Artificial intelligence (AI) can detect loneliness with 94 per cent accuracy from a person’s speech, a new scientific paper reports.
Researchers in the US used several AI tools, including IBM Watson, to analyse transcripts of older adults interviewed about feelings of loneliness.
By analysing words, phrases, and gaps of silence during the interviews, the AI assessed loneliness symptoms nearly as accurately as self-reports of loneliness and questionnaires completed by the participants themselves, which can be biased.
The AI also revealed that lonely individuals tend to have longer responses to direct questions about loneliness, and express more sadness in their answers.
A team led by researchers at University of California San Diego School of Medicine used artificial intelligence technologies to analyze natural language patterns (NLP) to discern degrees of loneliness in older adults
‘Most studies use either a direct question of “how often do you feel lonely”, which can lead to biased responses due to stigma associated with loneliness,’ said senior author Ellen Lee at UC San Diego (UCSD) School of Medicine.
‘For this project, we used natural language processing, an unbiased quantitative assessment of expressed emotion and sentiment, in concert with the usual loneliness measurement tools.
‘The interesting thing about the tool is that it does not just use a dictionary-based approach – for example searching for specific words that reflect fear – but it looks as patterns across the words used in the response.’
There has been a ‘loneliness pandemic’, marked by rising rates of suicides and opioid use, lost productivity, increased health care costs and rising mortality in the US, the experts say.
WHAT IS NATURAL LANGUAGE PROCESSING?
Natural language processing (NLP) is a branch of AI that helps computers understand, interpret and manipulate human language.
NLP helps computers communicate with humans in their own language and scales other language-related tasks.
For example, NLP makes it possible for computers to read text, hear speech, interpret it, measure sentiment and determine which parts are important.
A UC San Diego study published earlier this year found that 85 per cent of residents living in an independent senior housing community reported moderate to severe levels of loneliness.
The Covid-19 pandemic and resulting lockdowns have increased the amount of time people have been in solitude, making things worse.
Researchers wanted to know more about how natural language processing techniques and machine learning models can predict loneliness in older community-dwelling adults.
The study focused on 80 independent senior living residents aged between 66 and 94 years, with a mean age of 83.
Trained study staff conducted semi-structured interviews with participants before the pandemic, between April 2018 and August 2019.
Participants were asked 20 questions from the UCLA Loneliness Scale, which uses a four-point rating scale for questions such as ‘how often do you feel left out?’ and ‘how often do you feel part of a group of friends?’
None of the questions in the UCLA Loneliness Scale explicitly use the word ‘lonely’.
Participants were also interviewed during personal conversations, which were taped and manually transcribed.
Transcripts were then examined using natural language processing tools, including IBM’s Watson Natural Language Understanding (WNLU) software, to quantify sentiment and expressed emotions.
WNLU uses deep learning to extract metadata from keywords, categories, sentiment, emotion and syntax.
Participants completed semi-structured interviews regarding the experience of loneliness and a self-report scale (UCLA loneliness scale) to assess loneliness, which were then compared. Transcripts were fed into the IBM’s Watson Natural Language Understanding program (depicted)
‘Natural language patterns and machine learning allow us to systematically examine long interviews from many individuals and explore how subtle speech features like emotions may indicate loneliness,’ said first author Varsha Badal at UCSD.
‘Similar emotion analyses by humans would be open to bias, lack consistency and require extensive training to standardise.’
Using linguistic features, the AI could predict loneliness with 94 per cent precision when compared against the ‘quantative model’ – the scores from the UCLA Loneliness Scale.
The AI predicted self-acknowledged loneliness with 94 per cent precision and ‘quantitative loneliness’ (based on results from the UCLA Loneliness Scale) with 76 per cent precision.
Lonely individuals had longer responses in the personal interview and expressed greater sadness when answering direct questions about loneliness, they found.
The study also revealed differences between men and women – the latter were more likely than men to acknowledge feeling lonely during interviews.
And men used more fearful and joyful words in their responses compared to women, suggesting that their experiences of negative and positive emotions were more extreme, or even that men can express these emotions more freely.
‘There may be subtle sex differences in sentiment and emotion in how older men and women describe feeling lonely in response to a direct question,’ Lee told MailOnline.
The study highlights discrepancies between research assessments for loneliness and an individual’s subjective experience of loneliness, which AI could help identify.
There may be ‘lonely speech’ that could be used to detect loneliness in older adults, the researchers say.
IBM Watson lets users analyse text to extract metadata from content such as concepts, entities, keywords, categories, sentiment, emotion, relations, and semantic roles using natural language understanding
This could improve how clinicians and families assess and treat loneliness in older adults, especially during social isolation.
UCSD is now exploring natural language pattern signatures of loneliness and wisdom, which are inversely linked in older adults, meaning as one rises, the other falls.
‘Speech data can be combined with our other assessments of cognition, mobility, sleep, physical activity and mental health to improve our understanding of ageing and to help promote successful ageing,’ said study co-author Dilip Jeste at UCSD.
The study measured AI’s accuracy against the participants’ own reports of loneliness, which as pointed out, do not always reflect true feelings and emotions.
However, AI and self-reports can be used in tandem by psychologists and professionals to increase the accuracy of a diagnosis.
‘We agree that the [UCLA Loneliness Scale] score has some inaccuracies as it relies on self-report,’ Lee told MailOnline.
‘However it is one of the most popular tools used in research as it does not explicitly use the word ‘lonely’ and seems to capture the trait of loneliness consistently without a bias by sex.
‘We hope to develop more sensitive tools to assess the state of loneliness.’
The study has been published in the American Journal of Geriatric Psychiatry.
HOW ARTIFICIAL INTELLIGENCES LEARN USING NEURAL NETWORKS
AI systems rely on artificial neural networks (ANNs), which try to simulate the way the brain works in order to learn.
ANNs can be trained to recognise patterns in information – including speech, text data, or visual images – and are the basis for a large number of the developments in AI over recent years.
Conventional AI uses input to ‘teach’ an algorithm about a particular subject by feeding it massive amounts of information.
AI systems rely on artificial neural networks (ANNs), which try to simulate the way the brain works in order to learn. ANNs can be trained to recognise patterns in information – including speech, text data, or visual images
Practical applications include Google’s language translation services, Facebook’s facial recognition software and Snapchat’s image altering live filters.
The process of inputting this data can be extremely time consuming, and is limited to one type of knowledge.
A new breed of ANNs called Adversarial Neural Networks pits the wits of two AI bots against each other, which allows them to learn from each other.
This approach is designed to speed up the process of learning, as well as refining the output created by AI systems.