Srinagesh Satyanarayana

Ph.D

Kaya17

Chief Technology officer

PLCO (Learn more about this study)

PLCO-1000

Jul 1, 2022

Developing a multi-marker diagnostic test with predictive model enhancements for early-stage (Stage-1) Ovarian cancer

Kaya17 has developed a novel diagnostic platform for point of care testing. The test platform features high sensitivity and specificity with very a very low Limit of detection. Such high performance has been possible with the use of novel cartridge technology, Assay methods, and a high sensitivity reader. Armed with such performance improvements, early diagnosis of several diseases is possible. We have demonstrated the early detection abilities in covid testing.

With some modifications we have adapted our test platform and methods to protein marker detection. Early detection with certain cervical and ovarian cancer markers has shown very promising results. We plan to apply our method and platform to a variety of markers to develop a multi-marker ovarian cancer test. We also plan to develop a method to determine the stage of the cancer based on a combination of marker results and prediction models results.

One of the major challenges in ovarian cancer diagnosis is that Stage I diagnosis has been very difficult due to the lack of symptoms and lack of marker specificity. Our high sensitivity test can make testing for Stage I diagnosis possible. Machine learning based software algorithms have shown that predictive accuracy can be improved using models extracted from clinical trial data of previously diagnosed patients. Such models can be applied to both quantitative output vectors from multi-marker tests and non-molecular patient data like pre-existing conditions and non-specific symptoms.

Data from the PLCO study can help us with testing our predictive models – both on the marker data and patient symptoms and conditions. The PLCO study can also help us in selecting the right markers.

Our primary aim is to develop a multi-marker test for ovarian cancer. In the case of ovarian cancer, each individual marker has only a limited predictive capability and no single marker has a 100% predictive ability. Hence a predictive model based on a marker vector resulting from a panel of tests would be much more definitive. Marker efficacy is lower as the we move to earlier stages of Ovarian cancer. However, our test platform has very high sensitivity and the ability to quantitate. We plan to explore if we can take our diagnostic prediction to stage-1 of Ovarian cancer.

While marker-based diagnostics has been effective in diagnosis of ovarian cancer, it has been show that using non molecular methods like a software algorithm implementing a predictive model based on patient parameters, can increase the prediction accuracy. We plan to add this layer of prediction to our diagnostic system.