Medical Software Architecture

What kinds of architectural patterns are needed?

Manufacturers with a business interest in machine learning understand the architectural challenges of building and maintaining the distributed systems needed to support such systems.

A number of architectural and design patterns of interest

• Layers is the starting point for most scalable, distributed architectures

• Separated Presentation patterns (MVC, MVVM, MVU) have become increasingly important as front ends have become more diverse. 

  • Uni-directional data flow (MVU) is particularly amenable to a prolog implementation.  See the example using SWI prolog.

• Blackboard can be useful for a lot of knowledge-based systems

• Actors can be used for a lot of interesting things, and one of them is federated learning

• Ball of Mud is much maligned, but most systems start out this way.