OWKIN
MODELS

OWKIN LOOP DATA + OWKIN LAB ALGORITHM = OWKIN MODEL

Owkin has a catalog of 30 live models as well as 40 additional models in the pipeline. All models are built using public and Owkin Loop data. These models are available to medical researchers via Owkin Studio and to our pharma partners through Owkin Solutions. These models differ from traditional black-box models as they are built using interpretable AI, which enables us to go further in the research and identify biomarkers responsible for the predictions. The discovery of new multimodal biomarkers is essential to identify new biological targets, optimize clinical trial design using patients subgroups, and identify patients eligible for specific treatments.

Our catalog consists of three types of models: Data Enrichment, Outcome Prediction, and Patient Identification.

Owkin has developped a strong expertise in Hepatocellular Carcinoma (HCC), the most common form of primary liver cancer, and the fourth cause of cancer-related death worldwide. Discover how we work, and how we address HCC drug development with machine learning models, specifically built for increasing clinical trial efficiency.

Owkin HCC Expertise

Data enrichment models

Intended for translational researchers and pathologists, these models connect histology with molecular markers.
Data enrichment
Ongoing

Automatic Segmentation of Body Composition

CT Scans
Find out more
Data enrichment
Productized

HE2RNA Transcriptomic Prediction

Multimodal Project
- RNA-Seq Data
- H&E Slides
Find out more
Data enrichment
Productized

CD3 Virtual Staining

CD3 Detection
- H&E Slides
Find out more
Data enrichment
Ongoing

Virtual Staining Models

Biomarkers Detection
- CD8, D2-40, CD15, CD38, CD117, AE1/AE3, CD163
- H&E Slides
Find out more

Outcome prediction models

Intended for translational researchers, development leaders and commercial officers, these models can be implemented across all stages of the clinical drug development process to enhance clinical trial design and evaluation, as well as to optimize product strategy. In addition to the Oncology models outlined below, we also have various models for hematology and autoimmune diseases such as IBD and Lupus.
Outcome prediction
Productized

Meso-Net Predicting survival in Mesothelioma

Predicting Prognosis
- H&E Slides
Find out more
Outcome prediction
Ongoing

Predicting prognosis and treatment response in Melanoma

Federated Project
Multimodal Project
- Dermoscopy Images
- H&E Slides
Find out more
Outcome prediction
Ongoing

HCC-Net Predicting prognosis and treatment response in HCC

Multimodal Project
- H&E Slides
- Radiology MRIs
Find out more
Outcome prediction
Ongoing

Predicting relapse in Breast Cancer

Federated Project
- H&E Slides
Find out more
Outcome prediction
Ongoing

Predicting immunotherapy outcomes in NSCLC

Federated Project
Multimodal Project
- Radiology CT-Scans
- H&E Slides
Find out more

Patient identification models

Intended for commercial leaders and precision medicine heads, these models identify patients that would benefit from a specific treatment.
Patient identification
Ongoing

Predicting NTRK

Identifying NTRK Patients
Pan-Cancer Prediction
-H&E Slides
Find out more
Patient identification
Ongoing

Predicting HRD Status

Identifying HRD Patients
Pan-Cancer Prediction
- H&E Slides
Find out more