For that use case, my thought now is twofold. First, maybe they can do an open algorithm, meaning that they contribute their machine-learning algorithm, but for the data owners, the data stewards like hospitals and clinics to run that, and they share only the insight, not the raw data.