Dataset for the Evaluation of Autism-Related Mobile Applications

This dataset was developed as part of a research study evaluating mobile applications related to autism spectrum disorder (ASD). The data collection and evaluation process was conducted by two authors of the associated research article, both of whom are graduate students in Industrial Engineering at Amirkabir University of Technology (Tehran Polytechnic). The initial stage of data preparation involved systematically searching, reviewing, and selecting relevant applications available on two major mobile application platforms, Google Play and the Apple App Store. Following the screening and selection process, a total of 45 autism-related mobile applications were identified and included in the final dataset. To develop a comprehensive evaluation framework, 45 primary evaluation criteria were initially identified based on the objectives of the study and the characteristics of autism-related mobile applications. These primary criteria were subsequently decomposed into more specific sub-criteria and operational questions. Through this hierarchical process, a total of 78 final evaluation features were established for assessing each application. The 78 features cover different aspects of the selected applications, including descriptive and structural characteristics as well as functional characteristics. For functional features, the initial assessment was performed using a three-level scoring scheme: 0, 0.5, and 1. A score of 1 indicates that the corresponding feature or functionality was fully present in the application, a score of 0 indicates that the feature was completely absent, and a score of 0.5 represents a partial or intermediate level of implementation. To evaluate the applications consistently, the selected applications were downloaded and examined directly by both researchers. The evaluation questions were independently reviewed and assessed based on the actual characteristics and functionalities observed in each application. Particular attention was given to applying the evaluation criteria consistently and basing scores on observable evidence rather than subjective assumptions. Following the initial assessments, a review and discussion process was conducted by the two researchers to compare their evaluations and resolve discrepancies. In the initial assessment stage, each feature was assigned a score of 0, 0.5, or 1 by each researcher. The evaluations were then combined to obtain the final scores. As a result of this aggregation, the final scores could take any decimal value between 0 and 1, including 0 and 1. This procedure was adopted to obtain a consistent representation of the two researchers’ evaluations and to reduce the potential influence of individual evaluator bias. In some cases, values for particular features were unavailable for specific applications. These missing values were retained in the original dataset and subsequently addressed during the data preprocessing stage using the predefined preprocessing methodology of the study. The resulting dataset represents the raw evaluation data prior to the application of statistical analysis and machine learning procedures. Subsequent preprocessing steps, including the treatment of missing values, data scaling and standardization, feature extraction or transformation where applicable, and feature selection or dimensionality reduction, were performed as separate stages of the analytical workflow. Therefore, this dataset provides the underlying application-level observations and evaluation features used as the basis for the subsequent data analysis and modeling procedures reported in the associated research article.