We focus on the language of rules because the knowledge and the rule is not only the if-then part of it but it's actually what's inside both the left-hand side and the right-hand side. So, let me show a little bit what I'm talking about. We've discussed often this notion of intention and extension. So, when I tell you about the transcutaneous bilirubin I'm having in my mind a concept of transcutaneous bilirubin. We know it has to be encoded somehow in the EHR. I talked about the 24 different choices one has. That linking between the concept and the actual fields and values in the EHR, that's called binding, what often binds an abstraction to a specific instance. In the world of knowledge bases, rule bases and decision support, sometimes we talk about this intermediate level called the generalized EHR or virtual EHR. You'll notice I put them in curly braces, the reason is that the linking of that intentional concept to the specific was implemented in a decision support concept called Arden syntax and their instruction for linking those two together included putting inside curly braces the specifics of that particular installation. So, I wanted to move a rule from one place to another. Almost all the rule could be moved except for the part that's inside the curly braces. So, this problem of linking the intention to the specifics is sometimes called the curly brace problem. So, to give flesh to this let's use a different example that I've used so far. So, here's one. So, if isotretinoin po orally is ordered and the patient is pregnant, then alert absolute contraindication. The reason, which is not part of the rule, the reason is simply that a woman who is in the first trimester who takes this vitamin A derivative can actually cause a birth defect. Teratogenic is the way in medicine we say causing birth defects. So, for isotretinoin is a concept, administered orally is a concept, patient is pregnant is a concept. Alert absolute cantraindication is, remember what we've been calling, a local business rule or business concept, but certainly the left-hand side isotretinoin and patient and pregnant are concepts that need to be instantiated. So, what an order looks like, in case you haven't seen it, and you can see the patient ID and the actual instruction. The challenge is to link this patient's ID to the concept of the patient in the rule where it said, if the patient is pregnant. So I need to know what patient we're talking about, so I am binding the concept of patient to using this ID. Similarly, we can bind the drug isotretinoin and bind the specific order to whatever ID this order has. So, the question is, the doctor has written isotretinoin let's assume as a string of characters, rather than a pull down, a string of characters. So, now the machine has to link that string of characters isotretinoin etc to this more generic concept of the isotretinoin drug. Now, in this case it sounds obvious but if this doctor was in Brazil and ordering in Portuguese or if the choice was a trade name for the medication, Portuguese or trade name, they are still referring to the same drug which is conceptually isotretinoin. So, you still need to bind the concept of isotretinoin to this more local instantiation of that concept. So, we talked about instances being linked to concepts, or an instance is an instance of a more general concept. Concepts themselves can have relationships to each other. So, isotretinoin is a drug that treats acne, and then these concepts can have properties, trade names, generic names, their status. So, let's see some examples of what I am talking about. But these classes of concepts are what we'll be talking about the rest of this module. I hope you realize what I just did. I just talked about these concepts as instantiations of more generic concepts. So, it actually goes on without end. The core question for the rest of this module is going to be how should we organize concepts to be used by knowledge-based systems? And what I'm listing there are different ways that we do organize the concepts but it's in the language where our knowledge lives. So, that's why this is called knowledge for use and not just knowledge discovery or other sorts of knowledge.
