There are huge barriers to getting a healthcare AI product to market in the United States. First off, data is generally extremely difficult and expensive to ga

There are huge barriers to getting a healthcare AI product to market in the United States. First off, data is generally extremely difficult and expensive to gather due to the need to protect patient info according to the Health Insurance Portability and Accountability Act (HIPAA). To get representative data, you need to gather from many sites and the data needs to be anonymized and "truthed" by a medical expert who typically must examine multiple medical reports and case history for each patient. Each data collection site requires approval for data collection by an Institutional Review Board, which may take months or even years to get. Assuming you are able to get a reasonable amount of data for development and testing, you need to solve the actual AI problem (probably very difficult due to the extremely heterogeneous nature of disease you are dealing with) and then you need regulatory approval. Depending on the FDA classification of your product, this can be another many-month to many-year process of data collection and clinical studies.