Mental Health Prediction Using Support Vector

Machine

Suraj Rawal
Siddhesh Salgaonkar
Omkar Sawant
Shahid Shaikh
Under the guidence of
Prof.Mhapasekar D.P.
mproject2223@gmail.com

Sindhudurg Shikshan Prasarak Mandal’s

College of Engineering, Kankavli.

January 28, 2023

1 / 33
Outline

1 Introduction

2 Literature Review

3 Problem Statement/s

4 Proposed System

5 Methodologies

6 Conclusion

7 Bibliography
Bibliography32

2 / 33
Introduction

3 / 33
Introduction

Mental illness is a health problem that undoubtedly impacts

emotions, reasoning, and social interaction of a person.
Mental illness is usually diagnosed based on the individual
self-report that requires questionnaires designed for the detection
of the specific patterns of feeling or social interactions.
An activity tracker that will helps monitor some type of human
activity, such as walking or running, sleep quality or heart rate for
the input data.

4 / 33
Introduction(contd.)

Daily activity levels show significant pathways in terms of fitness,

and healthy life is concerned.
We propose a system to predict depression by taking motor
activity data collected by actigraph wearable wristbands.
The combination of the motor sensor data with other demographic
clinical data can be used for detection of depression in the
patients.
To develop a precise prediction model we use support vector
machine technique.

5 / 33
Literature Review

6 / 33
Literature Review

Literature Review[1]
Author Name:M. Raihan , Anupam Kumar Bairagi and
ShagotoRahman.
Title:A Machine Learning Based Study to Predict Depression with
Monitoring Actigraph Watch Data.
Work:With the help of wearable devices data of activity
monitoring. They are presenting a classification model and an
approach to extract classification models for predicting depression
based on agumentation system.

7 / 33
Literature Review 1

RESULT:They achieved the highest accuracy with both ANN and

AdaBoost with 98
CONCLUSION:The clinical Demographic data with these motor
sensor data can be deadly combination in terms of classification.
Cohen’s Kappa coefficient and Mathew’s correlation coeffect
factors helps in achieving more accuracy.

8 / 33
Literature Review 2

Literature Review[2]
TITLE:Monitoring Motor Activity Data for Detecting Patient’s
Depression Using Data Augmentation and Privacy-Preserving
Distributed Learning .
AUTHOR:Amin Aminifar, Fazle Rabbi, Violet Ka I Pun and Yngve
Lamo.
WORK:In this paper, they proposed an approach based on data
augmentation to analyze the Depression dataset and improve the
performance of detecting depression in subjects.

9 / 33
Literature Review 2

RESULT:The result substantial improvement in the classification

performance by employing our approach for generating data
records from raw data.Tree Based random forest and Xgboost
present more accurate result when trained on the data generated.
CONCLUSION:PPD- ERT distributed data analysis technique
ensures reservation of privacy of sensitive information and leads
to learning more accurate models with upto 7.9 .

10 / 33
Literature Review 3

Literature Review [3]

TITLE:Actigraphic recording of motor activity in depressed
inpatients: a novel computational approach to prediction of clinical
course and hospital discharge.
AUTHOR;Ignacio Peis, Javier-David Lopez-Morinigo, M.Mercedes
perez- rodriguez and Enrique baca- Garcia.
WORK:They have presented clinical course and hospital
discharge based on actigraph recording of motor activity using
Hierarchical generalised linear model.

11 / 33
Literature Review 3

RESULT:They achieved high level of accuracy (mean error of

Prediction) 0.23 days by using the HGLM.
CONCLUSION:In this paper, two main conclusions can be drawn
from the study results. 1st ACTi graphically recorded motor
activity increase Over the inpatient episodes, which was inline with
previous studies. 2nd based on the overall early activity data
estimated hospital discharge date with sufficient activity for it to be
used in routine clinical practice, reaching the highest predictive
value at day 7,at which most patient were still admitted to the
ward.

12 / 33
Literature Review 4

Literature Review 4[4]

title:Mental Health Prediction Using Machine Learning Taxonomy,
Applications, and Challenges.
AUTHOR:Jetli Chung and Jason Teo.
WORK:They have presented PRISMA methodology in conducting
this systematic review.

13 / 33
Literature Review 4

Literature Review [3]

Result:A total of 30 research articles were included in this review
paper.e research articles were divided and cate-gorized based on
the mental health problems such as schizophrenia, bipolar
disorder, anxiety and depression, posttraumatic stress disorder,
and mental health problems among children.
CONCLUSION:research show that machine learning can be a
useful tool in helping understand psy-chiatric disorders. Besides
that, it may also help distinguishand classify the mental health
problems among patients for further treatment. Newer approaches
that use data that arisefrom the integration of various sensor
modalities present in technologically advanced devices have
proven to be a convenient resource to recognize the mood state
14 / 33
Literature Review 5

Literature Review5[5]
TITLE: Studying depression using imaging and machine learning
methods.
AUTHOR: Meenal J. PatelAlexander KhalafHoward J. Aizenstein.
WORK: They have presented prompted the development of
multiple machine learning methods to help improve the
management of this disease. These methods utilize anatomical
and physiological data acquired from neuroimaging to create
models that can identify depressed patients vs. non-depressed
patients and predict treatment outcomes.

15 / 33
Literature Review 5

RESULT:They achieved high level of accuracy 88.9 by using the

Cross-validation method.
CONCLUSION:In this paper, two main conclusions can be drawn
from the study results. 88.9 high accuracy with cross validation,
feature reduction and learning method.

16 / 33
Problem Statement

17 / 33
Problem Statement

To develop a machine leaning models for prediction of mental

health (Depression) of a individual based on the motor activity
data collected by Actigraph wearable wristbands using support
vector machine .

18 / 33
Proposed System

19 / 33
Proposed System

Develope a system to predict weather a person is going through a

depression or not.
In existing system we saw that there are depression predicting
system based on social media content.
The dataset used in by different researchers are varying. So we
have decided to develop our system which works on motor activity
data.
The classification technique we are going to use is Support Vector
Machine.

20 / 33
Flow chart

21 / 33
Methodologies

22 / 33
Methodologies

The SVM algorithm is commonly used for classification purposes,

particularly those such as classifying objects from unseen data
samples.
SVM works by mapping data to a high-dimensional feature space
so that data points can be categorized, even when the data are
not otherwise linearly separable.
Support vector machine works comparably well when there is an
understandable margin of dissociation between classes.

23 / 33
How SVM works?

24 / 33
Dataset Descrption

Directories Conditions(23 csv files) Controls (32 csv files)

There are two main csv files in the dataset Patient.csv Score.csv
For each patient a csv file has been provided containing the
actigraph data collected over time.
Patient.csv flie includes the following data . The columns are:
timestamp (one minute intervals) date (date of measurement)
activity (activity measurement from the actigraph watch).

25 / 33
Dataset Descrption

scores.csv It contains the following columns

1 number (patient identifier)
2 days (number of days of measurements)
3 gender (female or male)
4 age (age in age groups)
5 afftype (1: bipolar II, 2: unipolar depressive, 3: bipolar I)
6 melanch (1: melancholia, 2: no melancholia)

26 / 33
Dataset Descrption

7 inpatient (1: inpatient, 2: outpatient)

8 edu (education grouped in years)
9 marriage (1: married or cohabiting, 2: single)
10 work (1: working or studying, 2: unemployed/sick
leave/pension)
11 madrs1 (MADRS score when measurement started)
12 madrs2 (MADRS when measurement stopped).

27 / 33
Project Plan and Schedule for next Semester

Well planned project schedule.

28 / 33
Conclusion

There are many kinds of expert systems used in the psychiatry

domain to predict diseases accurately at an early phase so that
therapy can be made high effectively.
It becomes clear that the SVM classifier can be chosen and
developed upon it to obtain very high accuracy.
It is clear from the results that the SVM classifier generates more
precise results than the others.

29 / 33
Bibliography

[1] P. M. Raihan , A. M. Raihan , and A. Shagoto Rahman.

A machine learning based study to predict depression with
monitoring actigraph watch data.
IEEE Access, pages 1–5, 2021.
[2] Amin Aminifar , Fazle Rabbi, Violet Ka I Pun, and title=Monitoring
Motor Activity Data for Detecting Patient’s Depression Using Data
Augmentation and Privacy-Preserving Distributed Learning.
year=2021 number=
pages=1-4 doi=10.1109/EMBC46164.2021.9630592 yngve Lamo,
journal=IEEE Access.

30 / 33
Bibliography (contd.)

[3] gnacio Peis, , Javier-David Lopez-Morinigo, and title=Actigraphic

recording of motor activity in depressed inpatients: a novel
computational approach to prediction of clinical course and hospital
discharge. year=2020 number=
pages=1-11 doi=10.1038/s41598-020-74425-x M.Mercedes perez-
rodriguezandM.Mercedes perez- rodriguez, journal=NATURE
Access.
[4] Jetli Chung, and title=Mental Health Prediction Using Machine
Learning: Taxonomy, Applications, and Challenges. year=2022
number= pages=1-19 doi=10.1155/2022/9970363 Jason Teo,
journal=hindvi Access.
[5]
31 / 33
Bibliography (contd.)

[1] [2] [?] [4]

[5]

32 / 33
Thank You
???

33 / 33