Davie Chen
Researcher · Faculty of Animation and Intermedia, University of Arts in Poznań
Studying how generative AI can assist scientific and technical communication—from figures and patents to instructions, editable diagrams, and manuscripts—and building the tools that put the research into practice.
Research Focus
My work sits at the intersection of generative AI and scientific and technical communication. I study how AI can create and recover structured visual artifacts—from biomedical figures and patent drawings to instruction illustrations and editable diagrams—while preserving clarity, domain constraints, and downstream editability.
These projects translate that research agenda into practical tools for researchers, inventors, documentation teams, and other knowledge workers. Some began as research questions; others extend the same inquiry into adjacent communication problems.
Projects
SciDraw AI →
AI scientific figure generation for researchers.
A platform that helps biomedical and life-science researchers generate publication-ready figures, graphical abstracts, and conceptual diagrams from natural-language descriptions. Built on a taxonomy of 2,792 real-world use cases.
PatentFig AI →
AI patent illustrations aligned with USPTO, EPO, and CNIPA standards.
Patent drawings have stricter formal requirements than any other figure category in technical communication. PatentFig AI generates compliant line-drawing illustrations and supports DXF/SVG export for downstream patent prosecution workflows.
ManualFig.ai →
AI instruction illustrations for manuals, IFUs, and assembly guides.
Turns product photos, CAD screenshots, and written steps into clean line drawings, step-by-step panels, callouts, warnings, and editable SVG figures for technical documentation.
LayerBack →
From diagram images to fully editable Visio and PowerPoint files.
Reconstructs screenshots and PDF diagrams as native shapes, live connectors, and editable text, with VSDX, PPTX, draw.io, and SVG exports plus browser-based editing.
Data2Paper →
From structured data to publication-ready manuscripts.
An automated drafting assistant for researchers who need to turn datasets, case records, or experimental results into well-structured academic prose. Currently focused on case reports and short-form scientific communications.
Selected Preprints
Data2Paper: An AI-Assisted Pipeline for Statistical Analysis and Research Paper Drafting from Survey and Clinical Data
AI-Assisted Scientific Illustration in Biomedical Research: A Practical Taxonomy and Prompt Engineering Guide Based on Analysis of 2,792 Real-World Use Cases
AI-Generated Figures for Patent Applications: Requirements, Challenges, and a Domain-Specific Approach
AI-Generated Figures in Academic Publishing: Policies, Tools, and Practical Guidelines