Fine-tuned large language models enhance influenza forecasting

Chenxiang Li, Wenjing Gao, Qiqiao Zhang, Yue Zhang, Bowen Zhao, Jule Yang, Li Qi, Jun Ding, Dechao Tian · 2026

Read Fine-tuned large language models enhance influenza forecasting by Chenxiang Li, Wenjing Gao, Qiqiao Zhang, Yue Zhang, Bowen Zhao, Jule Yang, Li Qi, Jun Ding, Dechao Tian (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/PMC13615498.txt.

Description

Chenxiang Li and colleagues test in 2026 whether fine-tuned large language models, the text systems behind modern chatbots, can forecast flu-like illness better than standard tools. They built a light training method that adds small embedding and prediction layers to models such as Llama2 and GPT2. Then they tested it on seven weekly flu surveillance datasets. Each set held only about 523 time points.

The tuned language models beat a bench of rivals, from classic statistical models to recent time-series networks, on accuracy and steadiness, above all for long-range forecasts. Even with no tuning, the raw models tracked broad epidemic trends about as well as the classic method. Since the setup needs no cloud transfer of data, it suits health agencies that must keep local data private and work with short records.

Catalog record

CREATORChenxiang Li, Wenjing Gao, Qiqiao Zhang, Yue Zhang, Bowen Zhao, Jule Yang, Li Qi, Jun Ding, Dechao Tian
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE65175
LICENSEhttps://creativecommons.org/licenses/by/4.0
RIGHTSCC BY
SOURCEEurope PMC

Source: https://europepmc.org/article/MED/PMC13615498

Browse the full library · About the Tower of Babel