patients. And that led one of our teams to ask: Is it possible to predict surgical implant needs in advance of surgeries—and make those surgeries safer and more cost effective in turn?
The answer appears to be yes, thanks to advanced data science algorithms together with state-of-the-art analytics capabilities.
By leveraging new touchpoints and data for insights, we should be able to understand the exact needs of each individual patient prior to surgery, and to make smarter shipment and inventory decisions based on that information. Rather than, say, shipping all possible sizes in advance of a given joint-replacement surgery, we can ship only what aligns with the requirements of each case. And that means less waste as well as lower costs—both for us and for the patients we serve.
Plus, the algorithm can account for the specific sizing preferences of individual surgeons, meaning it could help make these procedures safer and more effective in a host of other ways down the line, too.
Machine Learning for Enhanced IoT Cybersecurity
Cybersecurity attacks have been on the rise in a major way since the onset of COVID-19, with the FBI at one point announcing that they were fielding as many as 4,000 cybersecurity-related complaints per day, a 400 percent increase on pre-pandemic levels.
But that's not the only thing that has changed. With the proliferation of connected devices, these attacks—and the vulnerabilities they seek to exploit—continue to take on new forms.
All of this was top of mind for a team of information security risk management and automation professionals at Johnson & Johnson led by Judy Mason and Michael Syntax.
Looking at our manufacturing sites around the world, which are essential for producing life-saving and life-changing treatments for patients, the team wondered: How effectively are we monitoring connected devices on site? How are these devices protected? And what can be done to better identify and mitigate vulnerabilities?
Today, they've come back with an answer: a monitoring device that uses machine learning and advanced analytics capabilities to construct a model of all communications taking place between IoT devices at a worksite. As such, in addition to monitoring the overall health of connected devices, it can also alert our team to suspicious outliers and help us take preventative action.
It’s a great example of how interdisciplinary teams are applying data science at Johnson & Johnson to bring innovative ideas to life. Indeed, the project would never have even gotten off of the ground if it weren’t work the work of a massive cross-functional team, which included data scientists, network engineers, manufacturing site managers, logistics and supply chain professionals—as well as literally hundreds of others.
Join Our Team to Collaborate, Innovate and Make a Global Impact
With global teams assembled—and such an impressive range of entries across functions—it didn't make sense to announce "winners" at our Data Science and Intelligent Automation Showcases. Instead, the goal was to build connections, collaboration and camaraderie between data science practitioners. And after three days of eye-opening presentations, the event was above all else testimony to the vibrancy and strength of that community within Johnson & Johnson.
Looking to apply your data science expertise in a massively collaborative environment, see your ideas put into action and positively impact the lives of people everywhere? If so, we can't wait to meet you. Be sure to check out all the different ways you can join our community of data scientists at Johnson & Johnson today.
Plus, in the interim, be sure to sign up for our Global Talent Hub. It’s a great way to stay in touch, learn more about life at Johnson & Johnson—and even receive updates about jobs that might interest you in the future.