Machine Learning Techniques Used in Psychological Studies with Adolescents: A Systematic Review
Main Article Content
Abstract
During the last decade there has been an exponential growth of artificial intelligence models applied to the different fields of knowledge including psychology. Various studies have used these models in order to identify potential risks early. However, few studies focused on adolescents have used these techniques. This systematic review used the steps suggested by the Prisma model to identify studies that applied Machine Learning techniques to identify behavioral traits in adolescents. When applying the inclusion and exclusion criteria, 5 studies were identified in the PsycNET, PubMed, Scopus, Scielo, Web of Science and Science Direct databases. The main results show that the Machine learning algorithms mainly used individually or in combination, were logistic regression (n = 4) and Support Vector Machine SVM (n = 3) in addition to others such as Adaboost (n = 1) Nested ten-fold crossvalidation (n = 1), Random Forest (n = 1), Artificial Neural Network ANN (n = 1), and Extreme gradient boosting XGB (n = 1). This review highlights that the use of Machine learning methods provide reliable predictive tools as much or even more than the traditional statistical methods. Finally, the present review highlights the lack of studies using these tools in the field of psychology, mainly in adolescents.
Article Details
Section
The website https://journals.ucjc.edu/ (hereinafter the PLATFORM) contains open-access publications supported by the Camilo José Cela University's own resources. Therefore, special observance of current copyright legislation, as well as those regarding privacy and use of content, will be maintained at all times. In this regard, and as a medium designed for the transmission of scientific knowledge with no limitations other than those imposed by law, neither Camilo José Cela University nor any company affiliated with it, nor any member of its staff, assumes responsibility for the theories, approaches, opinions, and ideas expressed by the authors of articles published on said PLATFORM.
Publication on the PLATFORM entails the non-exclusive transfer of the exploitation rights of such publications, although the ownership of the moral rights of the author will correspond to the signatories of the articles under the conditions established by law. However, the UCJC imposes certain publication conditions, the violation of which will automatically invalidate the article even if it has been published, in which case it will be withdrawn from the PLATFORM automatically and without prior notice:
- Authors agree to submit original and unpublished articles in any format.
- They also agree not to submit their work for review in other publications or media while it is being reviewed by the PLATFORM. Violation of any of these conditions will automatically invalidate the article even if it has been published, in which case it will be withdrawn immediately.
- The PLATFORM is open access and non-profit. Therefore, it will not pay any money to authors who submit submissions to it, nor will it require any payment for publication.
- Authors who have published on the PLATFORM may have their work available for inclusion in repositories, databases, or other similar means of dissemination, provided that they appropriately reference the PLATFORM in the process.
In the event that any violation of the conditions described above is detected, the UCJC reserves the right to take appropriate legal action, completely independent of any action that may be taken by other affected parties, collaterally, for issues such as plagiarism or any other violations related to current legislation.