Inspire Uses ML to Connect Millions of Patients and Caregivers on AWS
Inspire, the vital online health community and important partner for life science companies, has a two-part mission. First, Inspire connects patients suffering from thousands of conditions and their caregivers with online tools, resources, and one another in condition-specific support groups. Second, Inspire connects pharmaceutical companies and other medical institutions conducting clinical-trials research—real-world evidence studies—to health-outcome studies on patients suffering from those diseases. “We seek to accelerate life-changing discoveries through our vital community of connected patients and caregivers,” says Richard Tsai, senior vice president of marketing at Inspire. More than 50 million people in 150 countries have used Inspire since 2015, and more than 2 million registered people with over 5 million reported medical conditions use the health community as of February 2021. Thousands more register every week, making it the largest and fastest-growing virtual support community for patients living with cancer, rare diseases, and chronic conditions, enabling them to actively share experiences and learn about diagnoses and treatments.
As Inspire looked to build on its success and keep up with its own growth, it needed to overcome the scaling challenges posed by its legacy on-premises infrastructure. Using managed solutions from Amazon Web Services (AWS), the company discovered numerous advantages in the cloud, including faster iteration, greater flexibility, and multiregion availability. Inspire found particular success using Amazon SageMaker, a fully managed service that provides every developer and data scientist with the ability to build, train, and deploy machine learning (ML) models quickly. Using a solution driven by Amazon SageMaker, the company saw a substantial increase in user engagement across all channels. Inspire’s use of AWS also helped streamline the process by which pharmaceutical companies conducting clinical trials or medical research could connect with relevant patient data—an important step in the development of life-saving therapies.
We’ve migrated most of our development to AWS Lambda functions. Between that, caching, and Aurora, we honestly don’t pay attention to scaling anymore.”
Vice President of Engineering, Development Infrastructure, and Data Science, Inspire
Scaling Automatically and Accelerating Innovation
Before using AWS, Inspire was operating around 20 physical servers in Ashburn, Virginia. The company ran into scaling problems, often having to wait 2 months to order and install a server and up to 3–6 months total to expand its capacity. In one instance, a database server upgrade resulted in a time-to-market delay of 3 months. Inspire explored the possibility of migrating to the cloud beginning in 2016, ultimately migrating its database to Amazon Aurora, a MySQL- and PostgreSQL-compatible relational database that is built for the cloud and combines the performance and availability of traditional enterprise databases with the simplicity and cost-effectiveness of open-source databases. “The most compelling thing to us was the database offering, Aurora,” says Brian Loew, founder and CEO of Inspire. “That blew everything else out of the water.”
Also involved in the new infrastructure was AWS Lambda, which lets customers run code without provisioning or managing servers. “We’ve migrated most of our development to AWS Lambda functions,” says Anthony Sheetz, vice president of engineering, development infrastructure, and data science at Inspire. “Between that, caching, and Aurora, we honestly don’t pay attention to scaling anymore.” Able to scale automatically and iterate more quickly on AWS, Inspire would eventually increase the frequency of its releases from one every 2 weeks to several a day, thus accelerating innovation and broadening its operations.
“We essentially forklifted our entire infrastructure onto AWS, keeping the same architecture,” adds Sheetz. “And then, once we were on AWS, we started playing with the toys.”
Using Machine Learning to Boost Engagement
A big part of Inspire is its content recommendation engine, through which it directs users living with particular conditions to relevant posts or articles. Integral to this engine is Amazon SageMaker, which Inspire uses in its development process to build and modify custom deep-learning models in 1–2-week cycles. “We’re now able to match users to relevant content by analyzing behavioral patterns and deploy these models with ease—all using Amazon SageMaker,” says Teja Talluri, director of data science at Inspire. “Amazon SageMaker provides a more scalable way of recommending content that we couldn’t manually curate ourselves.”
The sophisticated ML solution improved the content recommendation engine’s ability to suggest relevant content to two million registered users, pulling from Inspire’s massive library of 1.5 billion words written about 3,600 conditions. Ultimately, this solution enabled Inspire to accurately connect patients and caregivers with more personalized content and resources—including rare-disease information and treatment pathways.
When Inspire ran tests comparing new and old versions of its content recommendation engine, the metrics clearly showed more-robust engagement because of the company’s ML-driven personalization efforts. Personalized email subject lines led to a 281 percent email open rate increase. And once users opened these emails, the new recommendation engine boosted the click-through rate by 914 percent, contributing to a 119 percent increase in average page views on the site. Inspire also saw that its retention rate, the number of users who remained active after 4 weeks, had increased by 550 percent since adopting the new content-recommendation engine.
Although the numbers are impressive, the human impact they represent is what matters most to Inspire. “We’ve received a lot of testimonials in which patients have said, ‘The content you’ve recommended to me is so relevant,’” says John Novack, head of patient engagement and senior director of communications for Inspire. “We’ve never had that in the past. Now we have people telling us we’ve changed their lives—or even saved their lives.”
Changing the Way Pharmaceutical Companies Find Critical Research Data
Inspire’s other critical mission is to connect pharmaceutical companies researching new therapies with patients who may benefit from these therapies or at least provide useful data to those patients. Central to this use case are Amazon Redshift, a fast, simple, cost-effective data warehousing service, and Amazon Comprehend Medical, a natural language processing service that makes it simple to use ML to extract relevant medical information from unstructured text.
When Boston Children’s Hospital and the pharmaceutical company Pfizer sought specific insights to aid in the development of new lung cancer treatments, they faced a considerable challenge in finding data from a narrow set of patients: those with some combination of lung cancer and an autoimmune disease. Traditionally, researchers would have to reach out to individual investigators and clinicians to find patients who could supply relevant data—an extremely time-consuming process that could take years and still only find a few similar cases. However, Inspire’s AWS-backed natural language processing capability enabled Inspire to search the profiles of tens of thousands of users who had consented to show up in such searches, and the company ultimately found more than 100 qualifying participants within a few weeks. Stefan McDonough, who is no longer at Pfizer but was the executive director of genetics at the time of the project, described the Inspire community as “an extraordinary resource,” citing its rich pool of patients who are willing and eager to share medical information to advance treatments.
Connecting People with Profoundly Relevant and Impactful Information
After migrating to AWS, Inspire has seen a significant change in the way it does business. “AWS has given the whole software side of the house the ability to do what the business side needs done quickly and simply,” says Sheetz. “Now we can spend far more time focusing on giving our members new things.” Inspire expects that AWS data science–focused tools will help with the next phase of its business and play a key role in increasing revenue by an order of magnitude.
And at the center of Inspire’s mission are its users: the patients and caregivers who look to the company for help finding everything from practical information about rare illnesses to communities of people going through similar experiences. “It’s very important to be able to unite people around the world with similar health experiences,” says Sheetz. “This ability—to bring patients who have rare diseases together in a single place where they can share their experiences, regardless of where they live or what language they speak—has a profound impact.”
Inspire is an important partner for life science companies, offering a unique resource that collects permission-based data about patients’ rich and varied health journeys and provides insights into myriad conditions and their impact on patients.
Benefits of AWS
- Improved user experience
- Increased email open rate by 281%
- Increased email click-through rate by 914%
- Increased average page views on the site by 119%
- Increased retention rate by 550%
- Helped recruit all 100+ candidates for a Pfizer clinical trial in weeks
AWS Services Used
Amazon SageMaker helps data scientists and developers to prepare, build, train, and deploy high-quality machine learning (ML) models quickly by bringing together a broad set of capabilities purpose-built for ML.
Amazon Aurora is a MySQL and PostgreSQL-compatible relational database built for the cloud, that combines the performance and availability of traditional enterprise databases with the simplicity and cost-effectiveness of open source databases.
AWS Lambda is a serverless compute service that lets you run code without provisioning or managing servers, creating workload-aware cluster scaling logic, maintaining event integrations, or managing runtimes.
With Amazon Redshift, you can query and combine exabytes of structured and semi-structured data across your data warehouse, operational database, and data lake using standard SQL.
Companies of all sizes across all industries are transforming their businesses every day using AWS. Contact our experts and start your own AWS Cloud journey today.