Mental disorders, such as depression, are increasingly concerned and have significantly affected an individual's physical health. Artificial intelligence (AI) approaches have recently been developed to support mental health professionals, primarily psychiatrists and clinicians, with decision-making based on patients' historical data (e.g., clinical history, behavioural data, social media use, etc.). There is a significant need to cope with fundamental mental health issues in children that can lead to complicated, if not treated at an early stage. Hence, in this paper, Deep Learning assisted Integrated Prediction Model (DLIPM) has been proposed to early forecast and diagnose children's mental illness. In the suggested model, convolutional neural networks (CNN) is first constructed to learn deep-learned patient behavioural data features. By embedding semantic mathematical methods of behaviour or brain dynamic forces into a statistical deep learning framework, insights into disruption, effective classification, and forecast can be achieved. The simulation analysis shows that the proposed model enhances sensitivity rate of 97.9%, specificity rate of 96.7%, recall ratio of 95.6%, the precision ratio of 90.1% of F-measure rate of 95.6%, and less error rate of 9.2% than other existing methods.

精神障碍，例如抑郁症，越来越引起人们的关注，并严重影响了个人的身体健康。最近开发了人工智能（AI）方法，以支持心理健康专业人员（主要是精神科医生和临床医生）根据患者的历史数据（例如，临床历史，行为数据，社交媒体使用等）进行决策。迫切需要解决儿童的基本心理健康问题，如果不及早治疗，这可能会导致复杂的问题。因此，在本文中，提出了深度学习辅助的综合预测模型（DLIPM）来对儿童的精神疾病进行早期预测和诊断。在建议的模型中，首先构建卷积神经网络（CNN）以学习深度学习的患者行为数据特征。通过将行为或大脑动力的语义数学方法嵌入统计深度学习框架中，可以获得对破坏的见解，有效的分类和预测。仿真分析表明，提出的模型提高了F测评率的97.9％，特异度96.7％，召回率95.6％，准确率90.1％，95.6％，错误率9.2％。其他现有方法。