This capstone project is being implemented during my senior year for consideration towards the bachelor degree.

Role(s):

Data Analyst

Deep Learning Developer

Project Duration:

Feb 2022- Present

What did I do?

Literature Survey

Data Gathering

Storyboarding

Defining the workflow

Exploratory Data Analysis

Coding the module(s)

Tools:

Visual Studio Code

Google Colaboratory

GitHub

WhiteBoard

Pen and Paper

Team Members

ND Bhavana

Mrs. Devahema D (PhD) - Guide

What is Schizophrenia?

Schizophrenia (SZ) is a chronic, severe mental disorder that affects the way a person thinks, acts, expresses emotions, perceives reality, and relates to others. Though SZ isn’t as common as other major mental illnesses, it can be the most chronic and disabling. People with SZ often have problems doing well in society, at work, at school, and in relationships. They might feel frightened and withdrawn, and could appear to have lost touch with reality. This lifelong disease can’t be cured but can be controlled with proper treatment.

Problem Statement

Deep learning algorithms' extraordinary precision in classification and prediction tasks has transformed a wide number of scientific domains, including SZ research. Deep learning has the potential to be a useful tool for clinicians in the diagnosis, prediction, and treatment of SZ. Thus, the aim of this project is to detect children with SZ by analysing Electroencephalography (EEG).

💻 What is the project about ?

Predicting the risk of SZ in children by developing a deep learning Bi-LSTM framework to work on single electrode readings of EEG tests performed on children in the age group of 9-12.

🎯 Objectives

To develop a mechanism to detect SZ in children, which is:

• Simple to use and interpret
• Efficient
• Can be adopted for industrial applications

🧑‍⚕️👩‍⚕️ Target Audience

Clinical Psychologists

Psychiatrists

Lab Technicians

Hospitals and Clinics

👎Disadvantages of Existing Systems

• Maximizes the complexity of the problem
• Have to undergo extensive fine-tuning of the test dataset parameters.
• Unstable and difficult to train.
• High prediction complexity for large datasets
• Tedious message updating
• Solutions have been proved ineffective
• Narrowly specialized knowledge

👍Advantages of Proposed System

• Achieving better performance metrics than alternative state-of-the-art models.
• Simple to use and interpret
• Can achieve incredible accuracy
• Analyses the dynamic relationships between the features of high dimensional data.
• Learn long time dependencies with its feedback connections.
• Lead to high detection accuracy and a low false alarm rate
• Reduces the consumption of hardware resources
• Can be adopted for better prediction in industrial applications.

Architecture Diagram