使用人工智能在社交媒体上评估厌食症时的性别偏见:数据驱动研究
David Solans Noguero1,2, Diana Ramírez-Cifuentes2,3, Esteban Andrés Ríssola4
1Telefonica I+D, Telefónica Research, Barcelona, Spain.
Journal of medical Internet research
|June 8, 2023
概括
机器学习算法在社交媒体上检测神经性厌食症时显示性别偏见,女性的假负值较高. 缓解策略可以减少但不能消除这些差异.
科学领域:
- 计算语言学计算语言学
- 心理健康信息学心理健康信息学
- 机器学习伦理学 机器学习伦理学
背景情况:
- 社交媒体是心理健康信息的重要来源,包括神经性厌食症等饮食障碍.
- 人工智能 (AI) 和机器学习 (ML) 可以放大输入数据中存在的偏差.
- 心理健康检测的自动化方法需要仔细审查,以防止有偏见的歧视.
研究的目的:
- 从社交媒体中检测神经性厌食症的算法中检测和分析基于性别的绩效差异.
- 调查自动心理健康检测系统中的偏见.
- 为敏感领域开发更公平的自动分类器.
主要方法:
- 在西班牙的社交媒体帖子数据集上训练有素的自动预测器,来自有和没有神经性厌食症的用户.
- 检查了男性和女性用户之间的预测性绩效差异.
- 应用特征级偏差特征和具有临床相关性的比较分析.
- 展示了用于更公平的自动分类器的偏差缓解策略.
主要成果:
- 与男性样本 (0.005) 相比,女性样本 (0.082) 的假阴性率 (FNR) 明显较高.
- 对于男性来说,相关的特征包括生物过程和自杀风险;对于女性来说,年龄,情绪和个人担忧是关键.
- 提出了减轻偏差的技术,表明差异可以减少,但不能完全消除.
结论:
- 心理健康检测的自动化方法在临床部署之前需要严格的偏见评估.
- 人工智能驱动的诊断工具可以影响有风险的个体;公平是至关重要的.
- 需要进一步的研究,以确保人工智能在精神卫生保健中的公平表现.
相关概念视频
01:28Anorexia Nervosa
155
Anorexia nervosa is a complex and severe eating disorder characterized by an intense fear of weight gain, an unrelenting pursuit of thinness, and a distorted body image. It often leads to dangerously low body weight relative to an individual's age and height. This disorder is marked by significant physical and psychological consequences, making it one of the most life-threatening psychiatric illnesses.
Symptoms and Physical Effects
Individuals with anorexia nervosa commonly exhibit extreme...
Symptoms and Physical Effects
Individuals with anorexia nervosa commonly exhibit extreme...
155
02:16Stereotype Content Model
14.8K
The Stereotype Content Model (SCM) was first proposed by Susan Fiske and her colleagues (Fiske, Cuddy, Glick & Xu, 2002; see also Fiske, 2012 and Fiske, 2017). The SCM specifies that when someone encounters a new group, they will stereotype them based on two metrics: warmth—or that group’s perceived intent, and how likely they are to provide help or inflict harm—and competence—or their ability to carry out that objective. Depending on the warmth-competence...
14.8K
01:29Bias in Epidemiological Studies
384
Biases can arise at various stages of research, from study design and data collection to analysis and interpretation. Recognizing and addressing these biases is essential to ensure the validity and reliability of epidemiological findings.Broadly speaking, biases in epidemiology fall into three main categories: selection bias, information bias, and confounding. A more detailed description of possible biases is:
384
01:30Bulimia Nervosa
166
Bulimia nervosa is a complex and severe eating disorder characterized by a cyclical pattern of binge-and-purge eating pattern. It generally involves an episode of binge eating, followed by compensatory behaviors such as vomiting, excessive exercise, laxative use, or fasting, to prevent weight gain. Despite often maintaining a normal weight, individuals with bulimia are intensely preoccupied with their body image and harbor an overwhelming fear of gaining weight. This can contribute to the...
166
01:23Binge Eating Disorders
149
Binge eating disorder is a significant mental health condition characterized by recurrent episodes of excessive food consumption within a short period, accompanied by a perceived loss of control over eating behavior. Unlike occasional overeating, binge eating disorder is marked by distressing emotions such as guilt, shame, and anxiety following binge episodes. The disorder affects individuals across different ages and backgrounds, with profound implications for physical and psychological...
149
02:55Stereotypes, Prejudice, and Discrimination
90.3K
Humans are very diverse and although we share many similarities, we also have many differences. The social groups we belong to help form our identities (Tajfel, 1974). These differences may be difficult for some people to reconcile, which may lead to prejudice toward people who are different. Prejudice is a negative attitude and feeling toward an individual based solely on one’s membership in a particular social group (Allport, 1954; Brown, 2010). Prejudice is common against people who...
90.3K