EUVIMEDEUVIMED entdecken
Crossref

Exploring Machine Learning Techniques for Early Detection of Alzheimer’s Disease Using Multimodal Data

Shaurya Singhal, Monika Sen Chaudhary

IOSR Journal of Computer Engineering · 2026-03 · 28 (3) · 49-62

DOI: 10.9790/0661-2803034962 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IOSR Journal of Computer Engineering, 2026-03. Verfasst von Shaurya Singhal und 1 weiteren Autor:innen.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Background

Alzheimer’s disease (AD) is a progressive neurodegenerative disorder in which pathological changes can occur years before clinical symptoms become apparent. Early detection is therefore important for timely intervention. Current diagnostic methods include cognitive assessments, neuroimaging, and biomarker analysis, but these approaches may not reliably identify the disease during its earliest stages and can be expensive or difficult to implement for large-scale screening. Machine learning (ML) has increasingly been applied to this problem because it can identify patterns within complex biomedical datasets. Materials and Methods: This paper presents a structured review and comparative analysis of machine learning techniques applied to early AD detection using multimodal data. Studies were systematically analysed based on input modalities, including MRI, PET, EEG, CSF biomarkers, genetic data, and clinical scores. Model architectures evaluated included classical machine learning, deep learning, ensemble, and hybrid models. Datasets and reported evaluation metrics were also comparatively assessed to identify performance trends across approaches.

Results

The findings indicate consistent performance differences between single-modality and multimodal approaches. Multimodal fusion and longitudinal modelling demonstrated superior sensitivity for detecting early disease stages. Deep learning models, particularly convolutional and hybrid architectures, generally outperformed classical methods on high-dimensional imaging datasets. However, classical machine learning models showed greater interpretability and stability when applied to structured clinical data.

Conclusion

Despite promising diagnostic performance, significant challenges remain, including small and homogeneous cohorts, incomplete multimodal datasets, limited model interpretability, and inconsistent definitions of disease stages across studies. This review highlights the importance of standardized evaluation protocols, larger longitudinal datasets, and explainable multimodal models to support reliable clinical translation. Overall, the findings suggest that multimodal deep learning frameworks represent the most promising direction for nextgeneration AD diagnostic systems, provided current methodological and clinical limitations are effectively addressed.

Zitieren

APA

Singhal, S., & Chaudhary, M. S. (2026). Exploring Machine Learning Techniques for Early Detection of Alzheimer’s Disease Using Multimodal Data. IOSR Journal of Computer Engineering, 28(3), 49-62. https://doi.org/10.9790/0661-2803034962

Vancouver

Singhal S, Chaudhary MS. Exploring Machine Learning Techniques for Early Detection of Alzheimer’s Disease Using Multimodal Data. IOSR Journal of Computer Engineering. 2026;28(3):49-62. doi:10.9790/0661-2803034962

BibTeX
@article{Singhal2026,
  author = {Singhal, Shaurya and Chaudhary, Monika Sen},
  title = {Exploring Machine Learning Techniques for Early Detection of Alzheimer’s Disease Using Multimodal Data},
  journal = {IOSR Journal of Computer Engineering},
  year = {2026},
  volume = {28},
  number = {3},
  pages = {49--62},
  doi = {10.9790/0661-2803034962}
}
Kostenlos starten →