AI-Tool for Early-Stage Dementia Detection

Abstract

Now a days we are hearing a lot about the disease named dementia and we see from the fact that cognitive degeneration in people that have dementia progresses very slow and goes almost unobserved. To health care providers this means solving a complex puzzle of cognitive tests and brain scans. We see that to close this gap in diagnosis we have developed a project for an AI Powered Diagnostic System to put forth a smart and dependable second opinion for health care professionals. This system is to put together the big picture and give a full assessment of a patient\'s neurological health by using two different kinds of AI. The system will do this by using a combination of a Machine Learning ensemble of algorithms, acting as a Voting Classifier consisting of Logistic Regression, Random Forest, and XGBoost to assess everyday patient metrics such as memory test scores, age, and health history. Additionally, a ResNet50 Deep Learning model will be used to function as an expert set of eyes and specifically be trained to spot hidden and microscopic structural changes in MRI brain scans. This system has been developed with a user-friendly interface using React, Flask, and Python and instantly integrates these two types of data into a cohesive and high-precision risk profile to allow clinicians to visualize a patient\'s neurological health in an instant. Ultimately, this platform eliminates uncertainty in screening and allows doctors to make quick and informed decisions to ensure their patients get the care they need.

Introduction

The text describes an AI-powered system designed for early diagnosis and monitoring of dementia, addressing the limitations of traditional diagnostic methods that often fail to detect early cognitive decline. Conventional tools like manual cognitive tests and visual MRI analysis are limited in scalability and accuracy, leading to late diagnosis when treatment options are less effective.

To solve this, the proposed system uses artificial intelligence and machine learning to analyze both clinical data (such as cognitive test scores and patient history) and neuroimaging data (MRI scans). It combines multiple models, including a voting classifier (Logistic Regression, Random Forest, and XGBoost) and a deep learning model (ResNet50), to improve diagnostic accuracy. The system also generates consolidated reports, risk assessments, and progress tracking to support clinical decision-making and early intervention.

The literature review highlights that earlier research shows strong potential for AI in dementia detection, especially using machine learning for clinical data analysis and deep learning for MRI interpretation. However, most existing approaches focus on single data types (either clinical or imaging), and lack integrated, user-friendly systems. This motivates the need for a multimodal approach that combines different data sources for more reliable diagnosis.

The proposed system integrates these methods into a unified platform, including clinical analysis, MRI evaluation, dataset-based comparisons (using datasets like OASIS), and predictive risk assessment. It is designed to act as a “second opinion” tool for doctors, improving diagnostic confidence and supporting early intervention.

The system architecture follows an end-to-end pipeline: MRI images are preprocessed, features are extracted using ResNet50, and predictions are made using an ensemble of machine learning models. A frontend interface allows users to interact with the system, while backend APIs handle data processing and AI integration. A database stores medical and reference data for comparison and training support.

Conclusion

The AI-based dementia diagnostic platform was presented in this paper. The platform was developed to assist healthcare professionals in the early-stage dementia screening process. The platform was developed by integrating various diagnostic tools such as automated clinical metric analysis, structural MRI evaluation via deep learning algorithms, and OTP-based session management. The platform was developed using modern web technologies and multi-modal artificial intelligence. The platform provides data-based \"second opinions\" that assist healthcare professionals in the detection of early signs of neuro degeneration with greater accuracy. The implementation and testing of the platform proves that the AI-based platform performs the functions for which it was developed. The clinical assessment tool provides highly accurate classifications via the analysis of cognitive scores such as MMSE and CDR, and patient demographics via an ensemble-based machine learning approach. Concurrently, the platform’s neuroimaging tool uses the ResNet50 deep learning architecture for the detection of structural brain atrophy in the patient’s MRI scan. The diagnostic tool dashboard provides the healthcare professional with the ability to organize the patient’s data and obtain immediate feedback regarding the patient’s neurological condition. The platform bridges the gap between clinical observation and computational diagnostics. The platform provides an intelligent and interactive environment that assists the medical practitioners in providing precise diagnostic insights and facilitates the improvement of diagnostic accuracy. The AI-based platform provides precise diagnostic insights and facilitates the improvement of diagnostic accuracy. The platform assists the medical practitioners in the diagnosis and treatment of Dementia and various other disorders.

References

[1] Bayahya et al., \"multi-model deep learning for dementia detection,\" Front. Neurosis., vol. 19, 2025. Available: https://www.frontiersin.org/articles/10.3389/fnins.2025.1638022/full [2] A. Ksibi et al., \"Multimodal Siamese networks for dementia detection from speech patterns,\" Sci. Rep., vol. 15, 2025.Available: https://www.nature.com/articles/s41598-025-13902-7 [3] M. Al-Hammadi et al., \"Machine Learning Approaches for Dementia Detection,\" J. Alzheimer’s Dis., vol. 98, 2024. Available: https://journals.sagepub.com/doi/10.3233/JAD-231459 [4] A. Javeed, A. Dallora, U. Berglund, L. Ali, and P. Anderberg, \"Machine learning for dementia prediction: a systematic review and future research directions,\" J. Med. Syst., vol. 47, no. 6, 2023. [5] S. Grueso and R. Viejo-Sobera, \"Machine learning methods for predicting progression from mild cognitive impairment to Alzheimer’s disease dementia: a systematic review,\" Alzheimer’s Res. Ther., vol. 13, no. 162, 2021. [6] S. Basheer, S. Bhatia, and S. B. Sakri, \"Improving Alzheimer’s prediction using a novel capsule-network-based framework,\" in Proc. IEEE, 2021, Art. no. 9380278. [7] G. Battineni, N. Chintalapudi, and F. Amenta, \"Machine learning in medicine: Performance calculation of dementia prediction by support vector machines (SVM),\" Informat. Med. Unlocked, vol. 15, p. 100200, 2019. [8] A. Hashmi and O. Barukab, \"Dementia Classification Using Deep Reinforcement Learning for Early Diagnosis,\" Appl. Sci., vol. 13, no. 3, p. 1464, 2023. [9] A. So, D. Hooshyar, K.-W. Park, and H. S. Lim, \"Early diagnosis of dementia from clinical data by machine learning techniques,\" Appl. Sci., vol. 7, no. 7, p. 651, 2017. [10] M. R. Kumar et al., \"Dementia Detection from Speech Using Machine Learning and Deep Learning Architectures,\" Sensors, vol. 22, no. 23, p. 9311, 2022. [11] Z. Jahan et al., \"Early dementia detection with speech analysis and machine learning techniques,\" Discov. Sustain., vol.5,2024.Available: https://link.springer.com/article/10.1007/s43621-024-00217-2 [12] T. Ahmed et al., \"Detection of Dementia: Using Electroencephalography and Machine Learning,\" Open Bioinformat. J., vol. 17, 2024. [13] D. Ortiz-Perez et al., \"A Deep Learning-Based Multimodal Architecture to Predict Dementia,\" Neurocomputing, vol. 540, 2023. [14] K. AlHarkan et al., \"Artificial intelligence approaches for early detection of neurocognitive disorders among older adults,\" Front. Comput. Neurosci., vol. 18, p. 1307305, 2024

Copyright

Copyright © 2026 Mrs. Sumayya Samreen , Nyalati Sreekar, Mohammad Nawaz Khan, Agurla Harish. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.