CoChat

A high-generalizability machine learning framework for predicting the progression of Alzheimer’s disease using limited data

Open access
Wang, Caihua, Li, Yuanzhong, Tsuboshita, Yukihiro, Sakurai, Takuya, Goto, Tsubasa, Yamaguchi, Hiroyuki, Yamashita, Yuichi, Sekiguchi, Atsushi, Tachimori, Hisateru, for the Alzheimer’s Disease Neuroimaging Initiative, Wang, Caihua, Li, Yuanzhong, Goto, Tsubasa · 2022 · npj Digital Medicine

The DOI matches its Crossref record; 55 of the 55 references it cites were found in a registry or an index.

Abstract

Alzheimer’s disease is a neurodegenerative disease that imposes a substantial financial burden on society. A number of machine learning studies have been conducted to predict the speed of its progression, which varies widely among different individuals, for recruiting fast progressors in future clinical trials. However, because the data in this field are very limited, two problems have yet to be solved: the first is that models built on limited data tend to induce overfitting and have low generalizability, and the second is that no cross-cohort evaluations have been done. Here, to suppress the overfitting caused by limited data, we propose a hybrid machine learning framework consisting of multiple convolutional neural networks that automatically extract image features from the point of view of brain segments, which are relevant to cognitive decline according to clinical findings, and a linear support vector classifier that uses extracted image features together with non-image information to make robust final predictions. The experimental results indicate that our model achieves superior performance (accuracy: 0.88, area under the curve [AUC]: 0.95) compared with other state-of-the-art methods. Moreover, our framework demonstrates high generalizability as a result of evaluations using a completely different cohort dataset (accuracy: 0.84, AUC: 0.91) collected from a different population than that used for training.

Overview

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1038/s41746-022-00577-x

Cites

55 of 55 references verified0 citation errors · 0 not checkable

Statistics

Figures

Fig. 1 A machine learning framework for predicting the progression of Alzheimer’s disease. Multiple CNN-based image feature extractors with the same structure were trained on each brain segment, that is, the hippocampus and anterior temporal lobe in our study, to extract image features automatically. The CNN-based image feature extractor consists of a multi-layer convolutional neural network (CNN) block, a self-attention (SA) block, and an auto-encoder (AE) block. The feature vectors after the global average pooling (GAP) layer that encodes image features are passed to the PCA for dimensionality reduction. Finally, a linear SVM uses the low dimensional image features together with non-image information to make a final prediction.
Fig. 1 A machine learning framework for predicting the progression of Alzheimer’s disease. Multiple CNN-based image feature extractors with the same structure were trained on each brain segment, that is, the hippocampus and anterior temporal lobe in our study, to extract image features automatically. The CNN-based image feature extractor consists of a multi-layer convolutional neural network (CNN) block, a self-attention (SA) block, and an auto-encoder (AE) block. The feature vectors after the global average pooling (GAP) layer that encodes image features are passed to the PCA for dimensionality reduction. Finally, a linear SVM uses the low dimensional image features together with non-image information to make a final prediction.CC BY · PubMed Central
Fig. 2 Training, validation, and test datasets. We extracted data for training, validation, and test from the longitudinal datasets ADNI1, ADNI-GO, and ADNI2. If a participant was diagnosed as MCI and both the MRI T1-weighted images and other non-image information used in our model were available, we selected the related data as a candidate for this study. We evaluated our model in the following two ways: first, as in previous studies, we tested the model trained based on the NA-ADNI dataset using repeated 10-fold cross-validation and test; in each iteration, 80% of the samples were used for training, 10% for validation, and 10% for test; second, to evaluate generalization across cohorts, we tested the model, which was trained based on the ADNI dataset using a totally “unknown” J-ADNI dataset.
Fig. 2 Training, validation, and test datasets. We extracted data for training, validation, and test from the longitudinal datasets ADNI1, ADNI-GO, and ADNI2. If a participant was diagnosed as MCI and both the MRI T1-weighted images and other non-image information used in our model were available, we selected the related data as a candidate for this study. We evaluated our model in the following two ways: first, as in previous studies, we tested the model trained based on the NA-ADNI dataset using repeated 10-fold cross-validation and test; in each iteration, 80% of the samples were used for training, 10% for validation, and 10% for test; second, to evaluate generalization across cohorts, we tested the model, which was trained based on the ADNI dataset using a totally “unknown” J-ADNI dataset.CC BY · PubMed Central
Fig. 3 Cross-cohort evaluation results. ROC curves for both the NA-ADNI and J-ADNI are shown. The AUCs of the NA-ADNI and J-ADNI are 0.95 and 0.91, respectively.
Fig. 3 Cross-cohort evaluation results. ROC curves for both the NA-ADNI and J-ADNI are shown. The AUCs of the NA-ADNI and J-ADNI are 0.95 and 0.91, respectively.CC BY · PubMed Central
Fig. 4 Gradient maps of CNNs. a Gradient maps of CNNs of the whole brain. CNN visualization results of M1 (Supplementary Table 2) based on the whole brain. Red represents a high contribution ratio, and blue represents a low contribution ratio. b Gradient maps of CNNs of brain segments. CNN visualization results of M2 (Supplementary Table 2) based on brain segments. Red represents a high contribution ratio, and blue represents a low contribution ratio. Compared with the whole brain, the CNNs based on brain segments consistently focused on the hippocampi and amygdalae.
Fig. 4 Gradient maps of CNNs. a Gradient maps of CNNs of the whole brain. CNN visualization results of M1 (Supplementary Table 2) based on the whole brain. Red represents a high contribution ratio, and blue represents a low contribution ratio. b Gradient maps of CNNs of brain segments. CNN visualization results of M2 (Supplementary Table 2) based on brain segments. Red represents a high contribution ratio, and blue represents a low contribution ratio. Compared with the whole brain, the CNNs based on brain segments consistently focused on the hippocampi and amygdalae.CC BY · PubMed Central
Fig. 5 Visualization by UMAP (10 ⇒ 2 dimensions). Training samples were plotted after the dimensions of their image and non-image features were reduced to 2 by using UMAP. Blue dots represent sMCIs that were correctly classified as sMCIs by our model, and blue “xs” represent sMCIs that were incorrectly classified as pMCIs. Red dots represent pMCIs that were correctly classified as pMCIs by our model, and red “xs” represent pMCIs that were classified incorrectly as sMCIs. Samples S1 to S4 were selected for CNN visualization in Fig. 4.
Fig. 5 Visualization by UMAP (10 ⇒ 2 dimensions). Training samples were plotted after the dimensions of their image and non-image features were reduced to 2 by using UMAP. Blue dots represent sMCIs that were correctly classified as sMCIs by our model, and blue “xs” represent sMCIs that were incorrectly classified as pMCIs. Red dots represent pMCIs that were correctly classified as pMCIs by our model, and red “xs” represent pMCIs that were classified incorrectly as sMCIs. Samples S1 to S4 were selected for CNN visualization in Fig. 4.CC BY · PubMed Central

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

Show all 55 references

Research works printed but not linked: 5

  1. Goto, T., Wang, C., Li, Y. & Tsuboshita, Y. Multi-modal deep learning for predicting progression of Alzheimer’s disease using bi-linear shake fusion. Proc. SPIE (Med. Imaging) 11314, 452–457 (2020).
  2. Srivastava, N., Hinton, G., Krizhevsky, A., Sutskever, I. & Salakhutdinov, R. Dropout: a simple way to prevent neural networks from overfitting. J. Mach. Learn. Res. 15, 1929–1958 (2014).
  3. Vincent, P., Larochelle, H., Bengio, Y. & Manzagol, P. A. Stacked denoising autoencoders: learning useful representations in a deep network with a local denoising criterion. J. Mach. Learn. Res. 11, 3371–3408 (2010).
  4. Buolamwini, J. & Gebru, T. Gender shades: intersectional accuracy disparities in commercial gender classification. Proc. Mach. Learn. Res. 81, 1–15 (2018).
  5. Mattes, D., Haynor, D. R., Vesselle, H., Lewellen, T. K. & Eubank, W. Nonrigid multimodality image registration. Proc. SPIE 4322, Medical Imaging: Image Processing, 1609-1620(2001).

Cited by

See all 70

Metadata

Authors
Wang, C., Li, Y., Tsuboshita, Y., Sakurai, T., Goto, T., Yamaguchi, H., Yamashita, Y., Sekiguchi, A., Tachimori, H., Initiative, F. T. A. D. N., Wang, C., Li, Y., & Goto, T.
Affiliations
  • Fujifilm (Japan) (JP)
  • National Center of Neurology and Psychiatry (JP)
  • Keio University (JP)
Venue
npj Digital Medicine
Published
2022-04-12
Volume
5
Issue
1
Pages
43-43
ISSN
2398-6352 (ISSN-L)
OpenAlex
W4223442957
Cited by
51
References
54
Citations by year
2026: 10 · 2025: 16 · 2024: 9 · 2023: 13 · 2022: 3
FWCI
6.68
Type
journal article
Language
en
Open access
Licence
Locations
  • npj Digital Medicine (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
Funders
  • U.S. Department of Defense (U01 AG024904, W81XWH, W81XWH-12-2-, W81XWH-12, W81XWH1220012, AG024904)
  • Foundation for the National Institutes of Health (W81XWH-12-2-0012, U01AG024904, Grant U01 AG024904, AG024904)
  • Alzheimer's Association (U01 AG024904, W81XWH-12-2-0012)
  • Bristol-Myers Squibb (U01 AG024904, AG024904, W81XWH-12-2-0012)
  • Eli Lilly and Company (U01 AG024904, AG024904, W81XWH-12-2-0012)
  • Pfizer (AG024904, W81XWH-12-2-0012, U01 AG024904)
  • Biogen (W81XWH-12-2-0012, U01 AG024904, AG024904)
  • University of Southern California (W81XWH-12-2-0012, U01 AG024904)
  • F. Hoffmann-La Roche (W81XWH-12-2-0012, AG024904, U01 AG024904)
  • Meso Scale Diagnostics (U01 AG024904, AG024904, W81XWH-12-2-0012)
  • Alzheimer's Disease Neuroimaging Initiative (W81XWH-12-2–0012, U01AG024904, AG024904, W81XWH-12-2-, Department of Defense award number W81XWH-12-2-0012)
  • BioClinica (U01 AG024904, W81XWH-12-2-0012, AG024904)
  • Novartis Pharmaceuticals Corporation (W81XWH-12-2-0012, U01 AG024904, AG024904)
  • Northern California Institute for Research and Education (U01 AG024904, W81XWH-12-2-0012)
  • Fujifilm Corporation
  • Eisai (W81XWH-12-2-0012, U01 AG024904, AG024904)
  • Servier (AG024904, W81XWH-12-2-0012, U01 AG024904)
  • H. Lundbeck A/S (U01 AG024904, AG024904, W81XWH-12-2-0012)
  • IXICO (AG024904, W81XWH-12-2-0012, U01 AG024904)
  • National Institutes of Health (3u01ag024904-10s2, W81XWH, W81XWH-12-2-0012, U01-AG024904, Grant U01 AG024904, AG024904)
  • Genentech (W81XWH-12-2-0012, AG024904, U01 AG024904)
  • Canadian Institutes of Health Research (U01 AG024904, W81XWH-12-2-0012)
  • Japan Society for the Promotion of Science (22K15777)
  • Japan Science and Technology Agency (JPMJOP1842)
  • National Institute on Aging (U01-AG-024904, W81XWH-12-2–0012, AG 024904)
  • National Institute of Biomedical Imaging and Bioengineering (W81XWH-12-2-0012, U01 AG024904, AG024904)
  • Program on Open Innovation Platform with Enterprises, Research Institute and Academia (JPMJOP1842)
Topics
  • Health Sciences › Medicine › Psychiatry and Mental health › Dementia and Cognitive Impairment Research
  • Health Sciences › Health Professions › Health Information Management › Artificial Intelligence in Healthcare
  • Physical Sciences › Computer Science › Artificial Intelligence › Machine Learning in Healthcare
Keywords
Alzheimer's disease, disease progression prediction, fast progressors, data scarcity, overfitting, generalizability, convolutional neural networks, brain segmentation, image feature extraction, linear support vector machine, cognitive decline, clinical trial recruitment

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: A high-generalizability machine learning framework for predicting the progression of Alzheimer’s disease using limited data. https://app.cochat.ai/papers/10.1038/s41746-022-00577-x
The paper
Wang, C., Li, Y., Tsuboshita, Y., Sakurai, T., Goto, T., Yamaguchi, H., Yamashita, Y., Sekiguchi, A., Tachimori, H., Initiative, F. T. A. D. N., Wang, C., Li, Y., & Goto, T. (2022). A high-generalizability machine learning framework for predicting the progression of Alzheimer’s disease using limited data. npj Digital Medicine. https://doi.org/10.1038/s41746-022-00577-x

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.