IAES International Journal of Artificial Intelligence (IJ-AI)

Vol. 14, No. 2, April 2025, pp. 1531~1539

ISSN: 2252-8938, DOI: 10.11591/ijai.v14.i2.pp1531-1539  1531

Primary phase Alzheimer's disease detection using ensemble

learning model

Priya Dasarwar1, Uma Yadav2, Nekita Chavhan3

1
Department of Computer Science and Engineering, Symbiosis Institute of Technology, Symbiosis International (Deemed University),
Nagpur Campus, Pune, India
2
Department of Computer Science and Engineering, Shri Ramdeobaba College of Engineering and Management, Ramdeobaba University,
Nagpur, India
3
Department of Data Science, IoT, and Cyber Security, G H Raisoni College of Engineering, Nagpur, India

Article Info ABSTRACT

Article history: Alzheimer's disease (AD) is a noteworthy problem for public health. Older
people are most impacted by this neurological disease. It leads to memory
Received Sep 9, 2024 loss and various cognitive impairments, eventually hindering
Revised Oct 21, 2024 communication. As a result, research on early AD detection has intensified
Accepted Nov 14, 2024 in recent years. In current research work, we propose an ensemble learning
strategy to identify AD by classifying brain images into two groups: AD
brain and normal brain. Researchers have recently explored various machine
Keywords: learning (ML) and deep learning techniques to improve early disease
detection. Patients with AD can recover from it more successfully and with
Alzheimer's disease less damage if they receive early diagnosis and therapy. This research
Convolutional neural network presents an ensemble learning model to predict AD using decision trees
Deep learning (DT), logistic regression (LR), support vector machines (SVM), and
Machine learning convolutional neural networks (CNN). The open access series of imaging
Support vector machine studies (OASIS) dataset is used for model training, and performance is
measured in terms of various kinds of outcome namely accuracy, precision,
recall, and F1 score. Our results demonstrated that, for the AD dataset, the
CNN achieved the maximum validation accuracy of 90.32%. Thus, by
accurately detecting the condition, ensemble algorithms can potentially
significantly reduce the annual mortality rates associated with AD.
This is an open access article under the CC BY-SA license.

Corresponding Author:
Priya Dasarwar
Department of Computer Science and Engineering, Symbiosis Institute of Technology
Symbiosis International (Deemed University), Nagpur Campus
Pune, India
Email: [email protected]

1. INTRODUCTION
Millions of people worldwide suffer from the debilitating condition known as Alzheimer's disease
(AD). It is a incurable condition of the brain illness which progressively deteriorates memory and thinking
skills, making it challenging for sufferers to carry out even the most basic tasks. The illness causes brain cell
destruction and death, which lowers cognitive function and ultimately impairs one's capacity to carry out
daily tasks. As a condition worsens, people may have confusion, memory loss, language difficulties, and
changes in their typical behavior [1]. This disease is currently diagnosed mostly by cognitive impairment
testing, which regrettably does not provide an accurate diagnosis until the patient has advanced past the stage
of moderate AD [2]. Detecting primary phase AD is crucial for timely intervention and improved patient
outcomes.

Journal homepage: http://ijai.iaescore.com

1532  ISSN: 2252-8938

Traditional diagnostic methods often struggle due to subtle symptoms and overlap with other
cognitive disorders. Ensemble learning models offer a promising solution by combining predictions from
multiple algorithms and leveraging diverse data sources, such as neurosurgical and genetic information, to
capture complex patterns associated with early-stage AD. This approach aims to enhance diagnostic accuracy
and generalization across populations, enabling more effective and personalized interventions in Alzheimer's
care [3]. Deep learning, in particular convolutional neural networks (CNN), has revolutionized the area of
medical imaging. This cutting-edge neural network designs are especially well suited for interpretation of
medical imaging data since they are built to automatically learn and extract hierarchical characteristics from
massive volumes of data [4]. Using data from medical imaging tests like magnetic resonance imaging (MRI)
scans, CNN have lately shown a lot of promise in diagnosing AD [5], [6]. CNN are the subset of deep
learning algorithm that works well for medical image analysis because they can automatically recognize and
extract information from images. In this research we investigate the following: i) to train the model with
various machine learning (ML) methods on the open access series of imaging studies (OASIS) dataset; ii) to
apply a cutting-edge CNN based method for AD identification; iii) to identify AD with high accuracy at an
early stage; and iv) to improve the standard of life for patient and their families.
Several approaches have been studied in the literature to help doctors diagnose this illness. A thorough
analysis of AD was provided by Kavitha et al. [7], who also investigated the impact of two ML techniques that
aid in the early detection of the illness. The author covered crucial study subjects like the data sets utilized and
the methods of evaluation performed. They offer a paradigm that facilitates a deeper comprehension of the state
of the field and draws attention to the obstacles and possibilities for creative and valuable study. The study
demonstrated the optimal ML technique for the Alzheimer's disease neuroimaging initiative (ADNI) dataset.
Using the suggested classification scheme by Diogo et al. [8], clinicians can identify these illnesses. By using
these ML algorithms for early diagnosis, annual death rates from AD can be significantly reduced. On the AD
test data, the suggested approach yields superior findings with the highest evaluated average accuracy.
Compared to previous studies, the accuracy score on this test is significantly higher. Cheung et al. [9] addressed
the limitations of previous research by proposing an ML-based diagnostic tool for moderate cognitive
impairment (MCI) along with AD. The tool is specifically multi-diagnostic and was trained as well as evaluated
using two separate data sources. It has multiple acquisition protocols, and tests generalization across datasets
and protocols. The proposed method is entirely new in the context of AD and is based on baseline scans and a
follow-up diagnosis, regardless of progression, has transparent performance reports, and analyzes potential
clinical applicability. An unsupervised domain adaption deep learning technique was used by Zaabi et al. [10] to
improve the generalization of a novel retinal photograph-based deep learning architecture for the primary
detection of AD cases. This algorithm was created, validated, and tested. When distinguishing between patients
suffering from AD-dementia along with those without the condition, the proposed deep learning algorithm
performed consistently and accurately. Alroobaea et al. [11] used the CNN and transfer learning as two distinct
approaches for AD detection. There are two primary processes in the suggested method: region of interest
extraction and classification. To find the area of the image that contains the hippocampal region of the brain, the
image is first divided into blocks. Researchers evaluated CNN and transfer learning techniques in the second
phase. The achieved results demonstrate that the categorization of images employing transfer learning produces
a better result rate when compared to CNN. Singh et al. [12] proposed the use of ML techniques for AD
identification. The ADNI along with OASIS datasets are used in the classification model evaluation process.
The findings of the experiment demonstrated that, while employing the ADNI dataset, algorithms namely
logistic regression (LR) along with support vector machines (SVM) yielded the best accuracy values, while LR
and random forests produced the best results when using the OASIS dataset. Zhao et al. [13] highlighted the
summary of current studies on deep learning algorithms for diagnosing AD. From the literature survey done, it
is observed that detecting early-stage Alzheimer’s with ML faces certain challenges like limited data, poor
generalization across populations, and the black-box nature of models, which reduce clinical trust. Also, early
symptoms are subtle and overlap with other conditions thereby complicating detection. Therefore, this research
includes deep learning technique which can automatically extracting complex patterns from large datasets and
hence can improve accuracy by reducing the need for manual feature selection. The main contribution of the
current research is an analysis of how well ML along with deep learning detect AD. The study focuses on using
ML and artificial intelligence to diagnose AD.

2. PROPOSED METHOD
2.1. Proposed methodology
An effective strategy has been developed to improve the detection of AD. The proposed method is
divided into following steps as presented in Figure 1. The objective of this methodology is to build predictive
models using LR, decision trees (DT), SVM, and CNN to categorize patients with a likelihood of AD with

Int J Artif Intell, Vol. 14, No. 2, April 2025: 1531-1539

Int J Artif Intell ISSN: 2252-8938  1533

respect to clinical, genetic as well as imaging data. The steps in the methodology includes collection of data,
preprocessing it, extracting features, training the model, evaluating it, and comparing it.

Figure 1. Proposed methodology architecture

2.1.1. Data collection

In this research OASIS dataset is utilized which seeks to prepare MRI datasets of brain publicly available
for future research. There are three kinds of datasets available: OASIS-1 (cross-sectional) MRI data in young,
middle aged, non-demented and demented older adults, OASIS-2 (longitudinal) MRI data in non-demented and
demented older adults whereas OASIS-3 is an expansion of earlier dataset collections [14]. We have used OASIS-
2 and OASIS-3 datasets for the detection of Alzheimer. On these datasets, we apply both classical ML techniques
(LR, SVM, DT) and modern deep learning (CNN) for effective and comprehensive Alzheimer’s prediction.

2.1.2. Data pre-processing

Data preprocessing requires number of stages to develop the dataset before modeling. Handling
missing data is addressed using statistical techniques like mean, mode, and median imputation. Feature scaling
is applied to normalize or standardize continuous variables such as age, clinical scores, and genetic markers to
ensure consistency across models. To enhance model performance, data augmentation methods like random
rotations are applied to MRI/positron emission tomography (PET) images, reducing over-fitting, and improving
generalization in CNN models. Dimensionality reduction using method namely principal component analysis
(PCA) reduce the feature space for structured data, mitigating the curse of dimensionality, while CNN-based
feature extraction is employed to automatically capture relevant features from image data.

2.1.3. Model selection

Four predictive models are developed for AD classification. LR serves as a simple linear model to
classify binary outcomes (Alzheimer's vs. non-Alzheimer's) using clinical, demographic, and genetic
markers, with L1 or L2 regularization to prevent over-fitting. SVM is used with linear as well as radial basis
function (RBF) kernels to classify Alzheimer’s risk based on clinical, demographic, and genetic data, and
hyperparameter tuning will be done through grid search to optimize kernel parameters (C, gamma). CNN
focuses on analyzing MRI or PET scan data, leveraging convolutional along-with max-pooling layers for
feature extraction whereas fully connected layers to classify the data, with consideration of pretrained models
namely ResNet for transfer learning and optimization via the Adam optimizer with learning rate scheduling.
Lastly, DT models split data based on demographic, clinical, and genetic features, with cost-complexity
pruning to prevent over-fitting and hyperparameter tuning (e.g., tree depth, minimum samples split, and leaf
nodes) via cross-validation to refine model performance.

2.1.4. Model training

The training strategy begins with an 80:20 train-test splitting, where 80% associated with the dataset
is utilized for training and 20% dataset is applied for testing during initial model development. To ensure
robust model assessment, K-fold cross-validation (where K=5) is applied. In cases of class imbalance
between Alzheimer's and non-Alzheimer’s samples, technique namely synthetic minority over-sampling
technique (SMOTE) is utilized to address this issue. For hyperparameter tuning, grid search method is
combined with cross-validation to optimize the performance of each model.

2.1.5. Model evaluation

The proposed model will be evaluated employing several parameters to assess their performance.
Accuracy will provide an overall measure of model performance, while precision and recall will specifically
Primary phase Alzheimer's disease detection using ensemble learning model (Priya Dasarwar)
1534  ISSN: 2252-8938

evaluate the model’s capacity to correctly recognize true positives. The F1-score will be used to balance
precision and recall, which is particularly important in imbalanced datasets.

2.1.6. Comparison of models

Following model evaluation, the effectiveness of DT, SVM, CNN, and LR models are compared using
metrics comprising accuracy, precision, recall, and F1-score [15]. Beyond these indicators, factors like model
complexity (e.g., CNNs are more complex compared to LR), interpretability (e.g., DT and LR are easier to
interpret), and computational resource requirements (e.g., CNNs typically require more computational power
and memory) are considered in selecting the most suitable model for predicting AD. Furthermore, the ability of
each model to withstand overfitting and its generalizability to unknown data are also evaluated to ensure reliable
performance in predicting AD.

2.2. Description of dataset

We will employ longitudinal MRI data from OASIS-2. There are 150 people in the sample having
longitudinal MRI data, ranging in age from 60 to 96. Every patient underwent a minimum of one scan [16], [17].
Each individual has a right hand. 72 of the people underwent an investigation and were labeled as "nondemented".
64 of the participants in the study were initially diagnosed as "demented" at the time of their visits. 14 individuals
were first diagnosed as "non-demented," however during a follow-up visit, they were categorized as "demented".
These belong under the category of "converted". Table 1 shows the list of features from the dataset.
Figure 2 shows the dataset outcome for the different dementia stages. As seen in Figures 3 and 4, the
brain volume ratio of the non-demented category is greater than that of the demented category. This is
commonly believed to be the result of diseases that cause the tissue in the brain to shrink. When comparing
the demented patient category to the non-demented patient category, there is a greater age distribution of
70-80 years old. We assume that individuals with such an illness have a lesser chance of living, with only a
few ninety years remaining. Also, it illustrates that the non-demented category's mini-mental state
examination (MMSE) scores were significantly more than those of the demented category. Figure 5 shows
that the category with dementia has a slightly higher estimated total intracranial volume. Figure 6 shows that
the category with dementia has a higher normalize whole brain volume. Figure 7 depicts that demented
individual have longer lifespans than non-demented individuals.

Table 1. Qualities of the information

Attribute Description
EDUC Years of education
SES Socioeconomic Status
MMSE Mini mental state examination
CDR Clinical dementia rating
eTIV Estimated total intracranial volume
nWBV Normalize whole brain volume
ASF Atlas scaling factor

Figure 2. Dataset outcome of different dementia stages [7]

Int J Artif Intell, Vol. 14, No. 2, April 2025: 1531-1539

Int J Artif Intell ISSN: 2252-8938  1535

Figure 3. MMSE

Figure 4. Atlas (volume) scaling factor

Figure 5. Calculated total intracranial volume

Figure 6. Normalize whole brain volume

Primary phase Alzheimer's disease detection using ensemble learning model (Priya Dasarwar)
1536  ISSN: 2252-8938

Figure 7. Age in years

2.3. Algorithms utilized for Alzheimer detection

The process of classifying photographs involves using a classifier to divide the various objects into
discrete groups based on multiple classification criteria. In this instance, there are two classes: brains without
AD and brains with AD. Thus, four classification methods LR, DT, SVM, and CNN are used to complete the
categorization procedure once the blocks have been retrieved in order to compare their classification rates.

2.3.1. Logistic regression

LR is employed for binary classification issues and estimates the likelihood that a parameter belongs
to a specific group of data by applying the logistic (sigmoid) function. This function maps predicted values to
a probability between 0 and 1. The output probability is then compared to threshold to make a binary
decision, determining the class of the input based on whether the probability exceeds a certain threshold [18].
The model is represented in (1).
1 1
𝑃(𝑌 − 𝑋 ) = 1 + 𝑒 −(𝛽0+𝛽1 𝑋) (1)

Where β0 and β1 are the coefficients learned during training and X is the input feature. These coefficients are
learned using a technique like gradient descent that minimizes a cost function, typically the log-loss
(cross-entropy loss) [19].

2.3.2. Support vector machine

Although this technique is mostly employed for classification, SVM is applied for regression jobs as
well. It functions by locating the hyperplane that separates the data into the greatest number of classes.
This hyperplane is a line in two dimensions and turns into a plane in higher dimensions. In order to provide
the best feasible separation between the classes, SVM learning seeks to optimize the difference among this
hyperplane and the closest data points from every group [20]. For a given input X and weights w,
the decision function is given in (2).

𝑓(𝑋) = 𝑤. 𝑋 + 𝑏 (2)

Where b represents the bias term. The hyperplane is chosen so as to optimize the distance between it and the
closest data points, or support vectors.
To find the ideal hyperplane that divides the data for training a SVM, quadratic optimization
techniques are applied. The kernel approach is used to translate non-linear data into a higher-dimensional
space in which a linear separation is possible. This effectively converts the situation into one where linear
classification techniques may be used, enabling SVM to handle complex relationships between features [21].

2.3.3. Decision tree

A DT is used to create a tree-like model by separating the information among subsets according to
feature values. It can be applied to jobs involving both regression as well as classification.
An internal node indicates a feature, a branch expresses a decision rule, and a leaf node indicates a result or
class label. In order to attain the most information gain or impurity reduction, the method uses metrics like
mean squared error for regression and Gini impurity or entropy for classification to assess the appropriate
split at each node during training. Recursively building the tree results in a model that bases decisions on the
feature splits' hierarchical structure [22], [23].

Int J Artif Intell, Vol. 14, No. 2, April 2025: 1531-1539

Int J Artif Intell ISSN: 2252-8938  1537

2.3.4. Convolutional neural networks

While its major applications are in the field of image processing along with computer vision, CNNs
can also be employed in other fields, including as time-series data. CNNs work by employing convolutional
layers to intelligently acquire spatial hierarchies of characteristics from the input data, creating feature maps
through the application of convolutional filters. They lower the dimensionality of these feature maps by using
pooling layers like max pooling and non-linear activation functions like rectified linear unit (ReLU) to
introduce non-linearity. Predictions are based on the learnt features by fully linked layers at the end of the
network. Stochastic gradient descent and back-propagation are used in conjunction with end-to-end training
to optimize the CNN's performance by modifying its weights [24], [25].

3. RESULTS AND DISCUSSION

In this study, we assess a model to diagnose Alzheimer's using the patients' MRI images. A variety
of metrics, including specificity, F1-score, precision, accuracy as well as recall, are used to evaluate the
model [24], [25]. We divide our categorization rate into two distinct groups (normal range brain along with
brain having AD) with respect to the mentioned standards as (3).

(𝑁𝑜.𝑜𝑓 𝑤𝑒𝑙𝑙 𝑐𝑙𝑎𝑠𝑠𝑖𝑓𝑖𝑒𝑑 𝑖𝑚𝑎𝑔𝑒𝑠)

𝐶𝑙𝑎𝑠𝑠𝑖𝑓𝑖𝑐𝑎𝑡𝑖𝑜𝑛 𝑟𝑎𝑡𝑒 = ∗ 100 (3)
(𝑡𝑜𝑡𝑎𝑙 𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑖𝑚𝑎𝑔𝑒𝑠)

As demonstrated in Table 2, we evaluate our suggested method's performance against a few cutting-edge
techniques. The comparative analysis and the accuracy given are predicated on a testing dataset. The results
shows that the accuracy using the CNN (90.32%) is better than the accuracy using the remaining algorithms.
This shows the effectiveness of CNN algorithm. This increases the models' accuracy and decreases the
amount of time needed for learning. Thus, greater precision with less time and data. Figure 8 displays an
illustration of the accuracy, precision, recall, and F1-score attained by several classifiers.

Table 2. The comparative analysis of all algorithms

Algorithms Accuracy (%) Precision (%) Recall (%) F1-score (%)
CNN 90.32 90.54 91.06 49.40
SVM 81.57 82.22 70 37.80
LR 78.94 79.44 70 37.21
DT 81.56 82.50 65 36.35

Figure 8. Performance of algorithms

4. CONCLUSION
In our pursuit of advancing healthcare through intelligent machines, our research has underscored a
promising approach to enhancing sickness prognosis, with a particular focus on AD. We employed a variety
of ML along with deep learning methods in order to analyze patient data, considering a diverse array of
Primary phase Alzheimer's disease detection using ensemble learning model (Priya Dasarwar)
1538  ISSN: 2252-8938

symptoms and conditions to refine our predictive models. Our investigation yielded compelling results,
particularly with the CNN algorithm. This proposed model exhibited an exceptional accuracy rate of 90.32%
in predicting AD cases. This impressive performance underscores the CNN's capacity to effectively
categorize and identify early-phase Alzheimer's, marking it as a standout approach in our study. The future
research on ensemble learning for early Alzheimer's detection should improve the interpretability of the
models, applicability to different populations, and quality of the data. Ensemble approaches have the
potential to enhance diagnostic precision and include multi-modal data by merging various algorithms. It will
be possible to implement more precise, dependable, and secure models in healthcare settings by creating
explainable models and implementing privacy-preserving strategies. Also, future research should concentrate
on growing the dataset to encompass a wider demographic and looking into the use of more sophisticated
algorithms and hybrid models.

REFERENCES
[1] S. Liu et al., “Generalizable deep learning model for early alzheimer’s disease detection from structural MRIs,” Scientific Reports,
vol. 12, no. 1, Oct. 2022, doi: 10.1038/s41598-022-20674-x.
[2] S. Savaş, “Detecting the stages of alzheimer’s disease with pre-trained deep learning architectures,” Arabian Journal for Science
and Engineering, vol. 47, no. 2, pp. 2201–2218, Feb. 2022, doi: 10.1007/s13369-021-06131-3.
[3] D. AlSaeed and S. F. Omar, “Brain MRI analysis for alzheimer’s disease diagnosis using CNN-based feature extraction and
machine learning,” Sensors, vol. 22, no. 8, Apr. 2022, doi: 10.3390/s22082911.
[4] A. G. Vrahatis, K. Skolariki, M. G. Krokidis, K. Lazaros, T. P. Exarchos, and P. Vlamos, “Revolutionizing the early detection of
alzheimer’s disease through non-invasive biomarkers: the role of artificial intelligence and deep learning,” Sensors, vol. 23, no. 9,
Apr. 2023, doi: 10.3390/s23094184.
[5] Z. Zhao et al., “Conventional machine learning and deep learning in alzheimer’s disease diagnosis using neuroimaging: a review,”
Frontiers in Computational Neuroscience, vol. 17, Feb. 2023, doi: 10.3389/fncom.2023.1038636.
[6] V. Patil, M. Madgi, and A. Kiran, “Early prediction of alzheimer’s disease using convolutional neural network: a review,” The
Egyptian Journal of Neurology, Psychiatry and Neurosurgery, vol. 58, no. 1, pp. 1-10, Nov. 2022, doi: 10.1186/s41983-022-
00571-w.
[7] C. Kavitha, V. Mani, S. R. Srividhya, O. I. Khalaf, and C. A. T. Romero, “Early-stage alzheimer’s disease prediction using
machine learning models,” Frontiers in Public Health, vol. 10, Mar. 2022, doi: 10.3389/fpubh.2022.853294.
[8] V. S. Diogo, H. A. Ferreira, and D. Prata, “Early diagnosis of alzheimer’s disease using machine learning: a multi-diagnostic,
generalizable approach,” Alzheimer’s Research & Therapy, vol. 14, no. 1, Dec. 2022, doi: 10.1186/s13195-022-01047-y.
[9] C. Y. Cheung et al., “A deep learning model for detection of alzheimer’s disease based on retinal photographs: a retrospective,
multicentre case-control study,” The Lancet Digital Health, vol. 4, no. 11, pp. e806–e815, Nov. 2022, doi: 10.1016/S2589-
7500(22)00169-8.
[10] M. Zaabi, N. Smaoui, H. Derbel, and W. Hariri, “Alzheimer’s disease detection using convolutional neural networks and transfer
learning based methods,” in 2020 17th International Multi-Conference on Systems, Signals & Devices (SSD), Monastir, Tunisia,
Jul. 2020, pp. 939–943, doi: 10.1109/SSD49366.2020.9364155.
[11] R. Alroobaea et al., “Alzheimer’s disease early detection using machine learning techniques,” Research Square, pp. 1-16, Jun.
2021, doi: 10.21203/rs.3.rs-624520/v1.
[12] N. Singh, P. D, N. Soni, and A. Kapoor, “Automated detection of alzheimer disease using mri images and deep neural networks- a
review,” arXiv-Electrical Engineering and Systems Science, pp. 1-22, Sep. 2022.
[13] X. Zhao, C. K. E. Ang, U. R. Acharya, and K. H. Cheong, “Application of artificial intelligence techniques for the detection of
alzheimer’s disease using structural mri images,” Biocybernetics and Biomedical Engineering, vol. 41, no. 2, pp. 456–473, Apr.
2021, doi: 10.1016/j.bbe.2021.02.006.
[14] N. B. Balakrishnan, S. P S, and J. J. Panackal, “Alzheimer’s disease diagnosis using machine learning: a review,” International
Journal of Engineering Trends and Technology, vol. 71, no. 3, pp. 120–129, Mar. 2022, doi: 10.14445/22315381/IJETT-
V71I3P213.
[15] A. Sharma, S. Kaur, N. Memon, A. J. Fathima, S. Ray, and M. W. Bhatt, “Alzheimer’s patients detection using support vector
machine (SVM) with quantitative analysis,” Neuroscience Informatics, vol. 1, no. 3, Nov. 2021, doi: 10.1016/j.neuri.2021.100012.
[16] Q. Li and M. Q. Yang, “Comparison of machine learning approaches for enhancing alzheimer’s disease classification,” PeerJ,
vol. 9, Feb. 2021, doi: 10.7717/peerj.10549.
[17] M. EL-Geneedy, H. E.-D. Moustafa, F. Khalifa, H. Khater, and E. AbdElhalim, “An MRI-based deep learning approach for
accurate detection of alzheimer’s disease,” Alexandria Engineering Journal, vol. 63, pp. 211–221, Jan. 2023, doi:
10.1016/j.aej.2022.07.062.
[18] Z. Li, X. Jiang, Y. Wang, and Y. Kim, “Applied machine learning in alzheimer’s disease research: omics, imaging, and clinical
data,” Emerging Topics in Life Sciences, vol. 5, no. 6, pp. 765–777, Dec. 2021, doi: 10.1042/ETLS20210249.
[19] A. Shukla, R. Tiwari, and S. Tiwari, “Review on alzheimer disease detection methods: automatic pipelines and machine learning
techniques,” Sci, vol. 5, no. 1, Mar. 2023, doi: 10.3390/sci5010013.
[20] S. Harika, T. Yamini, T. Nagasaikamesh, S. H. Basha, S. S. Kumar, and M. S. S. DurgaKameswari, “Alzheimers disease detection
using different machine learning algorithms,” International Journal for Research in Applied Science and Engineering
Technology, vol. 10, no. 10, pp. 62–66, Oct. 2022, doi: 10.22214/ijraset.2022.46937.
[21] K. D. Silva and H. Kunz, “Prediction of alzheimer’s disease from magnetic resonance imaging using a convolutional neural
network,” Intelligence-Based Medicine, vol. 7, 2023, doi: 10.1016/j.ibmed.2023.100091.
[22] C. S. Eke, E. Jammeh, X. Li, C. Carroll, S. Pearson, and E. Ifeachor, “Early detection of Alzheimer's disease with blood plasma
proteins using support vector machines,” IEEE Journal of Biomedical and Health Informatics, vol. 25, no. 1, pp. 218-226, Jan.
2021, doi: 10.1109/JBHI.2020.2984355.
[23] M. Orouskhani, C. Zhu, S. Rostamian, F. Shomal Zadeh, M. Shafiei, and Y. Orouskhani, “Alzheimer’s disease detection from
structural mri using conditional deep triplet network,” Neuroscience Informatics, vol. 2, no. 4, Dec. 2022, doi:
10.1016/j.neuri.2022.100066.

Int J Artif Intell, Vol. 14, No. 2, April 2025: 1531-1539

Int J Artif Intell ISSN: 2252-8938  1539

[24] M. S. Bhargavi and B. Prabhakar, “Deep learning approaches for early detection of alzheimer’s disease using mri neuroimaging,”
in 2022 International Conference on Connected Systems & Intelligence (CSI), IEEE, Aug. 2022, pp. 1–6, doi:
10.1109/CSI54720.2022.9924058.
[25] H. A. Helaly, M. Badawy, and A. Y. Haikal, “Deep learning approach for early detection of alzheimer’s disease,” Cognitive
Computation, vol. 14, no. 5, pp. 1711–1727, Sep. 2022, doi: 10.1007/s12559-021-09946-2.

BIOGRAPHIES OF AUTHORS

Dr. Priya Dasarwar (Maidamwar) holds a doctoral degree in the area of

network security from G H Raisoni University, Amravati. She is currently working as an
Assistant Professor at Symbiosis Institute of Technology, Nagpur, India. She has also received
her B.Tech. and M.Tech. degrees with distinction from Nagpur University in 2010 and 2013,
respectively. She has published more than 25 research papers in reputable journals and at
national and international conferences. With over 13 years of experience, her research focuses
on machine learning, IoT, and network security. She can be contacted at email:
[email protected].

Dr. Uma Yadav is a distinguished Assistant Professor at Shri Ramdeobaba

College of Engineering and Management, Ramdeobaba University, Nagpur. She completed
her B.Tech. and M.Tech. degrees with distinction from Nagpur University in 2009 and 2013,
respectively, and recently earned her Ph.D. from GHRU Amravati. With over 14 years of
experience, her research interests include data science, machine learning, cloud computing, big
data management, and data mining. She has made significant contributions to her field,
authoring or co-authoring more than 20 research papers in reputed refereed journals and
conferences. She can be contacted at email: [email protected] or [email protected].

Prof. Nekita Chavhan received a Master of Engineering (M.E.) in wireless

communication and computing (CSE) from G H Raisoni College of Engineering in Nagpur,
Maharashtra, India. She is pursuing her doctoral degree from G H Raisoni University,
Amravati. She has authored 136 research papers published in international journals and
conferences. Currently, she serves as the Head of the Department in Data Science, IoT, and
Cyber Security (DIC) at G H Raisoni College of Engineering, Nagpur, India. Her research
areas encompass machine and deep learning, network security, Ad-hoc wireless networks,
wireless sensor networks, and mobile technology. She can be contacted at email:
[email protected].

Primary phase Alzheimer's disease detection using ensemble learning model (Priya Dasarwar)