Spatial biologists guide and check AI analyses of tissue data
How Spatial Biologists Direct and Verify AI-Assisted Analyses
Human-Computer Interaction
Summary
Spatial biologists study tissues by looking at where cells are and what they do. They often use computers to analyze complex tissue images, but this can be slow or hard. The authors studied how these biologists use an AI helper to do this work and how they make sure the AI's results are correct. They found that biologists like the AI’s help with drawing plots and finding cells, but they still carefully check the AI’s work using their own knowledge and extra tools. The authors also suggest how to design better tools that make it easier to control, see, and verify AI analyses.
What this means in practice
- •For biomedical data teams: Use AI helpers to speed up plotting and cell identification in spatial tissue analysis while ensuring data trustworthiness through guided verification.
- •For bioinformatics software developers: Design interactive tools that integrate AI assistance with clear execution control and verification features to support spatial biology workflows.$Commercial implications: Enables creation of advanced spatial biology analysis software products with AI collaboration features for life science research labs.
Authors
Ella Hugie, Alexandra Irger, Chiara Schiller, Lukas Hatscher, Denis Schapiro, Hanspeter Pfister
Abstract
Spatial biologists use visualization to assess computational analyses of tissue data. We examine how they direct and verify analyses when an AI agent performs this work. We synthesized workflows from fourteen contextual inquiries and conducted a formative pilot followed by an observational study with ten spatial biologists using Claude Science on their own data. Participants valued help with plotting, locating cells of interest, and tasks they found laborious or could not otherwise perform. Assessing the agent's work involved obtaining suitable evidence, sometimes through additional work in external tools, and interpreting it using knowledge of the tissue and its markers. Scientists also sought information about ongoing computation to decide how analysis should proceed. We contribute a workflow synthesis, an empirical account of scientists directing and verifying agentic analyses, and four design directions addressing execution control, familiar interactive views, source and execution information, and accessible verification across computing setups and experience.