Social network photos can determine whether suffering from depression: accuracy of 70%-depression, social networking-IT information

As a common mental illness of depression, external features is not obvious, and difficult to be clear judgment. Recently, however, researchers are different according to the color preferences in patients with depression, has developed a set of algorithms, through publishing of photos and pictures of "information" to identify patients with depression.

Medical studies have shown that people present effects of preference of color would be in the mood, in a pleasant mood tends to favor bright color; in a depressed mood when you choose dark colors. According to this theory, two scholars at Harvard University and the University of Vermont developed the "recognition of depression". In order to verify the accuracy of the algorithm they choose through a user's social networks on instagram photo released samples for testing. Researchers said the current algorithm for depression recognition accuracy is 70%.

Two researchers found 500 Amazon with Instagram account Turkey test robot team. To confirm in advance what subjects may be depressed, they provide all participants with a clinical depression questionnaire. Eventually, determined that there were 70 patients with clinical depression.

Sampling process, researchers from all instagram 40000 photos are selected in the test. In order to ensure the accuracy of tests, photos of healthy test subjects were selected from the latest release of 100 photographs, for patients with depression, selected in their treatment before the release of the 100 photos.

Through the use of color saturation, contrast, and facial recognition software to test and analysis of these photos, the researchers found that patients with depression tend to publish blue or grey dark color-tone photographs, and on the choice of filters also have certain preferences, such as "Inkwell" black and white filters is welcomed by patients with depression. By contrast, healthy people prefer higher brightness "Valencia" filter.

社交网络照片可判断是否患有抑郁症：准确率达70% - 抑郁症,社交网络 - IT资讯

抑郁症作为常见的精神疾病，由于外在表现特征不明显，从而不易被清楚判断。不过最近，有研究人员根据抑郁症患者对于颜色偏好的不同，研发了一套算法，通过人们发布的照片和照片中的“信息”来识别抑郁症患者。

医学界研究表明，人们当下喜好的颜色会受心情的影响，在心情愉悦时往往会偏向明亮的颜色；在心情低落的时候则会选择暗色系。根据这一理论，哈佛大学和佛蒙特大学的两位学者研发了“抑郁症识别算法”。为了验证这一算法的准确性，他们选择通过用户在社交网络instagram上发布的照片样本来作测试。研究人员表示，目前这一算法对于抑郁症患者的识别准确率为70%。

两位学者找到了500名拥有Instagram账号的亚马逊土耳其机器人团队员工进行测试。为了提前确认哪些测试者有可能是抑郁症患者，他们为所有测试者提供了临床抑郁症的调查问卷。最终，确定其中有70人为临床抑郁症患者。

取材过程中，研究人员从所有测试者的instagram中挑选了40000张照片。为了保证测试的准确性，健康测试者的照片选自最新发布的100张照片，对于抑郁症患者，挑选的是他们在治疗前发布的100张照片。

通过利用色彩饱和度、对比度及面部识别软件对这些照片的测试分析，研究人员发现抑郁症患者往往倾向于发布蓝色或灰色等暗色系颜色为主色调的照片，而且在滤镜的选择上也有一定的偏好，比如“inkwell”的黑白滤镜就比较受抑郁症患者的欢迎。相比之下，健康人更喜欢使用高亮度的“Valencia”滤镜。