ADHD与自闭症诊断陷阱：创伤被误诊
推荐指数 34.0 NO. 020 · 2026.10.04
发布2026/10/03Score160Comments130
为什么值得看
剑桥大学出版社论文指出，ADHD和自闭症症状与复杂性创伤（C-PTSD）高度重叠，临床误诊率惊人。对AI工程师而言，这解释了为何基于行为标签的「心理健康AI诊断工具」可能放大系统性误判风险。
编辑判断
心理健康AI赛道近年涌入大量创业团队，用问卷+语音/行为识别做ADHD/自闭症筛查。这篇论文揭示的「症状重叠」问题直接挑战这类产品的技术假设——如果训练数据本身基于误诊标签，模型只会固化错误模式。
更值得警惕的是监管套利：FDA对SaMD（软件即医疗器械）的审批门槛远低于药物，但诊断类AI的误差成本由患者承担。做数字疗法或企业EAP产品的团队，建议重新评估底层诊断逻辑是否经过创伤知情（trauma-informed）校准，否则可能面临集体诉讼风险。
一个可落地的方向：用LLM辅助构建「鉴别诊断决策树」，把创伤史纳入必须采集的维度，而非可选字段。
社区反馈
意见分歧 117 条评论
核心争论：症状重叠下，ADHD/自闭症诊断是否被过度使用，创伤因素是否被系统性忽视
Because this is a technical article with nonstandard jargon, I'm going to state the most important piece of tacit knowledge that I know psychology professionals have but do not communicate to laymen (I personally came across it somewhere when reading the complete works of Eric Berne, but I can't rec
> the innate parts that existed prior to birth What does this mean? Do you mean the congenital aspects of a disorder?
The environmental values of the living hood of where the child is to be born in to; is an element I would add to how I intercept it. Those who are to be born in to an abusive environment may be more prone to continuing the traits of abuse than those who come from reasonable families.