The research team says such systems could be used in a variety of ways, such as early
We tested the algorithm on two large databasesdata and compared our results with other depression detection methods. In all cases, we were able to outperform existing methods in terms of their classification accuracy.
Abdul Sadka, Professor and Director of the Brunel Institute for Digital Futures
The algorithm was trained using two databasesdata that contains the history of thousands of Twitter users, as well as additional information about the mental health of these users. 80% of the information in each database was used to train the bot, and the remaining 20% was then used to test its accuracy.
First, the bot excludes all users with less thanthan five tweets, then he corrects spelling errors and abbreviations. It then takes into account 38 different factors, such as the use of positive and negative words, the number of friends and followers, and the number of emojis, to determine the user's mental and emotional state.
The team says such a system could potentiallymay signal that a user is depressed before they post something publicly. This can help inform a person about potential mental health problems in advance.
Read more:
There is another “planet” inside the Earth: how it saved nascent life
Nuclear fusion no longer needs millions of degrees: how the new method works
New study refutes light energy transfer theory