<p> Medical diagnosis is a complex task, that requires integrating several sources and
types of information: a patient's description of their symptoms, lab results, and perhaps
even 'gut feelings' regarding potential diagnoses. From a cognitive psychology
perspective, diagnosis is a type of categorization and as such has been typically divided
into processes that are deliberate, rule oriented, and available to conscious control (often
called analytic processing) and processes that are rapid, outside of conscious awareness,
and typically based on similarity (often referred to as non-analytic processing).</p> <p> Traditionally, similarity has referred to whole-case similarity between a current and previously encountered case. However, this pattern matching to an entire previous case does not differentiate between diagnostic and non-diagnostic information, a distinction that is made clear in the rules taught to medical professionals. Since medicine does rely extensively on diagnostic rules, the research presented in this thesis will examine the effect of similarity of features relevant to the application of a diagnostic rule as well as the effect of similarity from patient identity, which is mnemonically salient but diagnostically irrelevant.</p> <p> The work presented in this thesis specifically examines the role of similarity in the categorization, or diagnostic decisions of novices. Medical students start training with the best available rules, standard diagnostic rules likely to be used in future practice. In the experiments reported in this thesis, participants are trained to competence on diagnostic rules using prototypical written case vignettes of simplified psychiatric diagnoses. Participants then evaluate cases in which clinical information supports two possible diagnoses, but in which either diagnostic features or diagnostically irrelevant identity information is similar to those seen in training. The results of these experiments indicate a strong reliance on familiar of rule-relevant symptom descriptions (i.e. similarity within the application of a diagnostic rule), supporting an adaptive role of similarity within the application of an analytical decision rule. Further, the influence of familiar diagnostically irrelevant information (i.e. similarity within the context of patient identity) demonstrates the maintenance of non-diagnostic information within memory,
and the possibility of matching to a previous exemplar on rule-irrelevant features. Familiarity, whether diagnostically relevant or not, increases the probability that clinically relevant features are mentioned in support of a diagnosis, which may indicate the disambiguation of features following previous experience with that feature, and a strong influence of familiar but non-diagnostic information on the interpretation of features.</p> <p> This thesis supports a model of medical decision making in which there is an effect of similarity to previous instantiations of clinically relevant features. That is, similarity is a basic component of decision making that is not limited to matching on entire previous instances. Previous research has suggested that analytic and non-analytic reasoning are competing or fundamentally separate processes, whereas the demonstration of similarity within the application of a diagnostic rule suggests that not only is similarity an adaptive strategy for learners, but the differentiation between similarity based and rule based processes may be less clear than previously suggested.</p> / Thesis / Doctor of Philosophy (PhD)
| Identifer | oai:union.ndltd.org:mcmaster.ca/oai:macsphere.mcmaster.ca:11375/17737 |
| Date | 06 1900 |
| Creators | Young, Meredith E. |
| Contributors | Norman, Geoffrey R., Psychology |
| Source Sets | McMaster University |
| Language | en_US |
| Detected Language | English |
| Type | Thesis |
Page generated in 0.0025 seconds