Scribd
Upload
28 views

Paper 2 Review

This study uses natural language processing (NLP) and machine learning techniques to analyze text messages in order to predict suicidal ideation and increased psychiatric symptoms. NLP is used to analyze responses to an open-ended question about how individuals are feeling. Trigrams, or strings of three words, are found to best predict outcomes. Machine learning is then used to train a model to forecast the likelihood of suicidal thoughts or more severe mental health issues based on the NLP analysis. The goal is to use this low-cost NLP approach to help detect and prevent suicide globally.

Uploaded by

Lamia
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
28 views

Paper 2 Review

This study uses natural language processing (NLP) and machine learning techniques to analyze text messages in order to predict suicidal ideation and increased psychiatric symptoms. NLP is used to analyze responses to an open-ended question about how individuals are feeling. Trigrams, or strings of three words, are found to best predict outcomes. Machine learning is then used to train a model to forecast the likelihood of suicidal thoughts or more severe mental health issues based on the NLP analysis. The goal is to use this low-cost NLP approach to help detect and prevent suicide globally.

Uploaded by

Lamia
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 7

Novel Use of Natural Language Processing (NLP) to

Predict Suicidal Ideation and Psychiatric Symptoms in a


Text-Based

Submitted by
Lamia Nurtaj
23141057; section 01 ; team24
Submitted to
Annajiat Alim Rasel
MD Sabbir Hossain
Introduction
Suicide accounts for 5-6% of all fatalities worldwide, making it the 13th
most common cause of death. The likelihood of a suicide being completed
varies globally according to a number of sociodemographic factors, with
young adults, teenagers, and men having the greatest rates of suicide
completion. Male suicide attempts are more likely to end in death, even
though females are much more likely than men to plot their suicides. The
difference between the number of male and female suicides that are
successfully completed is especially noticeable in higher-income nations.
Suicide rates are greater in places with sophisticated health systems, in
contrast to the hazards of many medical diseases, which are lower in
more industrialized countries. This research analyzes short message
service (SMS, or text messaging) that takes place outside of a clinical
environment by using an NLP/machine learning analytical technique. In
order to anticipate suicide thoughts and elevated mental symptoms, this
study reflects the early phases of creating and utilizing a prediction
algorithm in a free-text platform (i.e., medical notes in EHRs, messages,
and social media). SMS text notifications that are created in real-time may
help guide prompt therapeutic responses that avert suicide and worsen
psychological anguish.
How NLP work
Higher level semantic analysis of the entire text of respondents' responses to the open-ended question "How are you
feeling today?" was carried out using software created by Wired Informatics. This program generates NLP-based
algorithms by utilizing the clinical Text Analysis Knowledge Extract System. The model was trained using texts from a
randomly selected half of the dataset. The "n-grams" feature was employed in the construction of the NLP algorithm, which
meant that predictions were predicated on a continuous string of n words. In order to maximize the positive predictive value
(PPV) and maximize the number of true positives, various configurations for n-gram size (i.e., the number of words in a
contiguous string) were tested.
A "trigram," or zero words before and three words after a "token" word, was found to be
the n-gram configuration that maximized the PPV. Bigrams (two words after a token)
and single word settings often resulted in lower specificity but higher sensitivity,
according to our research.After that, n-grams were decoded and fed into a machine
learning (ML) system to forecast the likelihood that patients will have suicidal thoughts
or experience more severe mental health issues (GHQ-12 ≥ 4). In order to generate
predictive use cases, the machine learning software takes data from patients who have
and have not experienced these outcomes to train the model using a LIBLINEAR
machine learning protocol, a flexible linear classifier that can classify up to millions of
occurrences and attributes. The PPV, sensitivity, and specificity of the model were then
assessed on the positive instances in the other half of the data that had not been
utilized for training.
TABLE 01
TABLE 02
Conclusion
Natural language processing (NLP) based machine
learning prediction techniques may be used to forecast
the likelihood of suicide as well as increased mental
health symptoms in mobile phone replies that are
delivered in free-text format. Novel natural language
processing (NLP) techniques have the potential to
provide low-cost, high-impact substitutes for existing
data monitoring systems, which are resource-intensive,
in the global drive to prevent suicide and its devastating
emotional, social, and financial effects. In the end, using
NLP technology for this purpose might make a big
difference in detecting and helping those who are more
likely to commit suicide. Even in environments with
limited resources, this will help achieve the ultimate aim
of decreasing suicides and suicide attempts.

You might also like