| "At their core, guidelines are decision trees" That's wishful and perhaps not even helpful as a goal. Guidelines rarely have the data to cover all possible legs of decisions. They report on well-supported findings, offer expert opinions on some interpolated cases, and perhaps list factors to consider for some of the remainder. If you reduced this to a decision tree, you'd find many branches are not covered, and most experts could identify factors that should lead to a more complex tree. The reason is that branches are rarely definitive. It's more like quantum probabilities: you have to hold them all at once, and only when treatment works or doesn't does the disease (here cancer) declare itself as such. Until the true information architecture of guidelines is captured, they will be conveyed as authoritative and educational statements of the standard of care. In almost all cases, it's more important to reduce latency and increase transparency (i.e, publish faster but with references) than to simplify or operationalize in order to improve uptake. Most doctors in dynamic fields don't need the simplification; they rely on life-long self-discipline and diligence to overcome difficulty in the material, and use guidelines at most as a framework for communication and completion, i.e., for knowing when they're addressed known concerns. Structured guidelines mainly enable outsiders to observe and control in ways that are likely to be unproductive. |