Ionian University, Corfu, Greece.
Adv Exp Med Biol. 2021;1338:39-46. doi: 10.1007/978-3-030-78775-2_6.
The topic of recreational drug consumption is still largely controversial around the globe. Factors that predispose people and lead to initial drug use include, among others, personality traits. The study of personality is a well-established domain of psychology, with multiple models having been developed, which are capable of predicting predisposition to a certain degree. Furthermore, addiction and other mental health issues carry stigma, which inhibits affected people from reaching out for support. Online web-based tools and automated systems have shown to be fairly effective in tackling stigma by eliminating the human factor. As such, a web-based decision support system (DSS) is developed and made publicly available, in order to inform users about their drug predisposition through an online personality survey. To accomplish the latter, the DSS utilizes multiple machine learning algorithms to extract patterns of personality, as modeled by the Big Five personality traits. The utilized algorithms turn out to be effective at predicting drug use for most of the 17 drugs that are considered, even in cases of high-class imbalance.
娱乐性药物使用的话题在全球范围内仍然存在很大争议。导致人们初次使用药物的因素包括个性特征等。个性研究是心理学中一个成熟的领域，已经开发出多种能够在一定程度上预测倾向性的模型。此外，成瘾和其他心理健康问题带有污名，这使得受影响的人无法寻求支持。在线网络工具和自动化系统已被证明通过消除人为因素，在解决污名方面非常有效。因此，开发了一个基于网络的决策支持系统（DSS）并公开发布，以便通过在线个性调查使用户了解他们的药物倾向。为了实现后者，DSS 使用多种机器学习算法来提取由五大个性特征建模的个性模式。事实证明，即使在高级不平衡的情况下，所使用的算法也能有效地预测大多数 17 种被考虑的药物的使用情况。
