Biotechnology / Lab Automation
Automated PCR Results
CLIENT — Biotechnology start-up (UK)
An AI-driven solution to automate classification of PCR fluorescence curves, cutting manual interpretation from around 24 hours of work to about an hour.
The challenge
Laboratory staff faced bottlenecks manually classifying PCR fluorescence curves — a time-consuming, resource-intensive process.
Our approach
End-to-end analysis of data production, lab systems and user interactions; a neural-network classification model that reports confidence levels; a full application (web UI, ML model and database) integrated with upstream and downstream laboratory data flows; deployed as a standalone local server and later migrated to AWS.
Results
Manual interpretation that took around 24 hours of work was cut to about an hour.
Technologies
- Neural networks
- Django
- AWS
- Machine learning
Automating PCR fluorescence-curve classification — from around 24 hours of work to about an hour.