Testing domain knowledge and risk of bias of a large-scale general artificial intelligence model in mental health.
Testing domain knowledge and risk of bias of a large-scale general artificial intelligence model in mental health.
复制标题
测试心理健康领域大规模通用人工智能模型的领域知识和偏差风险。
DOI：
10.1177/20552076231170499
复制
发表时间：
2023-01
期刊：
影响因子：
3.9
通讯作者：
Jacobson, Nicholas C.
中科院分区：
文献类型：
--
作者：
Heinz, Michael V.;Bhattacharya, Sukanya;Trudeau, Brianna;Quist, Rachel;Song, Seo Ho;Lee, Camilla M.;Jacobson, Nicholas C.
关键词：
随着心理健康护理的需求和可用性之间的差距迅速扩大，人工智能（AI）为心理健康评估和治疗提供了一种有前途的可扩展解决方案。鉴于这种系统的新奇和不可理解的性质，探索性的措施，旨在了解领域知识和潜在的偏见，这样的系统是必要的，正在进行的翻译开发和未来的部署在高风险的医疗保健设置。我们研究了生成的AI模型的领域知识和人口统计学偏见，该模型使用具有系统性变化的人口统计学特征的人工临床小插曲。我们使用平衡精度（BAC）来量化模型的性能。我们使用广义线性混合效应模型来量化人口因素和模型解释之间的关系。我们发现不同诊断的模型表现不同;注意缺陷多动障碍、创伤后应激障碍、酒精使用障碍、自恋型人格障碍、暴食症和广泛性焦虑症显示出较高的BAC（0.70 ≤ BAC ≤ 0.82）;双相情感障碍、神经性贪食症、巴比妥酸盐使用障碍、行为障碍、躯体症状障碍、苯二氮卓类药物使用障碍、LSD使用障碍、表演性人格障碍和功能性神经症状障碍表现为低BAC（BAC ≤ 0.59）。我们的研究结果表明，在大型AI模型的领域知识中，性能变化可能是由于更突出的标志性症状，更窄的鉴别诊断以及某些疾病的更高患病率。我们发现模型人口统计学偏倚的证据有限，尽管我们确实观察到模型结果反映了真实世界差异患病率估计的一些性别和种族差异。
With a rapidly expanding gap between the need for and availability of mental health care, artificial intelligence (AI) presents a promising, scalable solution to mental health assessment and treatment. Given the novelty and inscrutable nature of such systems, exploratory measures aimed at understanding domain knowledge and potential biases of such systems are necessary for ongoing translational development and future deployment in high-stakes healthcare settings. We investigated the domain knowledge and demographic bias of a generative, AI model using contrived clinical vignettes with systematically varied demographic features. We used balanced accuracy (BAC) to quantify the model’s performance. We used generalized linear mixed-effects models to quantify the relationship between demographic factors and model interpretation. We found variable model performance across diagnoses; attention deficit hyperactivity disorder, posttraumatic stress disorder, alcohol use disorder, narcissistic personality disorder, binge eating disorder, and generalized anxiety disorder showed high BAC (0.70 ≤ BAC ≤ 0.82); bipolar disorder, bulimia nervosa, barbiturate use disorder, conduct disorder, somatic symptom disorder, benzodiazepine use disorder, LSD use disorder, histrionic personality disorder, and functional neurological symptom disorder showed low BAC (BAC ≤ 0.59). Our findings demonstrate initial promise in the domain knowledge of a large AI model, with performance variability perhaps due to the more salient hallmark symptoms, narrower differential diagnosis, and higher prevalence of some disorders. We found limited evidence of model demographic bias, although we do observe some gender and racial differences in model outcomes mirroring real-world differential prevalence estimates.
登录
查看更多内容
影响因子：
15.2
作者：
Korngiebel DM;Mooney SD
通讯作者：
Mooney SD
影响因子：
3.2
作者：
Sezgin E;Sirrianni J;Linwood SL
通讯作者：
Linwood SL
影响因子：
17.7
作者：
Regier, Darrel A.;Narrow, William E.;Kupfer, David J.
通讯作者：
Kupfer, David J.
影响因子：
6.2
作者：
Halladay AK;Bishop S;Constantino JN;Daniels AM;Koenig K;Palmer K;Messinger D;Pelphrey K;Sanders SJ;Singer AT;Taylor JL;Szatmari P
通讯作者：
Szatmari P
影响因子：
--
作者：
Vermani, Monica;Marcus, Madalyn;Katzman, Martin A
通讯作者：
Katzman, Martin A