262155
Beschreibung
Karteikarten von Michael Riben, aktualisiert more than 1 year ago
|
|
Erstellt von Michael Riben
vor etwa 11 Jahre
|
|
| Frage | Antworten |
| how is knowledge acquired? | narrow down from a broad subject to a specific topic with extraction of data |
| What are the 4 approaches to knowledge representation? | Clinical algorithms, bayesian statistitcs, production rules, and scoring/heuristics |
| What are benefits of clinical algorithms? | Knowledge is explicit, knowledge is easy to encode, uses a flow chart with information nodes and decision nodes |
| What are limitations of Clinical Algorithms | No accounting for prior results, no ability to pursue new etiologies, treatments, etc, and new knowledge is difficult to generate |
| What are Bayesian statistics? | based on bayes theorem, which calculates the probability based on prior probability and new informaiton |
| Give an example of Bayes Theorem CDS | Leeds Abdominal Pain System in 1970's, performed better than physicians |
| What are limitations of Bayesian Statistics? | findings are not conditionally independent, diseases may not be mutually exclusive, and multiple findings result in high computational complexity |
| What are production rules? | Knowledge encoded in If-Then rules to arrive at a diagnosis |
| What are the two types of Production rules systems | Forward Chaining and Backward chaining. |
| What is a backward chaining system? | systems pursues goals and asks questions to reach goal |
| what is forward chaining system? | similar to clinical algorithm with a computer following a prescribed path to reach an answer |
| what was the first rule based expert system in medicine? | Mycin |
| What was Mycin used for? | suggests treatments for infectious diseases in meningitis and bacteremia using a backward chaining approach by asking questions relentless to reach a diagnosis |
| What were the limitations of Rule Based Systems? | depth first searching could lead to focus in wrong area, large and difficult to maintain rule knowledge base, slow and time consuming to use the system |
| What is a scoring/heuristic knowlege representation? | knowledge is represented as "profiles" of findings that occur with measures of importance and frequency for each finding, most scalable |
| give 3 examples of heuristic knowlege representation expert systems? | Iliad, Dxplain, and QMR |
| what is the import of finding in QMR? | how important a measure of a finding is to explain |
| What is the Evoking Strength in QMR? | The likelihood of a disease given a finding = scored 0-non-specific to 5-pathognomonic |
| What is the Frequency score in QMR? | the likelihood of a finding given a disease 0-occurs rarely to 5(occurs in all cases) |
| What were the limitations of QMR? | Long learning curve, time consuming data entry, diagnosis is not a major issue for clinicians, incomplete knowledge bases, |
| What realization occurred in CDS in early 90's? | diagnostic process is too complex for computers, greek oracle model was inappropriate model for medical usefulness |
Möchten Sie mit GoConqr kostenlos Ihre eigenen Karteikarten erstellen? Mehr erfahren.