Artificial Intelligence in Action
Improved Colonoscopy Effectiveness
A leading research hospital sought to increase the detection of certain medical conditions from colonoscopy samples to improve patient outcomes.
Quantori created a model that analyzes colonoscopy images and predicts the presence of cancer, tumors or other diseases.
Based on just these images, the system is able to accurately diagnose ill patients and give healthy patients the all-clear.
Custom Engineering Solutions to Study Electronic Healthcare Records
A pharma startup integrates Real-World Data (RWD) with the latest biological and computational methods to discover novel therapeutics for a broad range of autoimmune diseases. In order to efficiently study various formats of RWD collected across a multitude of databases, the client required a broad range of custom engineering solutions (ETL Pipelines, Cloud Infrastructure, MLOps, etc.).
- ETL pipelines to group and post-verify patients for genomically distinguished populations (Python & AWS).
- Built cutting-edge tools for extracting, polishing, and consolidating data in different formats out of various RWD sources.
- Built an MLOps platform to easily manage experiments via an interactive user interface.
- Developed a standardized visualization toolbox to align with experimental needs.
- Integrated experimentation scripts within a High Performance Computing environment.
- The client gained high-performance standardize ETL pipelines to easily parse datasets from a multitude of sources.
- A highly customized ML platform was created to easily launch, monitor, and reproduce ML experiments.
AI/ML Application to Analyze Millions of Cell Nuclei
The client approached Quantori to implement a complex bio-computational analysis of high content images. It was a significant part of the research that required deep bioinformatic input. It was one of the projects where the team asked domain questions to unleash nuanced scenarios and suggest a balanced solution.
- Delivered a highly accurate image analysis model using AWS service.
- Prepared the molecular data for supervised machine learning.
- Implemented a pipeline with AWS SageMaker using Spot Instances.
- Introduced an effective labeling solution and onboarded biomedical specialists for the best accuracy.
Enabled the scientists to accelerate the analysis of nuclei images from thousands to millions per day and to achieve competitive precision on nuclei detection.