Kazem Sadegh-Zadeh     Philosophy of Medicine     HAPM




Medical Praxiology

Medical praxiology is defined to be the theory of medical practice, i.e., the philosophy, methodology, and logic of medical doing and acting ('praxis'). It comprises a theory of the patient, a theory of the physician, and a theory of clinical practice, i.e., clinical decision-making.

The patient is conceived as a bio-psycho-social and moral agent. This is explicated by presenting a theory of the organism. The organism turns out a fuzzy-causal system. The psyche is interpreted as a dynamic endomorphosis of this system, i.e., as a dynamic mapping of the totality of the states of the organism to a subsystem of itself. The current neurocentrism is rejected because the psyche is not a product, property, or faculty of the brain, but a systemic property of the entire organism in interaction with its social environment. As a social agent, a human being is subject to the consequences of her interactions. Through the perceptual system of the social agent, the interactions affect her emotional, cognitive, endocrine, and immune subsystems. Therefore, the concept of sociosomatics is suggested as a means of understanding the genesis of the so-called psychosomatic disorders and diseases.

Theories of health, illness, and disease are parts of the theory of the patient. A fuzzy-logical framework has been provided for these three notions which demonstrates their vague character. In this framework, health, illness, and disease turn out to be three completely different and independent categories. Health is not a complement of illness. Varieties of illness, characterized as ill, very ill, and so on, are different states of health. Nor is health a complement of disease. It is the complement of malady, which contains disease as one of its subcategories. The concept of disease denoting this subcategory has been explicated by our prototype resemblance theory of disease according to which the category emerges from a few socially constructed prototype diseases and a similarity relation between other human conditions and these prototypes. As a result of fuzzifying all three concepts, classical logic loses its capacity to be an appropriate logic of medicine. Particularly, the concept of fuzzy disease has consequences for clinical practice.

The notions of symptomatology, nosology, pathology, and nosological system are analyzed and the concepts of symptom, predictive value, sensitivity, specificity, and pathognomonicity of symptoms are explained. It is argued that all nosological systems are artifacts. In addition, an abstract geometry of diseases in the unit hypercube is introduced that enables the determination of similarity and dissimilarity, distance and proximity between different diseases and also individual disease states as particulars.

Causes of pathological states and diseases are viewed as additional information about a patient’s suffering which may enhance diagnostic, therapeutic, and preventive decision-making. To this end, a theory of etiology is presented that differentiates between deterministic, probabilistic, and fuzzy etiology. Particular attention is paid to probabilistic etiology, as the concept of cause is best conceived as a particular probabilistic relation between events. Specifically, we have chosen the familiar, well-defined concept of stochastic or probabilistic independence as our point of departure and have interpreted causation as a ternary relation of probabilistic dependence between an event A and an event B in a background context X consisting of one or more partial events, provided there is no other event C that renders this dependency relation between A and B spurious. Our ternary concept of causation also explicitly articulates the contextuality of causes in terms of the relationships between the cause event A and the context X. A complex system involving a number of events and a relationship of causation between them is called a causal structure. A causal structure is simply an extension of a usual probability space by including a period of time, a couple of events, and the information that there is a non-spurious, probabilistic dependency between these events. Because it is an extended probability space, the entire theory of probability is directly applicable to our theory of etiology. We have distinguished (i) positive and negative causation depending on whether an event A increases or decreases the probability of a later event B in a background context X, and (ii) many types of causes (potential, spurious, genuine, common, interactive, dominant, recessive, multifactorial, sufficient, necessary, conjectural). Our ternary relation of causation resolves many age-old, stubborn philosophical problems of causality. Moreover, it allows for a quantitative concept of causality that measures the degree of causal relevance of an event to its effect. With the aid of this numerical function, denoted cr, causes may be compared with each other regarding their causal impact. It also enables us to fuzzify causality by introducing a linguistic variable of causal relevance, CR.

After the preliminaries above, clinical decision-making is analyzed. The clinical encounter with the patient is conceived as a process of pathfinding between the initial patient data, A, and the final clinical decision, B. The space where the pathfinding occurs, is a dynamic, branching clinical questionnaire. This questionnaire has been conceived as a directed, acyclic graph that represents the procedural medical knowledge that the physician uses when decision-making. The nodes of the graph are the questions (examinations, tests) that the physician asks the patient. And an edge of the graph is constituted by the information that the physician obtains about the patient's health condition as an answer to a question (examination, test). It connects the question (examination, test) with the next one that is necessitated thereby. This process of knowledge-based action selection is logically analyzed. The analysis reveals that clinical judgment must be based on deontic logic because, as it is extensively demonstrated in the context and in later chapters, clinical knowledge in fact consists of ought-to-do rules. The syntax, semantics, and pragmatics of diagnosis is then analyzed. Precise concepts of differential diagnosis, indication, contra-indication, and differential indication are constructed. A theory of relativity of diagnosis is put forward according to which diagnosis is relative to a complex diagnostic structure, S, whose main components are the patient, patient data, the diagnostician, the underlying medical knowledge, and the reasoning methods used. By changing any of these parameters, the diagnosis may change. There is no 'true disease state' of the patient to uncover by the diagnosis. The supposed 'true disease state' of the patient is an artifactual construct of the structure S above. It is shown that clinical judgment is a computable process that could also be conducted by machines.

The concept, the logical structure, and the uncertainty of prognosis are analyzed. It is shown, that like diagnosis, prognosis is a speech act. Therapy and therapeutic decision-making are studied more extensively. Different types of decision-making are distinguished, and it is shown that in medicine expected value therapeutic decision-making is recommendable. Therapeutic efficacy research (randomized controlled clinical trial: RCCT) is critically examined, and the causal structure of RCCTs is reconstructed making them amenable to the philosophy of causation and causality. It is argued that RCCTs are not natural-scientific ("pharmacological") experiments, but practical-scientific investigations into the efficacy of human actions. Also the placebo and nocebo effects are logically analyzed. They turn out effects of human actions mentioned above. Considering the unavoidable placebo effect that accompanies every therapeutic action, a concept of net therapeutic efficiency is introduced.

The concepts of prevention, risk, risk factor, and risk ratio are the subject of the final section of Part II. Prevention is demonstrated to be a goal-driven practice of negative causation by human action. Probability-based and possibility-based concepts are distinguished. Improved concepts of risk and risk factor are put forward which make their causal nature and structure apparent.

This completes Part II that consists of 3 chapters and 279 pages.