Reinforcement Learning with ABC Algorithm-Optimized Rewards for Dynamic Difficulty Adjustment in Cognitive Memory Games using Synthetic Data Augmentation

Sustaining engagement in cognitive monitoring for aging populations requires assessment tools that continuously adapt to individual performance. We proposed a dynamic difficulty adjustment (DDA) framework for a tablet-based memory-matching game, leveraging synthetic data generation, swarm intelligence, and reinforcement learning. Addressing typical data limitations, we generated behaviorally consistent synthetic gameplay trajectories via parametric sampling, augmenting an original dataset from healthy cognitive (HC) and possible Mild Cognitive Impairment (pMCI) participants. This augmented data allowed us to robustly train a Q-learning agent. Its reward function integrated performance metrics, error rates, and difficulty changes, with parameter weights optimized using the Artificial Bee Colony (ABC) algorithm. The agent’s policies, refined over multiple training episodes to Q-table convergence, closely aligned with expert-defined suitability criteria for difficulty adjustments. Upon evaluation using a test dataset, the system yielded 98.55% positive rewards in the HC group and 94.29% in the pMCI group. This high positive feedback rate across diverse participant groups underscores the DDA framework’s success in promoting player success and fostering an encouraging environment, vital for long-term engagement in cognitive monitoring.