Diagnosis as Fiber Reduction

This repository contains the Python code used to generate the illustrative simulations accompanying the manuscript:

Diagnosis as Fiber Reduction: A Formal Theory of Clinical Reasoning

The simulations are not intended as clinical decision-support tools and do not model real diseases. Their purpose is conceptual: to illustrate how diagnostic reasoning can be represented as the progressive reduction of an underdetermined possibility space (the diagnostic fiber) through the integration of heterogeneous forms of evidence.

Included simulations

• Diagnostic fiber reduction
• Narrative and contextual evidence integration
• Diagnostic bias as premature fiber restriction
• Diagnostic gain of candidate observations
• Adaptive versus random diagnostic inquiry

Main concepts

For a biomedical representation d, the diagnostic fiber is:

F(d) = {z ∈ H : πD(z) = d}

Additional forms of evidence reduce the fiber:

F(d,e) ⊆ F(d)

F(d,e,s) ⊆ F(d,e)

where e denotes experiential information and s denotes social/contextual information.

Repository structure

• code/diagnosis_fiber_reduction.py
• README.md
• requirements.txt
• LICENSE

Citation

Galli C. Diagnosis as Fiber Reduction: A Formal Theory of Clinical Reasoning.