Task-Specific Model Allocation Medical Papers PICOS Information Extraction

In the field of medical research, extracting PICOS information which includes population, intervention, Comparison, outcomes, and study design from medical papers, plays a significant role in improving search efficiency and guiding clinical practice. Based on the PICOS key information extraction task released from the China Health Information Processing Conference (CHIP 2023), we proposed a method based on Task-Specific Model Allocation, which selects different models according to the characteristics of the abstract and title data. Additionally, incorporating techniques such as multi-task learning and model fusion. Our method achieved first place on the leaderboard with an F1 score of 0.78 on List A and 0.81 on List B.