Scribd
Upload
52 views18 pages

Presentation 1

The document summarizes a research project on knowledge discovery from mental health data in Bangladesh. The goals were to collect mental health patient data, predict length of patient stays using regression, predict suicidal attempts using classifiers, and identify influential attributes. Data was collected from a mental health hospital and preprocessed. Regression was used to predict stay duration and classifiers like SVM, KNN, NB were used to predict suicidal attempts. The models achieved accuracy of up to 79% for suicidal attempt prediction. Future work could involve social media data and ward-level patient forecasting.

Uploaded by

Nayeem
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
52 views18 pages

Presentation 1

The document summarizes a research project on knowledge discovery from mental health data in Bangladesh. The goals were to collect mental health patient data, predict length of patient stays using regression, predict suicidal attempts using classifiers, and identify influential attributes. Data was collected from a mental health hospital and preprocessed. Regression was used to predict stay duration and classifiers like SVM, KNN, NB were used to predict suicidal attempts. The models achieved accuracy of up to 79% for suicidal attempt prediction. Future work could involve social media data and ward-level patient forecasting.

Uploaded by

Nayeem
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 18

Knowledge Discovery From Mental

Health Data
Supervised By Submitted by
Shahidul Islam Khan Md Sohel Mahmud Avon (C151009)
Md Azizul Hakim (C151025)
Problem Definition
◦ In modern countries, people preserved and used mental health data in the
research field. By storing and analyzing the data they extract hidden useful
knowledge and improve their medical services.
◦ In this case, we are lagging behind. Our doctors cannot get enough help from
modern technology like data mining. By preserving mental health data in a
proper way and using the data for knowledge discovery by data mining
techniques we can help our doctors as well as improve our medical services.
Objective
◦ Collection of mental health data, digitized them and use them for knowledge discovery.
◦ Selecting the required attributes for the duration of stay of patient’s model using features
selection technique.
◦ Developing a model to predict the duration of stay of patients with a mental disorder in
hospitals.
◦ Selecting the required attribute for the suicidal attempts triggering factor using features
selection technique.
◦ Predicting suicidal attempts using classifiers.
◦ Among all the columns/features sorting out influential attributes for suicidal attempts.
Motivation
◦ According to the National Mental Health Survey in 2003-2005 about 16.05% of the adult population in
the country are suffering from mental disorders [2].

◦ Suicide attempt cases along with these mental diseases are also rising high in our country. Statistics say
that, every year 11,000 people committed suicide in Bangladesh, which means on average 172 people are
committing suicide in each district [1].
Related Work
◦ This paper “Supporting the Treatment of Mental Diseases using Data Mining” [3] analyzed 466 mental health
patient’s datasets to find the relation between diagnosis and attributes. They applied three machine-learning
techniques: Random forest, SVM, K-nearest neighbor and compared their performances on different measures of
accuracy in diagnosing mental health problems.
◦ Bhakta I. and Sau A. [4] developed a predictive model for prediction of depression among senior citizens of India
using machine learning classifiers. Data that were used in this study was collected from a slum in Kolkata. Naïve
Bayes (NB), logistic regression (LR), Multi-layer Perceptron (MLP), Support Vector Machines (SVM) and Decision
Trees (DT) machine learning techniques were used in this study.
◦ This paper “Predicting Generalized Anxiety Disorder among Women Using Random Forest Approach” [5] worked
with GAD data. They used machine learning approach Random Forest algorithm to find out the prediction model of
GAD.
Research Gap
Contribution
◦ We’ve collected real-life raw data (747) which includes suicidal data (59) from NIMH.

◦ We’ve used Regression model to predict the duration of stay of patients in mental hospital
based on their attributes.

◦ We’ve used feature selection technique to select triggering factors/attributes related to suicidal
attempt.

◦ We’ve used different classifiers to predict the suicide attempt tendency among the patients in
mental hospital based on their attributes.
Methodology
Flow Chart
Data Collection
◦ We’ve collected 747 patients data from NIMH.
◦ Each of the patients had average of 14 page full of
hand written information and diagnosis data.
◦ We found total 59 patients who attempted to commit
suicide.

Fig 1: National Institute of Mental Health and Hospital, Dhaka


Data preprocessing
◦ We’ve consulted with people who had domain knowledge about Mental health care and then
we sort out the symptoms and relevant data for our research.
◦ In our dataset, we cleaned up that data which are null and redundant. In the case of ages, we
have used the mean values.
◦ Categorical attributes like diagnosis, name, and gender are handled manually as random
replacement of these attributes’ missing values may occur significant change on the dataset.
Others attribute like the unit, symptoms are handled with zero values on a random basis.
◦ We have performed normalization process in our dataset. The process that we have applied is
Label Encoding.
Applied Model
◦ We’ve used feature selection technique to reduce the number of attributes and
to find out the triggering factors/attributes related to suicidal attempt. We had
total number of 86 columns. After applying the feature selection method it
reduced to only 23 columns.
◦ We used Multiple Linear Regression to predict the number of days a patient
stays in hospital with particular disease.
◦ Then we predict suicidal attempt using different classifiers such as SVM,
KNN, Gaussian NB.
Performance Measure
◦ In our study, we have used accuracy, precision, recall, and f1 score. Their
equations are:

• confusion matrix is a table that is often used to describe the performance of a classification model (or "classifier") on
a set of test data for which the true values are known.
Test Result
• How many days a patient stayed in
hospital (date difference).
After evaluation the RMSE was
36.99
• Suicidal Attempt prediction result is
shown in the figure

Fig 2: Suicidal Attempt prediction result evaluation with


Accuracy, Precision, Recall, F1 Score
Conclusion
◦ Almost every day in the daily newspaper we recognize the suicide news. Keeping these issues
in mind, we have collected real-life raw data from a mental health hospital.
◦ We have successfully used Regression model to predict the duration of stay of patients in
mental health hospital based on their symptoms, age, gender, previous mental health record,
and family mental health record. This model will help the hospital authorities to run the system
effectively.
◦ Also, we have succeeded to predict the mental health patients, who tend to attempt suicide
based on their symptoms, relevant information and diseases.
Future Work
◦ By analyzing social media data, we will try to extend our model to predict the
suicide attempts tendency among general people.
◦ Different ward wise patients number forecasting can be made.
◦ With a larger dataset accuracy and RMSE value may improve.
Reference
◦ [1] “Psychologist: 11,000 suicides every year in Bangladesh,” Dhaka Tribune, 20-Jan-2019. [Online].
Available: https://www.dhakatribune.com/bangladesh/nation/2019/01/20/psychologist-11-000-suicides-
every-year-in-bangladesh [Accessed: 6-oct-2019].
◦ [2] “WHO-AIMS Report on Mental Health System in Bangladesh.” [Online]. Available:
http://apps.searo.who.int/pds_docs/B0765.pdf. [Accessed: 18-Dec-2018].
◦ [3] Khan S. I. , Islam A. , Hossen A. , Zahangir T. I. , Hoque A. S. Md. L. “Supporting the treatment if
mental disease using data mining”. 2nd Int. Conf. on Innovations in Science, Engineering and
Technology (ICISET). Chittagong, Bangladesh, 2018.
◦ [4] Bhakta I. and Sau A., “Prediction of Depression among Senior Citizens using Machine Learning
Classifiers,” International Journal of Computer Applications, vol. 144, no. 7, pp. 11–16, 2016.
◦ [5] Wahidah Husain, Lee Ker Xim, Nur’ Aini Abdur Rashid, Neesha Jothi, “Predicting Generalized
Anxiety Disorder among Women Using Random Forest Approach”, 2016 3rd International Conference
on Computer and Information Sciences (ICCOINS).
Thank You

You might also like