Indexat a

Llicència i ús

Citacions

Altmetrics

Grant support

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Xiya Tao reports financial support was provided by Chinese Scholarship Council. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Anàlisi d'autories institucional

Compartir

Mapping the landscape of Artificial intelligence for serious games in Health: An enhanced meta review

Publicat a:Computers In Human Behavior Reports. 18 100696- - 2025-05-01 18(), DOI: 10.1016/j.chbr.2025.100696

Autors: Tao, Xiya; Saenz-Lechon, Nicolas; Eckert, Martina

Afiliacions

Resum

This paper presents an enhanced meta-review of artificial intelligence (AI) applications in serious games (SGs) for health, focusing on adaptation, personalisation, and real-time data processing to improve rehabilitation outcomes. A systematic review methodology, following PRISMA-ScR guidelines, was employed to analyse studies from 2017 to 2025, identifying the AI algorithms most frequently used to personalise gaming experiences, improve patient engagement, and increase treatment efficacy. Our proposal categorises the algorithms based on their role in the game environment, which can either be game control or user assessment. Game control algorithms adapt the game environment and difficulty, while user assessment algorithms gather information about the player's state, such as performance, mood, or physiological data, to evaluate the treatment progress. The review also examines the growing trend of using multimodal data as input for machine learning models. Results show that a few well-known algorithms, such as Decision Trees (DT), Artificial Neural Networks (ANN), Fuzzy Logic (FL), Na & iuml;ve Bayes (NB), and Support Vector Machines (SVM), are frequently used. However, there is a clear distinction in their purposes: while FL is typically applied to game control tasks, SVMs are mainly used for user assessment. This review offers valuable insights for researchers in the field, providing a comprehensive overview of the suitability of different AI algorithms for various tasks in SGs, with a particular focus on personalisation and motivation.

Paraules clau