Talk Pedometer: LENA

The Arizona State University Artificial Intelligence Cloud Innovation Center, powered by Amazon Web Services (AWS), collaborated with LENA, a nonprofit organization dedicated to transforming children's futures through early talk technology, to explore innovative solutions for supporting early childhood educators’ use of data. 

Early talk is one of the most important factors shaping children’s brain development during the first few years of life. Research shows that back-and-forth interactions with adult caregivers are one of the most powerful tools to create a firm foundation of healthy brain architecture for children. During the first years of life, infant and toddler brains are forming more than 1 million neural connections every second, according to the Harvard Center on the Developing Child. That means that these years offer a unique developmental opportunity.

LENA’s unique talk pedometer technology measures these interactions and provides data to early childhood educators that in turn helps them better support brain development.

The collaboration between the ASU AI CIC and LENA focusd on making LENA data easier for teachers to understand and act on by generating textual insights from  early talk data collected in classrooms.

While LENA’s reports are designed to be easy to digest, they comprise both individual child and classroom interaction, and can run many pages in length. Gaining insight from this data and setting appropriate goals based on it can be overwhelming for busy teachers. Additionally, many  teachers are not trained in data analysis and interpretation. LENA is seeking ways to make it easier for teachers to quickly glean the most important insights from the data while minimizing mental load.

The ASU AI CIC, in partnership with LENA, developed a proof of concept (POC) that leverages large language models and sophisticated prompt engineering to generate key insights from complex classroom- and child-level data that teachers can mor eeasily understand and act on. To achieve these objectives, the project utilizes several key AWS services. Amazon S3 is used for secure storage of LENA’s datasets, ensuring they are easily accessible and scalable. Amazon RDS manages and analyzes the structured data collected from classrooms, while Amazon Bedrock was used for  insight generation. Together, these services enhance LENA’s ability to generate accurate textual insights while maintaining privacy and security.

The automation of data insight generation in early childhood education significantly increases the usability of large data sets in a field that has struggled to make use of them in the past. By harnessing the power of AI and machine learning, LENA can better support teachers’ professional development, ensuring that more children benefit from improved early learning experiences. This collaboration not only enhances LENA’s ability to deliver on its mission but also sets a precedent for how data-driven approaches can be applied in education to achieve better outcomes.

“Working with the CIC really opened our eyes to new ways of making data more accessible to teachers. It also showcased how quickly and easily we can build these innovations into our programs within AWS. As a Product Manager, I now have a much clearer vision for how we will integrate AI and ML to support more effective and impactful adult learning.” 

- Liz Pettit, Senior Product Manager, LENA