Critical Thinking and Ontology

Content: Research methods lecture by Dr Greenes introduced us to understanding the definition of research and the steps involved in a successful research idea. To summarize my understanding, the first step is understanding the reality model or the process which exists. Then identify the possible loopholes and deficiencies in the process and build models to address the same. These models which are based on reality could be probabilistic prediction, rule based reasoning or an algorithmic calculation.
The other method to approach a problem is to manipulate reality by building simulation models. These are models which are as close to reality as possible and then study how well they work.
The third and final method introduced was empirical studies. This is the hypothesis based approach where we look at real life scenarios and build hypothesis to describe and understand these phenomena or build interventions to study if these scenarios can be modified. The final step to these hypothesis/intervention models is evaluation of these models by applying these models in real life situations.
BMI poses special challenges as it covers more than one domain and Dr Greenes mentioned the importance of identifying the research domain early. Since this is a coming together of various fields, it is important for a BMI researcher to understand the limits and range of each domain. This will help in defining our own research area and research question in a scientific manner.
I would suggest to all that they visit the website http://www.criticalthinking.org/ which introduces researchers to the methodical approach to research idea development.
The second lecture on ontology was almost an eyeopener for me. I thought that ontology was just another fancy name for etymology (history of how a word came into existence). However ontology has deeply philosophical as well as overtly scientific meanings. Leaving the philosophical aspect aside, ontology is simply the relationship of various concepts within a domain (Wikipedia). Dr Fridsma explained this concept very simplistically using the "wine/winery" example.
The lecture explained the difference between taxonomy and ontology wherein taxonomy was a simple classification without a well defined relationship between the hierarchies. On the other hand ontology explicitly defines the relationship between every member of the ontology domain.
There was a very good question posed in the class which asked us to think about the difference between database schema and knowledge based system, ontology being a part of both. Database schema (as introduced in EVM class) was obviously describing about the relationship between the tables and content within the database. Knowledge based system store data in a manner such that there is a logical deduction which can be applied to all elements of the stored data (Wikipedia). Ontology gives a structure to the terms stored in this knowledge based form.
The lecture introduced us to controlled vocabulary system like National Drug Codes, taxonomy like SNOWMED and ICD-9 which is more organized is a hierarchical manner and the advantages and disadvantages of these systems.
Understanding the components of an ontology based schema; instance, class, attribute and relation was also stressed.
Obviously storing medical terms in manner that allows structure and links between these terms is very important for easy retrieval of the same. Just having a data dictionary of terms without having any relationship defined leaves the terms open to varied interpretations. This is the reason for current confusions arising from ICD codes and DRG codes used for different purposes. The current research and brain storming on development of an ontology system which can be standardized and acceptable stems from the need to reduce this confusion.  
Unified Medical Language System is a National Library of Medicine initiative which is like a knowledge based system. Meta thesaurus uses the vocabulary from all major systems like SNOWMED, ICD-9, 10, etc. Semantic network tries to explain the relationship between the terms in both hierarchical and non hierarchical manner. Specialist lexicon uses both common English language usage and medical words to facilitate natural language processors. MetamorphoSys allows the user to download UMLS on your system. A very good resource for this information is the NLM website.
http://www.nlm.nih.gov/research/umls/about_umls.html#Specialist   
A lot of the concepts introduced in this lecture had good references in Wikipedia. 

Posted by: Sheetal Shetty