Medical Software Architecture

What kinds of architectural patterns are needed?

Vendors 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. *** SEE Uno.Prism *** for some fun things happening...

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

• Blackboard is an under-rated pattern

• 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 this way.

Prototypes

• E2EImage Review

• IsocurveMvvm

• PainttoolMvvm: Distributing as a signed click-once package at: myapp @ d4r4k.net

• FsBlackboard

• SWI Prolog MVU Tic-Tac-Toe example

• Actor Model Clue game