Rely on findings from published data, OR
Wetlab experiments in simplistic models, OR
Generate bespoke data in isolated data modalities (e.g. genomics) and use traditional statistics to measure correlation between target expression and clinical outcomes
Data modalities used: 1-2
Generate bespoke multi-modal data and use advanced, integrated ML algorithms to identify targets via multimodal “explainable” AI
Data modalities used: 6+
Clinical signals (e.g. biomarkers) are assessed via manual pathologist review (histology), limiting potential to effectively leverage complex signals
These signals for precision medicine are becoming increasingly complex (e.g. TME)
More complex signals may be challenging to measure accurately or scalably, especially if they aren’t standard of care
Leverage machine learning to detect and measure novel, complex signals quickly and with high fidelity
In addition to custom analyses, we are building a suite of “off-the-shelf” translational research applications:
Tissue QC
Cell detection
Cell classification
Tissue segmentation
Mitosis detection
Complex IHC algorithms
(membrane, cytoplasm, nucleus intensity scoring)
Membrane segmentation
Omics analyses
Biomarkers in clinical trials and companion diagnostics are assessed via manual review
Generate advanced machine learning algorithms that power:
GCP-compliant biomarker scoring in clinical trials
Interim analyses for ongoing clinical trials
Complementary diagnostics
(e.g. patient pre-selection/screening)
Companion diagnostics
