PGAE-ICA_A simplified digital system for intellectual measurement-assessment in children and adolescents using cognitive testing and machine learning techniques

Measuring and assessing intelligence level in children and adolescents is crucial for monitoring their developmental progress, identifying intellectual disabilities, and implementing early interventions. To date, there is no digital and simplified tool specifically designed to evaluate whether intelligence is normal or abnormal in these age stages. The present study aims to develop an intelligence measurement-assessment system based on primary cognitive ability tests across four cognitive domains and a machine learning model to distinguish between normal and abnormal intelligence. A total of 103 participants aged 9 to 17, were recruited for this study, including 39 with abnormal intelligence and 64 with normal intelligence. Participants completed the Chinese Wechsler Intelligence Scale for Children and primary cognitive ability tests in a randomly counterbalanced order. Independent samples t-tests and partial correlation analyses were employed to validate whether the cognitive tests effectively reflected individual differences in intelligence and to examine the correlation between cognitive tests and intelligence, while excluding ineffective tests. A genetic algorithm-optimized extreme learning machine model was then constructed and trained to predict intellectual status of children and adolescents. The results indicated that after excluding the time selection task, the remaining eleven cognitive tests effectively reflected the differences between individuals with normal and abnormal intelligence, with significant positive correlations to intelligence. Meanwhile, the optimized extreme learning machine model achieved an overall prediction accuracy rate of 92.63%, outperforming the unoptimized basic extreme learning machine model as well as traditional logistic regression model and support vector machine model. Therefore, the validity of the digital and simplified intelligence measurement-assessment system for children and adolescents, named PGAE-ICA, developed in the present study has been confirmed, supporting its further application in clinical and scientific research fields. Furthermore, the PGAE-ICA establishes a foundation for the future development of artificial intelligence expert diagnostic systems for identifying intellectual abnormalities.儿童与青少年智力水平的测评，对于监测其发育进程、识别智力障碍并开展早期干预均具有重要意义。迄今为止，尚未有专门针对该年龄段人群设计的数字化轻量化智力正常/异常评估工具。本研究旨在开发一套基于四大认知领域基础认知能力测试的智力测评系统，并构建机器学习模型以区分智力正常与异常状态。本研究共招募103名9至17岁的参与者，其中智力异常者39名、智力正常者64名。所有参与者按照随机平衡顺序完成中文版韦氏儿童智力量表（Chinese Wechsler Intelligence Scale for Children）与基础认知能力测试。研究采用独立样本t检验与偏相关分析，验证认知测试能否有效反映个体智力差异，并考察认知测试与智力水平的相关性，同时剔除无效测试项。随后构建并训练了基于遗传算法（genetic algorithm）优化的极限学习机（extreme learning machine）模型，用于预测儿童青少年的智力状态。研究结果显示，剔除时间选择任务后，剩余11项认知测试可有效反映智力正常与异常群体间的个体差异，且与智力水平呈显著正相关。同时，经优化的极限学习机模型整体预测准确率达92.63%，优于未优化的基础极限学习机模型、传统逻辑回归（logistic regression）模型与支持向量机（support vector machine）模型。综上，本研究所开发的命名为PGAE-ICA的数字化轻量化儿童青少年智力测评系统已验证具备有效性，可为其在临床与科研领域的进一步应用提供支撑。此外，PGAE-ICA还为未来开发用于识别智力异常的人工智能专家诊断系统奠定了基础。