PatSight

PatSight is an AI-driven pharmaceutical patent data mining platform developed by 晶泰科技 (XtalPi). Designed for small molecule drug discovery, it addresses the significant bottleneck of manual patent analysis — a process that traditionally requires medicinal chemistry experts to spend days or even weeks manually compiling patent data, drawing chemical structures, annotating bioactivity data, and conducting structure-activity relationship (SAR) analyses. PatSight automates this entire workflow, reducing the time required to under one hour.

The platform is purpose-built for drug chemists, data scientists, IP professionals, and project managers who need rapid, accurate extraction and analysis of chemical and biological data from patents. PatSight automatically extracts key data from patents and exports results as CSV or SDF files containing structural SMILES, enabling seamless downstream analysis.

核心功能 — Core Capabilities

• High-precision molecular structure recognition: Leverages advanced AI models to accurately identify and extract compound structures from patent documents.
• Automated integration of structure and activity data: Automatically extracts relevant bioactivity information from patents and associates it with the corresponding molecular structures.
• User-friendly interface: Provides one-click navigation to molecular structure views and bioactivity data tables for easy review and manual verification.
• Significant time savings: Capable of extracting all compounds and associated bioactivity data from a patent within one hour — a process that previously took days to weeks.
• Multi-language patent support: Offers comprehensive analysis across patents in multiple languages including Chinese, English, and Japanese, covering a wide range of patent sources.
• Data management and analysis: Centralises molecular structure and activity data management, and supports systematic structure-activity relationship (SAR) analysis.

具体应用 — Key Applications

• SAR analysis: Intelligently extracts structural and molecular data from patents and literature, and integrates deeply with 晶泰科技's analysis tool MolValley to enable systematic structure-activity relationship studies — identifying correlations between molecular structure and biological activity to support rational drug design.
• Large-scale data extraction: Rapidly extracts large volumes of structural and molecular feature information from patents, providing rich training datasets for AI model optimisation and development.
• Patent landscape analysis: Performs in-depth analysis of patent data to provide a comprehensive overview of the intellectual property landscape, precisely identifying innovation opportunities and white spaces.
• Patent molecular library construction: Identifies and compiles molecular structures, bioactivity data, and related information from project-relevant patents to build a dedicated patent molecular library for R&D projects.

实际案例与用户反馈 — Case Study and User Feedback

• Demonstrated performance: In a documented case study, PatSight successfully extracted 1,500 molecular structures and associated bioactivity data within one hour.
• Medicinal chemist (Dr. Li): Highlighted PatSight's high molecular recognition rate, accuracy, and convenience for manual review and verification, noting significant time savings in patent analysis.
• Data scientist (Dr. Jiang): Praised the speed and accuracy of patent data acquisition, emphasising its value in providing large-scale data for data analysis and AI model construction.
• IP expert (Dr. Peng): Noted the system's high recognition capability and its convenience for manual review workflows.
• Project manager (Dr. Wang): Described PatSight as highly valuable for large-scale patent analysis and particularly useful in supporting go/no-go decision-making during drug development projects.

PatSight is available for free trial and is offered as part of 晶泰科技's broader suite of drug discovery software tools, which includes MolValley for integrated molecular analysis. The platform supports multiple patent languages and sources, making it suitable for global pharmaceutical R&D teams engaged in competitive intelligence, IP strategy, and data-driven drug design.