OncoSight: AI Screening for Early Cancer Detection

OncoSight leverages existing patient data from electronic health records, identifying high-risk patients using commonly available data points such as:

• Demographic information

• Past diagnosis codes

• Medications

• Lab values

• Procedure codes

We use time-series machine learning models to find predictive trends for future cancer risk using this data. It routinely scans the electronic health records to identify patients at high risk of developing cancer within the next 1-3 years without any patient-initiated action.

Integrates seamlessly with the EHR system

Our SMART on FHIR app integrates into any EHR provider system and serves a dashboard for the clinician to review the explanation behind the predictions for various different cancers (such as pancreatic, ovarian, prostate, lung, colon, etc.).

What next?

At that point, the physician can recommend a "best-next-step" confirmatory test such as a liquid biopsy (e.g., pancreatic) or medical imaging (e.g., lung). In this way, we are building an automated, actionable early cancer detection layer for electronic health records. 

Using the AI screen as a first-pass high recall funnel, people without any symptoms can also be caught. These patients would normally not get screened and be diagnosed at stages III and IV, when outcomes are poor. Our approach enables these patients to get confirmatory tests and necessary treatment if required when it's still early enough for curative treatment, leading to significantly higher chances of survival. This approach has the potential to massively scale and expand to detect other cancers that also lack effective screening, addressing a $100B+ market for cancer screening in the US.