In the ever-evolving world of healthcare analytics, ZS strives to continually stay at the forefront of innovation. We support our people in exploring new technologies and applying their learnings to solve complex problems. Recently, a team of ZSers participated in a case challenge sponsored by the Centers for Disease Control and Prevention’s (CDC) National Center for Injury Prevention and Control (NCIPC). The team used unsupervised machine learning—a machine learning model that can learn without human supervision—to understand medical records’ narrative data about why older adults fall, ultimately winning “Most Novel Approach” in this case challenge.
Where passion changes lives
Falls among adults aged 65 and older are the leading cause of injury-related deaths in the United States, yet many falls can be avoided with proper preventative measures. To better understand this issue, the CDC and NCIPC—in conjunction with the National Electronic Injury Surveillance System (NEISS)—collected tens of thousands of medical record narratives and asked participants to use unsupervised machine learning to comb through the records for insights.
Narrative data sources are often difficult to work with because of the intensive manual coding required to make sense of the research. This case challenge tasked participants with using AI to improve efficiency when sorting through this underused well of information.
ZSers Yash Aggarwal, Umar Faruque, Aayush Khurana and Gaurav Yadav partnered to tackle this challenge. Guided by years of expertise in AI and analytics, the team took a unique approach by first splitting the plethora of data into manageable sample sizes. From there, they were able to understand the three key learnings the data offered: ranking the severity of the fall, activity performed before the fall and reason for the fall.
Still, the team was faced with a difficult challenge—how to decide on comprehensive labels to categorize the data. Seeing an opportunity to think differently, the team used AI with large language models (LLMs) to read each narrative, develop labels for categorization and tag each data set accordingly.
“ZS has used AI in many projects, from market research to clinical trials,” Umar says. “This challenge gave us the opportunity to expand our knowledge and use our expertise in a new way.”
The team’s use of sampling and their unique categorization made their research stand out from other competitors, earning the team the “Most Novel Approach” award.
“This is what ZS is known for,” says Gaurav. “ZSers thrive on looking at our own research and finding new ways of doing things, rather than relying on the status quo.”
Yash echoes this sentiment. “We take complex problems and figure out how to solve them in an innovative way using the latest technologies.”
The team’s unique approach to understanding large data sets enabled it to uncover patterns, painting a comprehensive picture of fall risks. Thinking differently allowed the team to discern critical insights that hold the potential to revolutionize fall prevention strategies.
This novel approach to organizing large data sets holds immense potential for application in similar situations across various domains. By harnessing the power of AI and machine learning, this innovative methodology enables efficient extraction of insights from complex data sets.
“Medical record narratives, like the ones we classified, are rich in potential, but often under-explored due to the challenges associated with manual coding procedures,” Aayush explains. “This new approach not only enhances the accuracy of analysis but also unveils patterns and relationships within the data that might otherwise go unnoticed.”
Leveraging such techniques in future projects can empower ZSers and their clients to navigate vast data sets, extract meaningful insights and make informed decisions, ultimately contributing to improved health outcomes around the world.
The team wants to extend a special thank you to Sagar Madgi and Mayank Shah.
To stay updated on how ZS is leading innovation, follow us on social media.
Life at ZS
Learn what it’s like to be a ZSer: Facebook, LinkedIn and Instagram
View ZS’s open roles.
Add insights to your inbox
We’ll send you content you’ll want to read – and put to use.
