Predict
H&E-based Biomarker Profiling
OI Predict uses H&E images to predict genomic alterations, protein expression, and clinical outcomes
Discuss your needs
What Can Be Predicted From an H&E Image?
With a powerful histopathology foundation model and the right training data, standard H&E images can open the door to an extraordinary range of molecular and clinical insights. Models can be trained to predict molecular alterations, including mutations, gene fusions, and exon-skipping events; protein expression and overexpression; and clinical outcomes such as treatment response.
Research Applications (RUO)
- Sample triage for molecular analysis
- Retrospective biomarker analysis
- Cohort enrichment for research
- Biomarker discovery
- Real-World Data (RWD) studies
Investigational and Clinical Applications (LDT, IVD & CDx)
- Clinical-trial prescreening
- Site feasibility assessment
- Molecular testing prioritization
- Rule-out prescreening
- Rule-in or reflex-testing trigger
Research Applications (RUO)
- IHC testing prioritization
- ADC biomarker research
- Virtual protein expression profiling
- Expression prevalence studies
Investigational and Clinical Applications (LDT, IVD & CDx)
- IHC pre-screening or rule-out
- Adjunct to pathologist scoring
- Classification of clinical expression status (0, 1+, 2+, 3+)
- Semi-quantitative expression score
Research Applications (RUO)
- Prognostic research on retrospective cohorts
- Trial-enrichment modeling
- Risk-based cohort stratification
- Biomarker discovery for resistance mechanisms
Investigational and Clinical Applications (LDT, IVD & CDx)
- Prognostic risk assessment
- Predictive treatment benefit assessment
- Estimate likelihood of response or resistance to a defined therapy
Predict
Solutions
OI Predict includes validated products as well as a custom panel development offering. The breadth and depth of our CanvOI foundation model enables development of custom classifiers tailored to specific tissue types, indications, and use cases.
The Lung Prediction Panel predicts eight genomic alteration biomarkers in NSCLC directly from H&E images, delivering results in under 20 minutes. Research applications include sample triage, retrospective biomarker analysis, and real-world data studies.
LungOI is an AI-powered, multigene test that screens for actionable genomic alterations in non-small cell lung cancer (NSCLC). Available as a laboratory-developed test through a CLIA-certified, CAP-accredited laboratory, LungOI has an established CMS reimbursement rate. With a demonstrated negative predictive value of 99%, a negative result can support timely treatment initiation when clinically appropriate. A positive result can prompt rapid confirmatory testing; if the alteration is confirmed, targeted therapy may be initiated as clinically appropriate.
AI-Enabled Trial Prescreening
Prioritize samples for molecular confirmation
Potential Investigational Benefits
Prioritize samples for confirmation
Streamline enrollment
Focus testing on prioritized samples
Reduce enrollment costs
-
01Analyze
NSCLC
H&E Cohort Analyze with the Lung
Prediction Panel -
02Prioritize
Prioritized
Sample Subset Advance to Molecular
Confirmation -
03Confirm
Protocol-required
Molecular Testing Determine
Trial Eligibility
AI-Enabled Trial Planning
Optimize site feasibility
Potential Site Feasibility Benefits
Inform sample-access and feasibility planning- Compare predicted alteration prevalence across candidate sites
- Prioritize sites for deeper molecular feasibility assessment
-
01Access
Historical
H&E Slides Access available
site archives -
02Predict
EGFR Mutation
Status Predict directly
from H&E -
03Estimate
Site-level EGFR+
prevalence Estimate expected
positive cases -
04Prioritize
Rank and
prioritize site Focus enrollment
resources -
05Confirm
Molecular
testing Confirm
EGFR status
The breadth and depth of CanvOI enables development of custom classifiers tailored to specific tissue types, indications, and use cases. For example, a classifier can be trained to detect KRAS and PIK3CA mutations in colorectal cancer tissue, enabling efficient screening from pathology images.
Beyond mutation detection, custom classifiers can help predict other genomic alterations, protein expression, and clinically relevant outcomes in tissue samples.
Start with a proof-of-concept study to assess feasibility and guide the scope and design of subsequent development and validation efforts.
Initiate a Proof-of-Concept
Lung Prediction Panel (RUO) is For Research Use Only. Not for use in diagnostic procedures. Lung Prediction Panel results are intended for research sample prioritization and are not intended for patient diagnosis, treatment decisions, or determination of clinical-trial eligibility.
LungOI is performed as a laboratory developed test in a CLIA-certified, CAP-accredited laboratory and has not been cleared or approved by the United States Food and Drug Administration (FDA).
Lung Prediction Panel (IVD) is currently under development and has not been cleared by the FDA. Following FDA clearance, the test is intended to be offered for the indications specified in its FDA-cleared labeling.