Utilization of machine learning classifiers to predict different forms of mental illness: schizophrenia, PTSD, bipolar disorder and depression
Abstract
The most alarming, yet abstained issue of our so-called ‘Generation Z’ is mental
health. While there are seminars, psychotherapy and awareness procedures initi ated to tackle this issue in many developed countries, it is unfortunately treated
as a mere joke to a majority of the population among the developing nations. Ac cording to various research, the probability of depression is highly prone to younger
ones, however it can occur to any individual at any age category whether the person
is of 13 years old or late 60’s. The only way to tackle this is to find out the correct
mental illness associated with an individual and gradually provide a systematic so lution as early as possible before it gets to a stage we cannot bring them back from.
In our paper, we have emphasized on the category of a disease rather than just gen eralizing it as depression. We came up with four highly anticipated mental health
statuses which are Schizophrenia, PTSD, Bipolar Disorder and lastly, Depression.
Our research proposes to identify, or in other words “Classify” which of these mental
illnesses a person is most likely to be diagnosed with, if not a mentally healthy per son. We do this by examining the language patterns of such self-reported diagnosed
people from a corpus of Reddit posts. We also researched multiple classification
algorithms and state-of-art technologies to identify individuals with mental illness
through their language and discovered better outcomes. Our approaches and results
may be valuable not only in the development of tools by healthcare organizations
for detecting mental disorders but also in assisting the individuals, the ones affected,
to be more proactive in their life.