AWS Startups Blog
Syllable Automates Healthcare’s Frontline with AWS
(This article was created in partnership with Chirag Dhull, Principal Product Marketing Manager for Amazon Quicksight)
Given the times we are living in, healthcare organizations are going through digital transformation at a faster rate than ever before.
And that was before the pandemic. Almost overnight, the healthcare system was hit with a new wave of demand, a lack of resources, and the need to separate the non-urgent services from the essential.
Syllable was perfectly poised to help. Founded in 2016, the Bay Area-based company works on automating the “frontline” of healthcare, or the first point of contact between patients and providers. These are the everyday tasks—looking for a medical record, scheduling an appointment, or getting information from your doctor—that are often more difficult than they should be.
Built on AWS since day one, Syllable uses natural language processing to create bots that can speak to patients, understand what they’re asking, and help them get what they came for. “It’s all about helping patients and providers get the information that they need as quickly as possible,” says Andrew Rogers, Syllable’s co-founder.
Sharing the burden
The nature of the COVID-19 pandemic meant Syllable had a new opportunity to help ease some of the load placed on the healthcare system. “The onset of COVID-19 reduced access to healthcare providers, and restricted bandwidth for the patients due to additional safety procedures,” Rogers says.
The company quickly offered informational web bots to help concerned patients figure out whether or not they needed in-person care, which could offload some of the requests swamping hospital workers. But that new service might have been impossible two years ago, when Syllable was hitting a wall with its existing data infrastructure.
By serving as the frontline for entire healthcare systems that encompass networks of hospitals, Syllable brings in massive amounts of data on patient behavior.
Their old data pipeline was struggling to keep up with the volume of that data, sometimes taking hours to create new customer reports. And a previous BI tool forced the team to create and manage separate reports for every user, which was soon unsustainable. “Every customer we added was an extra burden on not only our engineering team, but on our product team to be able to support the reports that we were generating,” says Rogers.
Syllable needed a new pipeline that could post-process and store data in near real time, and then could feed data into reporting dashboards to be created once for multiple customers.
One dataset, many insights
The company now streams and processes its data with Amazon Kinesis Data Firehose, and uses Redshift for a data warehouse. But because of Syllable’s business model, the move to Amazon QuickSight may have made the most visible difference.
Whenever Syllable’s bots help patients book an appointment faster or find the right doctor, that can save time and stress. Every time that happens, the platform also adds to a valuable store of data about what patients want from their providers, and what they’re missing.
“We can extract semantic meaning from phone conversations by transcribing them and classifying them, which is not something that the healthcare system has ever had access to before,” says Rogers. What are the top specialities patients are calling about? How many patients are searching for information about specific exams? How long are they spending on the site?
Syllable’s AWS Architecture
Using QuickSight, Syllable can produce individualized reports based on questions like these, helping healthcare systems understand how to better serve patients. With the row-level security feature, Syllable’s engineers can build a single dataset and give each customer access to only the data they need to see. And Rogers credits SPICE—the tool’s Super-fast, Parallel, In-memory Calculation Engine—for speedy visualizations and analysis, processing tens of millions of rows in seconds.
“We don’t have to spend all of our time worrying about pipelines and restarting failed jobs,” he says. “It’s working behind the scenes for us so we can actually make forward progress, like looking at our data and coming up with new, innovative ideas.”