How Juisci saved 100 hours by automating large volumes of text synthesis

Automating the synthesis of a large volume of texts

Juisci is a mobile app that allows healthcare professionals to keep up to date with the latest publications and clinical studies t founded in 0 by Robin Roumengas, an healthcare entrepreneur, and Dr. David Luu, a heart surgeon and entrepreneur.

Easily access scientific publications and journals

Every year, healthcare professionals have to navigate through the nearly 2.5 million new scientific papers that appear! With that kind of volume, it's easy to understand how difficult it is to keep up with the latest advances and to deliver important information to patients. Juisci's job is to summarize these scientific papers and make them available to users via an ergonomic and enjoyable mobile application.

Fresh squeezed, straight from the source

This mission leads to a number of challenges, which is why the Juisci team brings together complementary expertise to meet at the crossroads of mobile development, UX design, and medical experience. The missing element was an NLP technology allowing them to structure and analyze the scientific papers in an automated way. That's where Lettria stepped in.

A fruitful collaboration between Lettria and Juisci

Over the first several weeks, Lettria developed tools to synthesize scientific papers (for more technical details on text synthesis, click here). The deliverable, in API format, was then integrated into the Juisci application. That way, each scientific paper selected by Juisci is directly sent to Lettria's analysis engine, which transforms a document of several dozen pages into less than 15 lines.

To check the quality of the abstracts, Lettria's abstracts for 100 papers were compared with the handwritten abstracts of 100 papers. Results showed that the information quality of a Lettria abstract is identical, if not better, than the handmade abstract. The time saved is also considerable: the solution designed to summarize 100 scientific papers took 200x less time than its handmade equivalent (100 hours). In practice, the application has been acclaimed by users, obtaining an NPS score of 8.5.

"The team of data scientists at Lettria was able to quickly understand the language processing challenges at Juisci and come up with a solution that allows for the efficient and scalable synthesis of scientific papers." -- Robin Roumengas, co-founder and CEO of Juisci

Next steps

Juisci isn't stopping there, and intends to strengthen its collaboration with Lettria following this successful first phase. The average volume of articles processed for synthesis will increase considerably, from one hundred to several thousand. New functionalities and expanded medical fields will also be developed.

‍