tutorsindia

1
MACHINE-LEARNING IN DEMENTIA INFORMATICS
RESEARCH
An Academic presentation by Dr. Nancy Agens,
Head, Technical Operations, Tutors India
Group www.tutorsindia.com Email
info_at_tutorsindia.com
2
Today's Discussion
OUTLINE
In Brief Introduction Methods Developing a
ML-based Model to Identify Dementia
Machine-Learning Models Limitations Conclusion
3
In Brief

• Machine Learning (ML) is the modern method used
  to predict, recognize, and to assess the disease
  correctly without the participation of
  humankind. ML is emerging rapidly in the field
  of medicine to diagnose the disease, to visualize
  the disease, and to examine the transmission of
  disease. Dementia is a chronic disease that
  affects millions of people worldwide.
• Machine learning is used in neuro-imaging data
  analysis for the treatment of dementia patients.

4
Introduction
The diagnosis usually includes set of clinical tes
ts like cognitive assessments, history of
patients, and neuro-imaging. This is a very
time-consuming process and also quite
expensive. Due to the progress and development
in the technologies, to advance in information
technology, the field of medical sciences has
created huge set of data related to this
disease. The main objective is to create a model
based on machine-learning that can be used to
predict Alzheimers disease, cognitive
impairment, and associated diseases of dementia.
5
Methods
A methodology was used to detect dementia
patients who are not diagnosed. This is done
with the help of read-encoded data that are
collected regularly in primary care. The
methodology is as follows A list of Read codes
that are related to the disease was gathered and
this data was used to detect dementia
patients. The dataset was investigated to
explore other Read codes that were allotted to
the patients who have dementia. Contd..
6
A division of Read codes was then established
that has an association with
dementia patients.
This division of codes are said to be the
read-encoded risk factors associated with
dementia patients. The dataset that was obtained
was then used to build a model based on machine-
learning to identify dementia patients. The
proposed model was then tested and assessed for
performance. The status of dementia patients
that have been predicted by the model was then
evaluated further. Contd..
7
Figure 1 Overview of Methodology
8
Developing a ML-based Model
to Identify Dementia
A classifier was derived using Machine-learning
that was used to characterize dementia patients,
and this classifier can also be used to detect
all possible cases associated with the
disease. The Read codes are used for building a
dementia classification model. The dataset
guides the classifier derived using
machine-learning to distinguish the patients who
have dementia and those who are healthy. These
classifiers would be influenced to recognize
healthy patients.
9
Machine-Learning Models
Support Vector Machine (SVM) is a
machine-learning model and is widely used for
pattern recognition and diagnosing the problems
of dementia from data. The Naïve Bayes (NB)
classifier is a machine-learning approach that
can be used to acquire probabilistic knowledge
to categorize unseen data. Random Forest (RF) is
an algorithm based on machine-learning that uses
data to construct decision trees (DTs), and
organize unseen data by merging each decision
trees. Contd..
10
Logistic Regression (LR) is an easy
machine-learning model that has been mostly used
in binary classification problems and for early
detection of the disease.
There are four criterions that can be used to
examine the performance of the machine-learning
classification specificity, accuracy,
sensitivity, and area under the curve
(AUC). After this preliminary assessment, the
model was checked on the primary care dataset to
decide undiagnosed dementia patients. Contd..
11
Figure 2 Number of Potentially Undiagnosed
Dementia Cases
12
Limitations
Dementia is one of the ill health problems that
have been a great challenge for the health
experts worldwide. Additionally, the disease
affects mostly older people above age of
60. This disease is worse that it can even cause
damage to brain and reduces the patients
ability to do their activities. Still, the
research is carried on to find a cure for this
disease from the past 2 decades
13
Conclusion
The healthcare practitioner determines the
accuracy and efficiency of the diagnosis. As
there is a lack of practitioners in some areas,
it is even more difficult for the diagnosis of
the disease. Machine-Learning helps in the
progress of the analysis of medical data, and
automatically make the decision for
diagnosis. The machine-learning field has become
very active recently that it uses variety of
patients data to discover new biomarkers for
diagnosis and to improve the diagnostic ability.
14
CONTACT US
UNITED KINGDOM 44-1143520021 INDIA 91-444813707
0 EMAIL info_at_tutorsindia.com