Empowered Patient Podcast Advancing Synthetic Data for Paradigm Shift in Medical Research with Daniel Blumenthal MDClone
Daniel Blumenthal, VP of Strategy at MDClone, is building synthetic data models to mimic real patient data without the need to protect the identity of individuals. This type of data is being utilized in drug discovery and development, building predictive tools and new care models, and to better understand biases in clinical trials making them more representative. AI tools play an essential role in generating and validating synthetic data as well as integrating it into various processes and environments.
Daniel explains, "Synthetic data means different things to different people. So I definitely would want to start with sharing what it usually means to MDClone, which is creating data based on real populations of patients, real patient data. Learning from that real patient data, extracting essentially the statistics from that data, understanding the relevant patterns that exist inside that data, and then, based on the statistics alone, generating a brand-new set of data, a brand-new set of people, and the synthetic data that we generate."
"If you look at it, essentially with your naked eye, you wouldn't know it's not real people. It has the same structure, has the same content appearance, and yet, inside, synthetic data contains no real people. So very different from other approaches to protecting patient privacy, which take a patient data set and hash data, mask data, or remove information like IDs, names, or locations."
"Instead of taking that sort of approach, synthetic data actually generates a brand-new set of people, fake people, that in our case, as we build it, maintain the same statistical properties, such that any analysis you want to complete, you could accomplish using the synthetic dataset just as you would with the original dataset. That it will yield the same meaningful conclusions as the synthetic that it would with the original."
#MDClone #MedAI #AI #SyntheticData #DrugDevelopment #DrugDiscovery #ClinicalTrials #PatientPrivacy