Dynamic inference (usually causal) is the flow of thought where we use premises (statements with a probability of truth) to draw inferences and conclusions. Conclusions are utilized as premises in continued inferential processes and determine, but do not restrict, practice. To be more concrete, in clinical practice these inferences are usually causal. But even as causal inferences there are at least three types as we have already discussed: deductive, inductive and abductive.
A quick recap and in a causal framework, inductive infers general conclusions (premises about a causal association) from specific data; this is what emerges from research studies and systematic reviews. Hill’s criteria of causation, risks of bias, and sets of application help us weigh the applicability and strength of such general premises. Deductive inference estimates outcomes - future events - based on general premises (typically from inductive inference) and specific observations. Abductive inference is an attempt to infer effect from cause and does not have a well-established rule for justification (inverse problem).
Dynamic inference is simply a term for inference as used in clinical practice, which requires a dynamic flow of induction, deduction and abduction in processing information – interpretation – action. The influence of evidence for a given premise in the process is dependent on the justification of the premises and thus there are critical bottlenecks in any given clinical scenario to the application of evidence-based practice. But we know something about the general premise, we have some reason to believe the general premise and these reasons also have a hierarchy of justification. Don’t get me wrong this hierarchy is not separate from empirical observations. After all, we do exist in a world (ontology) and its reality (realism) gives rise methods what can be used to know it (epistemology), and those methods involve a a balance of observation and making models about the observations (empiricism balanced with rationalism).
At CSM in a few weeks I will be making a presentation on Dynamic Inference. The presentation will explain these bottlenecks using causal graph models with emphasis on the combinatorial explosion of cardiopulmonary cases as inherently complex.
I will then present two possible uses of the concept of dynamic inference. First the presentation will encourage thinking about reasoning in complex cases to leverage the use of heuristics. Heuristics reduce information demand in complex cases. A low probability of error necessitates the proper application of heuristics. Thus, implications for evidence interpretation to application should be considered in going from evidence to practice guidelines. Second the presentation hopes to demonstrate how the web of causal interactions and number of abductive inferences can be used to quantify complexity of a clinical case and with increasing complexity of a case there is an exponential rise in the probability of error.
Each of the uses of applying an analysis of the dynamic inference of clinical reasoning are independently important to members. First related to clinical practice recommendations and second related to consideration of the complexity of clinical cases. In addition, this approach demonstrates that cardiopulmonary cases are intrinsically complex and this influences the types of evidence required and the process of taking evidence to practice recommendations. So there is an interactive effect between the evidence interpretation to application and the clinical complexity components of dynamic inference in general, and cardiopulmonary PT in particular.
Once I have the presentation completed I will be sure to post it with a link from the blog. The abstract (used in creating the above post) was submitted in July 2014, about 6 months before I started the blog but is consistent with what I am trying to develop overall with a knowledge based practice system of thought.