The document provides an overview of schizophrenia, detailing its definition as a severe mental illness affecting perception, behavior, and thoughts, with insights into potential causes, symptoms, and treatment options. It elaborates on early warning signs, different types of schizophrenia, and emphasizes the importance of genetic and environmental factors in its development. Additionally, it discusses treatment methods and includes a statistical analysis comparing classifier models for predicting the disorder's characteristics.

1. Schizophrenia
PRESENTED BY: GUIDED BY:
AVINASH M. HANWATE PROF. B. R. BOMBADE
2016MNS013

2. Contents:
 Introduction
 Causes
 Symptoms
 Types
 Implementation
 Treatment
 Conclusion

3. What is Schizophrenia?
 Schizophrenia is a serious mental illness that has an effect on how a person feels,
behaves, and thinks.
 Brain disorder
 Often confused with “Multiple Personality Disorder.”
 Patients of Schizophrenia are not violent and do not pose a danger to others.

4. Causes:
 No one knows the exact causes of Schizophrenia, but multiple possible factors have
been discovered.
 These factors include:
1. Genetics
2. Brain chemical imbalance
3. Environmental factors
4. Family history

5. 1. Genetics:
 It may appear when the body undergoes hormonal and physical changes.
 It may also appear after dealing with highly stressful situations.
 Increase the incidence of schizophrenia in subjects related to an affected person.

6. 2. Brain chemical imbalance:
 Abnormal production or reaction to the important chemical neurotransmitter
dopamine and maybe others.
 Dopamine is responsible in sending messages through the brain and effects how
the brain perceives things.
 Imbalance between dopamine and serotonin.

7. 3. Environmental factors:
 Stress inducing environmental factors such as social pressure, physical or sexual
abuse, loss of loved ones, hormones, malnutrition, and early exposure to viruses.
 The major brain changes that occur during puberty has also been identified as a
possible contributing factor.

8. 4. Family history:
 If ones mother, father, brother or sister has the illness, statistics show a 10 percent
chance of developing it.
 If ones twin, or both father and mother has the disorder then the probability of
developing it rises to 40 percent.

9. Early warning signs of schizophrenia
 Hearing or seeing something that isn’t there.
 A constant feeling of being watched.
 Peculiar or nonsensical way of speaking or writing.
 Strange body positioning.
 Feeling indifferent to very important situations.
 Worse in academic or work performance.

10. Symptoms:
 The symptoms of schizophrenia can be categorized into three wide-ranging groups.
1. Positive symptoms
2. Negative symptoms
3. Cognitive symptoms

11. Types:
1. Catatonic : Consists of activity levels on either end of the spectrum; either a dazed,
coma-like state or a hyperactive state
2. Paranoid : Typically consists of false beliefs and hearing things that aren’t there.
3. Disorganized : Typically consists of thoughts, speech and behavior that is
inappropriate and incomprehensible
4. Residual : Consists of some schizophrenic symptoms of lesser severity
5. Undifferentiated : A form of schizophrenia that does not match any of the above
types; sometimes this is called schizophrenia not otherwise specified

12. Implementation
 Preparatory step:
 Load the libraries:
 library(caret), library(randomForest), library(e1071), library(kernlab), library(doMC),
library(foreach), library(RColorBrewer).
 Read the data:
 Training set:
trFC <- read.csv('/Your-Path-To-Train-Data/train_FNC.csv')
trSBM <- read.csv('/Your-Path-To-Train-Data/train_SBM.csv')
tr <- merge(trFC, trSBM, by='Id')
 Test set:
tstFC <- read.csv('/Your-Path-To-Test-Data/test_FNC.csv')
tstSBM <- read.csv('/Your-Path-To-Test-Data/test_SBM.csv')
tst <- merge(tstFC, tstSBM, by='Id')
y <- read.csv('/Your-Path-To-Train-Data/train_labels.csv')

13.  Analysis
 "Feature Trimming”
 Registering 6 cores to speed up my computations:
registerDoMC(cores=6)
 Converting a y-label vector into appropriate format:
y <- as.factor(paste('X.', y[,2], sep = ‘ '))
 Introducing a random vector into feature set:
all <- cbind(tr, rnorm(1:dim(tr)[1]))
colnames(all)[412] <- 'rand‘
 Now train Random Forest with this (full) feature set:
rf.mod <- foreach(ntree=rep(25000, 6), .combine=combine, .multicombine=TRUE,
.packages='randomForest') %dopar% {
randomForest(all[,2:412], y, ntree=ntree)
}

14. • Looking at the feature importances:
color <- brewer.pal(n = 8, "Dark2")
imp <-as.data.frame(rf.mod$importance[order(rf.mod$importance),])
barplot(t(imp),col=color[1])
points(which(imp==imp['rand',]),0.6,col=color[2],type='h',lwd=2)
• Everything below importance of our "dummy" feature (random vector) can likely be ignored.
imp <- subset(imp, imp>imp['rand',])

15. • Saving the data in one rda-file for further analyses:
save('all', 'y', 'tst', 'imp', file = ‘./AllData.rda')
• Now reduce the feature set:
dat <- all[,rownames(imp)]
 Final Model
• First, estimate "sigma" (inverse width parameter for the RBF-SVM).
sigDist <- sigest(y ~ as.matrix(dat), data=dat, frac = 1)
• Creating a tune grid for further C-parameter selection):
svmTuneGrid <- data.frame(.sigma = sigDist[1], .C = 2^(-20:100))

16. • Training the final RBF-SVM model:
svmFit <- train(dat,y, method = "svmRadial",
preProc = c("center", "scale"), tuneGrid = svmTuneGrid,
trControl = trainControl(method = "cv", number = 86, classProbs = TRUE))
• Making predictions:
ttst <- tst[,rownames(imp)]
predTst <- predict(svmFit, ttst, type='prob')
predTst <- predTst[,2]
• Formatting submission:
pred <- cbind(as.integer(tst$Id), as.numeric(predTst))
colnames(pred) <- c('Id', 'Probability')
• Writing:
write.table(pred, file = ‘./submission.csv', sep=',', quote=F, row.names=F, fileEncoding = 'UTF-16LE')

17. Diagnosis:
 Two or more of the following must be
present:
1) Hallucinations.
2) Delusions.
3) Disorganized behavior or catatonic behavior.
4) Disorganized speech.
5) Negative symptoms
The symptoms must last at least for 6
months.

18. Treatment:
 Medications
 Antipsychotic Drugs
 Psychotherapy
 Hospitalization
 Electroconvulsive Therapy (ECT)

19. Conclusion:
 As per our observations the result we got
by comparing GP classifier and SVM
classifier, SVM gives accuracy greater
than GP classifier.

20. References
[1] "Schizophrenia Fact sheet N°397", WHO.
[2] "Schizophrenia". National Institute of Mental Health.
[3] Schizophrenia dataset from “kaggle.com/c/mlsp-2014-mri/data”
[4] “National Institute of Medical Health”
https://www.nimh.nih.gov/health/topics/schizophrenia/index.shtml
[5] “National Alliance on Mental Illness”
https://www.nami.org/Learn-More/Mental-Health-Conditions/Schizophrenia
[6] “Schizophrenia –Wikipedia” https://en.wikipedia.org/wiki/Schizophrenia
[7] http://www.mentalhealthamerica.net/conditions/schizophrenia

21. Thank you!!!