Depression and Anxiety Detection from Blog Posts Data
Name
Yevhen Tyshchenko
Abstract
Depression and anxiety affect the life of many individuals and if the diagnosis is not
stated in time it could lead to considerable health decline and even suicide. Nowadays,
mental health specialists, as well as data scientists, work towards analyzing social
media sources and, in particular, publicly available text messages and blogs to identify
depressed people and provide them with necessary treatment and support. In this work,
we adopt an experimental data collection approach to gather a corpus of blog posts from
clinical and control subjects. Ill people are considered as clinical subjects while control
subjects refer to healthy individuals. We inspect the latent topics found in collected
data to analyze the blog’ content according to themes covered by blog authors. We
experiment with various text encoding techniques such as Bag-of-Words (BOW), Term
Frequency-Inverse Document Frequency (TFIDF) and topic model’s features. We apply
Support Vector Machines (SVM) and Convolutional Neural Network (CNN) classifiers
to discriminate between clinical and control subjects. Additionally, we explore the
classification performance of CNNs trained on blog post texts of different size. The
best accuracy and recall scores of 78% and 0.72 respectively were obtained with a
Convolutional Neural Network (CNN) classifier initialised with pretrained GloVe word
vectors
stated in time it could lead to considerable health decline and even suicide. Nowadays,
mental health specialists, as well as data scientists, work towards analyzing social
media sources and, in particular, publicly available text messages and blogs to identify
depressed people and provide them with necessary treatment and support. In this work,
we adopt an experimental data collection approach to gather a corpus of blog posts from
clinical and control subjects. Ill people are considered as clinical subjects while control
subjects refer to healthy individuals. We inspect the latent topics found in collected
data to analyze the blog’ content according to themes covered by blog authors. We
experiment with various text encoding techniques such as Bag-of-Words (BOW), Term
Frequency-Inverse Document Frequency (TFIDF) and topic model’s features. We apply
Support Vector Machines (SVM) and Convolutional Neural Network (CNN) classifiers
to discriminate between clinical and control subjects. Additionally, we explore the
classification performance of CNNs trained on blog post texts of different size. The
best accuracy and recall scores of 78% and 0.72 respectively were obtained with a
Convolutional Neural Network (CNN) classifier initialised with pretrained GloVe word
vectors
Graduation Thesis language
English
Graduation Thesis type
Master - Computer Science
Supervisor(s)
Kairit Sirts
Defence year
2018