Abstract：
The Diagnostic Statistical Manual of Mental Disorders (DSM-5) recommends diagnosing neurocognitive disorders (i.e., cognitive impairment) when a patient scores beyond - 1 SD below neurotypical norms on two tests. I review how this approach will fail due to cognitive tests\' power limitations, validity issues, imperfect reliabilities, and biases, before summarizing their resulting negative consequences. As a proof of concept, I use developmental prosopagnosia, a condition characterized by difficulties recognizing faces, to show the DSM-5 only diagnoses 62-70% (n1 = 61, n2 = 165) versus 100% (n1 = 61) through symptoms alone. Pooling the DSM-5 missed cases confirmed the presence of group-level impairments on objective tests, which were further evidenced through meta-analyses, thus validating their highly atypical symptoms. These findings support a paradigm shift towards bespoke diagnostic approaches for distinct cognitive impairments, including a symptom-based method when validated effective. I reject dogmatic adherence to the DSM-5 approach to neurocognitive disorders, and underscore the importance of a data driven, transdiagnostic approach to understanding patients\' subjective cognitive impairments. This will ultimately benefit patients, their families, clinicians, and scientific progress.
摘要：
精神疾病诊断统计手册（DSM-5）建议诊断神经认知障碍（即，认知障碍）当患者在两项测试中得分超过-1SD低于神经典型标准时。我回顾了这种方法是如何由于认知测试的功率限制而失败的，有效性问题，不完美的可靠性，和偏见，在总结其由此产生的负面后果之前。作为概念的证明，我用发育性前失认症,一种以识别面孔困难为特征的状况，显示DSM-5仅通过症状诊断62-70％（n1=61，n2=165）和100％（n1=61）。合并DSM-5漏诊病例证实在客观测试中存在组级别的损伤，通过荟萃分析进一步证明了这一点，从而验证了他们的高度非典型症状。这些发现支持针对不同认知障碍的定制诊断方法的范式转变，包括在验证有效时基于症状的方法。我拒绝教条式地坚持DSM-5治疗神经认知障碍的方法，强调数据驱动的重要性，了解患者主观认知障碍的综合诊断方法。这最终会使患者受益，他们的家人,临床医生,和科学进步。
