Large Language Model for Automating the Analysis of Cryoprotectants.

Area of Science:

• Biotechnology
• Bioinformatics
• Scientific Data Management

Background:

• The exponential growth of scientific literature presents challenges for manual data extraction.
• Efficient retrieval of specific information, such as cryoprotectant data, is crucial for research advancement.
• Existing methods for data extraction can be time-consuming and labor-intensive.

Purpose of the Study:

• To develop and validate an automated system for extracting cryoprotectant information from scientific publications.
• To leverage artificial intelligence, specifically large language models (LLMs), for enhanced data analysis.
• To create a user-friendly interface (Telegram bot) for accessible data retrieval.

Main Methods:

• Utilized a generative pretrained transformer (GPT) model for natural language processing.
• Integrated the GPT model with a Telegram bot for an interactive user interface.
• Developed data preparation, algorithm development, and system validation protocols.
• Trained and tested the system on a substantial dataset of scientific articles.

Main Results:

• The automated system successfully extracted relevant cryoprotectant and bacteria data from scientific articles.
• Significant reduction in the time required for researchers to gather essential information was observed.
• The system demonstrated accuracy in recognizing and extracting critical data points.
• Identified limitations in term specificity, indicating areas for further model refinement.

Conclusions:

• The AI-driven approach effectively automates the extraction of cryoprotectant information, optimizing research workflows.
• The user-friendly Telegram bot enhances accessibility and efficiency for researchers in cryopreservation and related fields.
• Further model refinement and specialized training are recommended to improve accuracy and reliability for specific scientific applications.