Advanced Curricular Design and Educational Technologies (ACDET) provide DDX training based upon the seven previously described cognitive principles. We have developed an artificial intelligence agent/tool (called KBIT - Knowledge Based Inference Tool) that utilizes the knowledge base of clinical experts to guide training in the DDX of common and important clinical problems. Key features of KBIT's approach to DDX training include:
A. Over 25 years of bench and applied research into the cognitive factors underlying the development of DDX competence.
B. Supports development of DDX competence one-problem-at-a-time.
C. Use of an artificial intelligence (AI) system to capture the declarative knowledge base of clinical experts with respect to the common and important disease differentials associated with each problem, and the key features (signs, symptoms, lab findings, radiological findings, etc) that characterize and distinguish each disease differential
D. Use of a second AI system to translate each expert's declarative knowledge base into advanced DDX training. This training consists (in part) of a description of the problem's disease differentials, each disease differential's prototypical signs and symptoms, the identification of 'key competitors' for each disease differential, and, a description of the signs and symptoms that can be used to discriminate one disease from another.
E. Fully supports the role of 'Deliberate Practice' in the development of DDX competence via the generation and availability of a large number of practice cases.
F. Provides immediate, individually tailored, formative feedback regarding how to correct any diagnostic error associated with any given practice case.
G. Use of AI techniques that support students develop procedural knowledge involving two key pattern recognition capabilities (pattern matching and pattern discrimination)
Currently, KBIT contains over 30 problem-specific modules, which encompass more than 225 of the most common and important disease differentials encountered during medical training. Many more modules are currently under construction and will be made available to subscribers as they are completed (see current list of problem modules and disease differentials on home page). This material is accessible via the internet, so training is available anywhere, and at any time.
KBIT supports learning DDX within the context of problem-specific training modules (by definition, a 'problem' is any patient complaint or abnormal finding (lab, radiologic, etc). For each problem module, KBIT provides two fundamental declarative knowledge components for that problem. These consist of: 1) a list of common and important disease differentials that could cause the problem, and 2) a list prototypical signs and symptoms for each of these disease differentials. KBIT also provides on average, one case example demonstrating how each disease might be portrayed in terms of a typical case vignette.
KBIT then provides numerous opportunities to practice DDX and thereby develop procedural knowledge involving pattern recognition. That is, training in how to utilize the learners inherent pattern matching and pattern discrimination capabilities as the basis for recognizing the disease pattern hidden within each case vignette. The more practice cases attempted, the better your declarative knowledge consisting of the 'what' about each disease differential. Also, with practice your memory of those disease patterns will become more enduring. Finally, the more practice at, and training in, pattern matching and pattern discrimination you receive, the more refined those procedural skills will be (knowing 'how') when confronted with new problems and diseases with which you have had little or even no prior experience or practice.
It is also important to note that these practice cases vary such that prototypical cases are experienced first, followed by the slow introduction of intermediate and, eventually, less typical case presentations. Thus, as case-based experience accumulates, KBIT progressively challenges the learner with less typical (more difficult) cases. The student's knowledge of disease patterns is expanded until it becomes easier to diagnose atypical case presentations.
With every practice case, KBIT reinforces your evolving declarative knowledge of both disease patterns, and, procedural knowledge capabilities (pattern recognition capabilities). When a correct diagnosis is made, KBIT notes the prototypical signs and symptoms associated with the disease just diagnosed and highlights within the listed disease prototype, those S/S that are present in the case at hand.
When an incorrect diagnosis occurs, KBIT provides immediate, individually tailored, corrective feedback designed to both refine your evolving declarative knowledge base of disease patterns, and, to improve your pattern matching and pattern discrimination capabilities. During this 'corrective feedback' mode, KBIT compares the features of the case in terms of both the misdiagnosis made by the learner (the incorrect diagnosis) and the diagnosis that should have been made. This feedback focuses on signs and symptoms in the case that should have enabled you to match or rule in (R/I) the proper diagnosis, as well as signs and symptoms that should have been used to discriminate or rule out (R/O) the incorrect diagnosis.
ACDET's ongoing research with medical students strongly suggests that it is this immediate, individually tailored, corrective feedback that expedites the transformation of naive and novice students into diagnostically competent students. Such training is not possible in a large classroom because a faculty member in front of a large group usually does not know the mistakes of individual students. Even if all the students' diagnostic errors were known for a given case, the faculty member certainly would not have time to provide a 'likely explanation' for each particular incorrect diagnosis in a large group setting. Similarly, this level of individually tailored feedback is generally not available even in small-group, problem-solving settings, because the large number of faculty required results in a low likelihood that all students will have access to an expert in the problem at hand in their particular small group session.
Because KBIT training is derived from an expert specialist in the problem at hand, students receive instruction that closely approximates the personal guidance only an expert can provide. Research has shown that students trained with KBIT are likely to outperform students trained by an expert using traditional instructional approaches. Simply put, KBIT provides unparalleled advantages in training to, and learning DDX.