Gallery
Shahram Tahmasebian defended his Thesis titled “Designing and implementation of Case-Based Reasoning system using electronic patient data, Case Mining and Case Similarity Algorithms”
Case-based reasoning is one of the artificial intelligence techniques that are used to implement decision support systems in the medical field. In this study, attempts are first made to comprehensively review case-based reasoning systems ever implemented and used in the medical field in order to identify their strengths and weaknesses and then propose a technical infrastructure, including case storage structures, distributed systems architecture, optimized fuzzy algorithms and implementation on mobile devices. Finally, the proposed structure was evaluated according to the data obtained from chronic kidney disease patients.
Firstly, expert meetings were held to investigate the current state of the case-based inference systems in the health field. For maximum efficiency, some modifications needed to be made in the structure of these systems and the way they are implemented. These modifications include: modification of the database structure, modification of the case storage and retrieval techniques in the database. More flexible inference algorithms that function based on operating systems proved to be more suitable. Discharge summary document is used as the basis for storing and retrieving the data. However, this document has some basic weaknesses and attempts were made to eliminate them based on a Semantic Web structure and data management software of this document. First, a Semantic Web-based structure was proposed for the unit summary document, then a unit summary extraction software based on the Java programming language and a document-oriented database were proposed to develop the unit summary. In addition, a software based on the Android operating system was designed and implemented to facilitate access to the document and simplify its development and management. The structure of the Semantic Web-based discharge summary and the discharge summary extraction software were evaluated by designing a questionnaire. A distributed service-oriented architecture was designed for the case-based inference system. A fuzzy case-based reasoning algorithm was designed. This algorithm is designed based on the importance of parameters and their weight. Data mining technique was used on the patients’ data to determine the weight and importance of the parameters. The proposed algorithm based on the Android operating system software was implemented and evaluated.
A semantic web-based unit summary structure was defined. Moreover, a software was developed to extract unit summaries on the basis of Java programming language. The features of this software include, utilization of the MongoDB database that is a document-oriented database. The structure of the unit summary and the designed software was evaluated using a questionnaire. The proposed structure for the unit summary has several advantages. The previous studies on the patients’ unit summary did not focus on the structure of them and no specific data structure was presented for them.
A service-oriented architecture for the case-based inference systems led to resolution of many of the problems in the previous case-based inference systems implemented in the health field has designed. This architecture was also used to design an android-based software to extract patients’ unit summaries. Then, the designed software and architectural design were evaluated using a questionnaire. The software and designed architectural were evaluated using a questionnaire.
Data mining process was conducted on the data obtained from patients with chronic renal failure. After this process, important parameters and their weight were extracted. Using the extracted parameters and their weight, the fuzzy algorithms designed for searching the case-based inference systems will be strengthened. Then the algorithm was implemented on the patients’ unit summary and compared to other search techniques.
Based on the semantic web structure standard, the layers existing in this structure will be added to the unit summary of metadata patient. This process will have the following advantages:
1. Development of fully structured unit summary of patients
2. Simpler implementation of processes such as data mining
3. Simpler transition of patient data through unit summaries
4. Utilization of all the semantic web processing facilities and technologies for data processing purposes
Moreover, the designed architecture for knowledge management in organizations was an important and effectively step. Knowledge management includes a set of strategies and techniques used to capture, create, represent, distribute and match insights within organizations. The Action Plan for scientific reserves management is based on two basic components: processes that produce scientific reserves and tools and methods that facilitate access to these reserves.
Many Knowledge management systems don’t have any suitable architecture, and this makes it very time-consuming and expensive to update their databases. Therefore, these systems are only used for a short period of time. The update process of the system presented in the present study was much more cost-effective and less time-consuming, therefore it could be used by users for a longer period of time. The fuzzy algorithm presented in the present study, is flexible according to the time slot considered for each variable and its normal range. Therefore, after each run, we can strengthened the system and increase the search accuracy. On the other hand, due to the fact that the proposed algorithm identifies the weight and importance of each variable using data mining techniques and applies them on the fuzzy search algorithm, its accuracy is much higher than other algorithms.
Supervisors: M. Langarizadeh, , Assistant Professor
M. Ghazi-Saeedi, Assistant Professor
Advisors: M. Mahdavi-Mazdeh, Professor
Reviewers: R. Safdari, Professor
Sh. Rostam niakan, Assistant Professor
F. Sadooghi, Professor
H. Rahimzadeh, , Assistant Professor