AWERProcedia Information Technology & Computer Science Vol 04 (2013) 1074-1078
3rd World Conference on Innovation and Computer Sciences 2013
Designing an expert system to screen for autism: Investigating psychometric properties Mohsen Shokoohi-Yekta *, Associate professor, University of Tehran, Chamran Hwy, Tehran, Iran Maryam Mahmoudi, Ph.D student, Allameh Tabatabaei University, Tehran, Iran. Bagher Ghobari-Bonab, Associate professor, University of Tehran, Chamran Hwy, Tehran, Iran. Saeed Akbari Zardkhaneh, Ph.D student, Allameh Tabatabaei University, Tehran, Iran. Salahadin Lotfi, Supervisor of Neurofeedback and Cognitive Training Section, Atieh NeuroscienceBased Center, Tehran, Iran. Suggested Citation: Yekta S., M., Mahmoudi M., Bonab G., B., Zardkhaneh A., S. & Lotfi S. Designing an expert system to screen for autism: Investigating psychometric properties. AWERProcedia Information Technology & Computer Science. [Online]. 2013, 04, pp 1074-1078 Available from: www.awer-center.org/pitcs Received December 23, 2012; revised February 10, 2013; accepted March 05, 2013 Selection and peer review under responsibility of Prof. Dr. Fahrettin Sadıkoglu, Near East University. ©2013 Academic World Education & Research Center. All rights reserved. Abstract On the condition of educational modernization and joining the Republic of Kazakhstan to the Bologna process, A. Yasawi International Kazakh-Turkish University points out as one of its main problem the usage of information technologies in highly qualified training of the specialists of international level in the sphere of ecology and increasing the quality of their professional training, because the use of information technologies emphasizes practical-oriented education, its professional, pragmatic, subject-professional aspect. Including some elements of well known approaches in pedagogical sciences such as systematic, personal-oriented and synergetic and others the significance of competence-based approach as one of necessary base of formation of communicative competence and professional training of future ecologists is stated out. Because it has humanistic, professional, pragmatic and practical features showing its integration and systematic advantages. Accepted as one of new scientific approach, the use of information technologies of is one of the most effective means of developing methodological and creative thinking activity, professional and cultural norms of behavior, correction of oriented values on Keywords: Architectural design education, generative design , computer aided design, typological data, algoritmic design;
*
ADDRESS FOR CORRESPONDENCE: Mohsen Shokoohi-Yekta, Faculty of Psychology, University of Tehran, Chamran Hwy, Tehran, Iran, E-mail address:
[email protected] / Tel.: +98-21-6111-7452
Yekta S., M., Mahmoudi M., Bonab G., B., Zardkhaneh A., S. & Lotfi S. Designing an expert system to screen for autism: Investigating psychometric properties. AWERProcedia Information Technology & Computer Science. [Online]. 2013, 04, pp 1074-1078. Available from: www.awer-center.org/pitcs
1. Introduction Autism spectrum disorders are a group of neurodevelopmental disorders marked by impairments in communication, social interactions, limited interests and repetitive behaviors (Duffy & Healy, 2011). Findings suggest increasing prevalence of autism throughout the world (Prior, 2003). Many countries such as Japan, Sweden, England and America, reported the prevalence of autism to be even more 1 percent (Baron-Cohen et al., 2009). In Iran, Prevalence of autism by screening children aging 5 has been 0.24 percent using Social Communication Questionnaire (Samadi, Mahmoodizadeh & Mcconkey, 2011). Because there is no biological marker or test for identifying autism, diagnosis of the disorder rely on behavioral assessments based on observations and historical reports collected from several sources. This process is complex and between the first time parents become concern about problems in their children and the time autism being diagnosed, can take up to about 4 years (Howlin & Asgharian, 1999). However, in order to have the best therapeutic outcome, children have to be diagnosed before six years of age (Filipik et al., 2000). Due to the limitations of the diagnostic tools such as taking time for their implementation, it is irrational and unfeasible that all children being fully diagnosed for autism disorders. Therefore, at first it is better that all children be identified by screening process and then refer for more evaluations (Norris & Lecavalier, 2010). In Iran, Social Communication Questionnaire is used for screening autism in children up to 5 years. Internal consistency for this questionnaire was reported about 0.81 (Sasanfar & Ghadami, 2006 site in Samadi et al., 2011). Minimum age for diagnosis of autism formally is about 4 years old in Iran. Consequently, a tool is needed to cover lower ages and the aim of the current study was to develop a tool that helps early diagnosis of autism in children aging 2 to 6 years old. 2. Method 2.1. Design and sample group. This research is among research and development studies and data collection methods is a descriptive-survey one. The study population included all of the mothers of children with autistic disorder aging 2 to 6 years old in Tehran. Sample groups included 53 mothers of children with autism, 49 mothers of normal children and 20 mothers of children with Down syndrome. Sampling method was an available sampling for all three groups. It is necessary to indicate that sample demographic properties and stages of developing the tool, was presented in the article titled ‘Online expert system: An items analysis’. 2.2. Procedure. The aim of the current study was to investigate the psychometric properties of subscales and categories designed to be used in the knowledge base of an expert system, thus we started with investigating a preliminary tool designed for this purpose. Autistic children were selected from the “Center for Treatment of Autistic Disorder” (CTAD), children with Down syndrome from special preschools and normal children from kindergartens throughout Tehran. In the analysis of subscales and categories, discriminant analysis was used in order to know which compound subscales and with which coefficients get together to discriminate between autistic group and the two other groups, as well as measuring internal consistency of each category computed by Cronbach’s alpha to show whether each category assesses exactly the right construct. The study tool is an expert system )a software program(, which can stimulate professionals reasoning and make decision through rules providing to it. The program includes two main parts such as knowledge base and engine motor. The knowledge base include the items that are the most important in autism diagnosis and the inference engine compounds of procedural rules that direct decision making process based on the information acquired. 1075
Yekta S., M., Mahmoudi M., Bonab G., B., Zardkhaneh A., S. & Lotfi S. Designing an expert system to screen for autism: Investigating psychometric properties. AWERProcedia Information Technology & Computer Science. [Online]. 2013, 04, pp 1074-1078. Available from: www.awer-center.org/pitcs
3. Results The tool being used in this study has four categories, which their internal consistencies were at the appropriate level. That is, items in each of the categories have high correlation among themselves. Findings related to internal consistency coefficients and correlation coefficients of the items with the total score of each category are sited in the article titled ‘Online expert system: an item analysis’. In order to find out whether this diagnostic test has appropriate diagnostic power to identify and distinguish normal, autistic and Down syndrome children from each other, stepwise discriminant analysis was used. Thus, discriminant analysis performed for each category at the level of subscales separately. Using discriminant analysis, showed specified more important subscales in predicting autism in each of the categories. Four subscales out of five, were related to social interaction category (Lack of sharing enjoyment with others, Theory of mind, Lack of social and emotional interaction, Failure to relate with peers), three subscales out of four related to communication (Delay in verbal communication, Lack of play, Stereotyping language), all of the three subscales of Stereotyping behaviors (Occupation, Inflexible following rituals, Repetitive gestures) and the two subscales in the Developmental delay category (Delay, Regulation factors) had fairly high tolerance coefficients in the regression four fold steps, that means it is considered important in predicting criterion variable (diagnosis of autism). In addition, two functions were computed to discriminate between groups at P
