Exploring the Polyethylene Glycol-Modified Drug Patent Landscape by Deep Learning

Area of Science:

• Biopharmaceutical science
• Drug development
• Intellectual property analysis

Background:

• Polyethylene glycol (PEG) modification is crucial for improving the pharmacokinetics and clinical utility of biopharmaceuticals.
• Understanding the patent landscape of PEG-modified drugs is essential for strategic R&D and commercialization.

Purpose of the Study:

• To develop a comprehensive analytical framework for characterizing the patent landscape of PEG-modified drugs.
• To integrate current status analysis, technology flow, and value assessment of PEG-related patents.
• To provide a step-by-step analysis of the global patent landscape for PEG-modified drugs.

Main Methods:

• Compiled and analyzed 99,540 PEG-related patents filed globally between 2014 and 2023 using the Derwent patent database.
• Employed descriptive statistics, social network analysis, machine learning, and deep learning techniques for patent analysis.
• Utilized XGBoost for patent transfer prediction and RoBERTa-BiLSTM-MLP for early-stage patent quality assessment.

Main Results:

• A significant increase in PEG-related patents over the last decade, with China filing the most but having narrower technological scope.
• The United States leads in inventor and assignee numbers; patents from developed regions receive more citations.
• PEG-modified proteins and peptides are the most commercially active category, with advanced machine learning models showing high accuracy in prediction and quality assessment.

Conclusions:

• The study offers a multi-layered understanding of the PEG-modified drug patent landscape, encompassing static features, dynamic trends, and qualitative/quantitative evaluations.
• Findings provide a valuable analytical reference for identifying technology trends in the field of PEG-modified drugs.
• This research supports strategic decision-making in the development and commercialization of PEGylated therapeutics.