Design of an interactive system for aging cognitive training based on the concept of digital therapy

With the increasingly serious aging of the world's population, mild cognitive impairment (MCI) has become a public health problem affecting the health of the elderly. Traditional cognitive intervention methods have high cost, difficulty in implementation, lack of individual differences in training, etc., which cannot be applied to the needs of modern society. This paper designs a cognitive training system suitable for the elderly. The system is based on the Flutter-cross-platform framework, integrating cognitive assessment with adaptive training that modifies task complexity based on ongoing user assessment results, and supports multimodal interaction. After introducing the reinforcement learning algorithm based on proximal strategy optimization (PPO), the system dynamically adjusts training difficulty in real time by analyzing each user's performance data, tailoring exercises to their individual progress and needs. The implementation method of visual specifications for aging security protection of server-side proxy rack API is gradually summarized. The test results show that the system's adaptive training features, such as real-time difficulty adjustment, enhance user engagement and participation continuity, providing a practical direction for digital therapy cognitive intervention.