This is basically an effort that started a couple of years ago. We were looking into ways that could really facilitate data sharing between multiple stakeholders. And we’re looking specifically at real-world data. So the reason for that is that given the recent acceleration in the approvals for drugs and oncology, we really have like this vicious cycle of unmet needs and exponential increase in the demand for data...
This is basically an effort that started a couple of years ago. We were looking into ways that could really facilitate data sharing between multiple stakeholders. And we’re looking specifically at real-world data. So the reason for that is that given the recent acceleration in the approvals for drugs and oncology, we really have like this vicious cycle of unmet needs and exponential increase in the demand for data. And while clinical trials are great because they are the gold standard, they really require a lot of resources that oftentimes we do not have. So our aim was to try to get real-world data, but to try to bypass some of the regulatory challenges that come with these. And this is how we wanted to implement artificial intelligence models to try and take the real-world data and derive from it a synthetic version that replicates the same statistical patterns without compromising patients’ privacy. And this is exactly what we show. We show that a specific AI model that we call classification and regression trees is really able to do just that. And then we offer a framework of how this model can be integrated in a sort of a platform where this data could be continuously ingested, processed by AI, and then used to derive these synthetic data sets. Using these synthetic data sets, we are able to model the clinical outcomes for patients in the real world, and we can use that to predict clinical outcomes for a specific real patient that comes into our clinic.
This transcript is AI-generated. While we strive for accuracy, please verify this copy with the video.
