AI can detect how lonely you are by analyzing your speech

According to a new scientific paper, artificial intelligence (AI) can detect loneliness from human speech with 94% accuracy.

U.S. researchers used several AI tools, including IBM Watson, to analyze the writings of older people interviewed about loneliness...

By analyzing the gaps in words, phrases, and silence during the interview, AI assessed the symptoms of loneliness and loneliness as accurately as the prejudiced participants themselves filled out and self-reported in the questionnaire.

AI also revealed that lonely people have a longer response to direct questions about loneliness and tend to express more sadness in their answers.

A team led by researchers at the University of California, San Diego Medical School used artificial intelligence technology to analyze natural language patterns (NLPs) to identify the loneliness of older people.

“Most studies use the direct question,” How often do you feel lonely? “, Said Ellenley, the lead author of the University of California, San Diego School of Medicine.

This project used natural language processing, a fair quantitative assessment of expressed emotions and emotions, in conjunction with regular loneliness measurement tools.

“The interesting thing about this tool is that it not only uses a dictionary-based approach, such as searching for specific words that reflect fear, but also looks like a pattern of the entire word used in the response.”

Experts say there has been a “lonely pandemic” characterized by rising suicide and opioid usage, lower productivity, higher medical costs, and higher mortality in the United States.

According to a UC San Diego survey released earlier this year, 85% of residents in an independent elderly housing community report moderate to severe loneliness.

The Covid-19 pandemic and the associated lockdowns have made people more lonely and worse.

Researchers wanted to know more about how natural language processing technology and machine learning models can predict the loneliness of older people in the community.

This study focused on 80 independent elderly people between the ages of 66 and 94, with an average age of 83 years.

Trained research staff conducted a structured interview with participants between April 2018 and August 2019 prior to the pandemic.

Participants were asked 20 questions from the UCLA Loneliness Scale. In this question, “How often do you feel left behind?” And “How often do you feel you are part of a group of friends?”

The UCLA loneliness scale question does not explicitly use the word “loneliness.”

Participants were also interviewed, recorded and manually transcribed during a personal conversation.

The transcript was then inspected using a natural language processing tool such as: IBM’s Watson Natural Language Understanding (WNLU) Software that quantifies emotions and expressed emotions.

WNLU uses deep learning to extract metadata from keywords, categories, emotions, emotions, and syntax.

Participants conducted a semi-structured interview on loneliness and self-reporting scale (UCLA loneliness scale) to assess loneliness before comparison. The writing record was sent to IBM’s Watson Natural Language Understanding Program (shown in the figure).

“Natural language patterns and machine learning allow us to systematically investigate long interviews from many individuals and explore how subtle speech functions such as emotions indicate loneliness,” said UCSD’s lead. Author Varsha Badal states.

“Similar sentiment analysis by humans is prejudiced, inconsistent, and requires extensive training to standardize.”

Using language capabilities, AI was able to predict loneliness with 94% accuracy when compared to a “quantitative model” (UCLA loneliness scale score).

AI predicted self-aware loneliness with 94% accuracy and (quantitative loneliness) with 76% accuracy (based on the results of the UCLA loneliness scale).

They found that lonely people had longer answers in personal interviews and were more sad when they answered direct questions about loneliness.

The study also revealed the difference between men and women. The latter was more likely than men to feel lonely during interviews.

Men also use more terrifying and joyful language than women, suggesting that the experience of negative and positive emotions is more extreme or that men are more free to express these emotions. I will.

“There may be subtle gender differences between emotions in the way older men and women express themselves as feeling lonely in response to direct questions,” Lee told Mail Online.

This study highlights the differences between research assessments of loneliness and individual subjective experiences of loneliness.

According to researchers, there may be “lonely speeches” that can be used to detect loneliness in the elderly.

With IBM Watson, users analyze text and use natural language understanding to extract metadata from content such as concepts, entities, keywords, categories, emotions, emotions, relationships, and semantic roles. I can do it

This may improve the way clinicians and families assess and treat loneliness in older people, especially during social isolation.

UCSD is currently investigating the reversely linked natural language pattern signatures of loneliness and wisdom in the elderly.

“Combine voice data with other assessments of cognition, motility, sleep, physical activity, and mental health to better understand and accelerate aging,” said study co-author Dilip Jeste of UCSD. Can be done. ”

In this study, AI accuracy was measured by comparing it to participants’ own reports of loneliness, but as pointed out, it does not necessarily reflect true emotions or emotions.

However, psychologists and professionals can use AI with self-reporting to improve the accuracy of their diagnosis.

“We [UCLA Loneliness Scale] There are some inaccuracies because the score depends on self-reporting, “Lee told Mail Online.

“But it is one of the most popular tools used in research because it does not explicitly use the word” loneliness “and consistently captures the traits of loneliness in gender prejudice.

“We want to develop more sensitive tools for assessing loneliness.”

This study American Journal of Geriatric Psychiatry..

How to Learn Artificial Intelligence Using Neural Networks

AI systems rely on artificial neural networks (ANNs) that try to simulate how the brain works to learn.

ANN can be trained to recognize patterns of information such as voice, text data, and visual images, which is the basis of many developments in AI in recent years.

Traditional AI “teach” an algorithm on a particular subject by using input to supply large amounts of information.

AI systems rely on artificial neural networks (ANNs) that try to simulate how the brain works to learn. ANNs can be trained to recognize patterns of information such as audio, text data, and visual images.

Practical applications include Google’s language translation service, Facebook’s facial recognition software, and Snapchat’s live image modification filters.

The process of entering this data can be very time consuming and is limited to one type of knowledge.

A new kind of ANN, called a hostile neural network, can compete with each other for the wisdom of two AI bots and learn from each other.

This approach is designed to speed up the learning process and improve the output produced by AI systems.