Using Texting Mining and Machine Learning Techniques to Discover the Cognitive Behavioral Differences between Insomnia and Non-Insomnia

Motivation

Insomnia is a big issue in modern society. Various studies worldwide have shown the prevalence of insomnia in 10%–30% of the population, some even as high as 50%–60%. There are unknown reasons that cause insomnia, maybe it is caused by inheritance, anxiety or bad sleep hygiene…etc. Dan Isaac Slobin, an
authority linguistics dubbed "thinking for speaking". What this means is that the language we learn shapes the way we perceive reality and think about it. Therefore this research translates the interview with insomnia and non-insomnia people to scripts, and using “text-mining” to study the differences between insomnia and non-insomnia cognituive speaking differences. And try to understand what causes insomnia. More detail >>

Research Scheme

This research used Pittsburgh Sleep Quality Index (PSQI) to identify insomnia and non-insonia people. And used Mroin and Espir insomnia interview outline to interview 25 people. The average interview time was 29 minutes 58 seconds. Then the research type each interview to manuscripts. The research used Pennebaker and Chin-Lan Huang scholar Chinese Linguistic Inquiry and Word Count as a dictionary to classify words as 79 categories, then used feature selections to choose the categories that could make prediction insomnia and non-insonia people. And the research used T-Test and Man-Whiney U Test to find out whether the categorie words are different from insomnia and non-insonia people.

Research Result

“Think for speaking” How we think affect the way we speak. This research is using text mining and machine learning techniques to analyze insomnia interviews to find out the cognitive behavioral differences between insomnia and non-insomnia and to predict whether if the person is suffering from insomnia. The result indicated that this research can predict insomnia by using SVM with I-question, I-numbers, I-cause and I-past 4 attributes, and the accuracy of it is 92%. The pattern of how people suffer from insomnia is “they ask fewer questions during the interview”. They like to repeat the questions that are being asked by the interviewer instead of asking questions. The patterns and behavioral differences of non-insomnia people, who are middle-aged women with stable jobs and income. On the contrary, the patterns of people suffer from insomnia are middle-aged women with no job and no income and suffering from insomnia about 0.6 years. The hidden rule behind insomnia is taking hypnotics cannot provide people suffer from insomnia a better sleep - the support is 0.32 and confidence is 0.8.