Systematic Modality Ablation of Multimodal Machine Learning for Predicting 24-Month Progression from Mild Cognitive Impairment to Alzheimer's Disease
Choe, S.. medRxiv (preprint), 2026-09-04
Topic: Amyloid, tau and blood-test diagnosis
INTRODUCTIONMultimodal biomarkers have transformed Alzheimers disease research, but the incremental contribution of each modality to predicting progression from mild cognitive impairment (MCI) remains unclear. We systematically evaluated individual biomarker modalities using a comprehensive ablation framework. METHODSWe analyzed 2,430 participants with MCI from the Alzheimers Disease Neuroimaging Initiative with known 24-month progression status. XGBoost models were trained using combinations of demographic variables, cognitive assessments, apolipoprotein E (APOE) genotype, structural MRI, cerebrospinal fluid (CSF) biomarkers, and PET biomarkers. Performance was evaluated using repeated stratified 5x10 cross-validation; discrimination was assessed using the area under the receiver operating characteristic curve (AUC), with pairwise comparisons via DeLongs test on single-pass out-of-fold predictions, Holm-Bonferroni corrected. Complete-case analyses evaluated the impact of missing data and median imputation. RESULTSThe full multimodal model achieved the highest discrimination (AUC=0.934). Excluding cognitive assessments produced the largest reduction (AUC=0.883, P<0.001 vs. multimodal), the only comparison to remain significant after correction. Removing APOE, CSF, or MRI produced only modest, statistically indistinguishable reductions (AUC=0.933, 0.931, 0.932). Removing PET produced a similarly small reduction in the primary analysis (AUC=0.932), but complete-case sensitivity analysis showed imputation significantly inflated this estimate (P=0.005); PETs true contribution may exceed the other three. The baseline clinical model performed near chance (AUC=0.556). DISCUSSIONCognitive assessment contributes substantially more predictive information for 24-month progression than structural imaging, molecular biomarkers, or genetic risk, whose removal produces only modest, largely statistically indistinguishable performance loss -- with the exception of PET, whose apparent equivalence may be an imputation artifact. These findings establish an evidence-based hierarchy of biomarker utility and provide a quantitative framework for designing cost-effective prediction models and prioritizing biomarker acquisition in clinical research and multimodal AI systems.
Key findings
- Excluding cognitive assessments produced the largest reduction (AUC=0.883, P<0.001 vs. multimodal), the only comparison to remain significant after correction.
- Removing PET produced a similarly small reduction in the primary analysis (AUC=0.932), but complete-case sensitivity analysis showed imputation significantly inflated this estimate (P=0.005); PETs true contribution may exceed the other three.
- These findings establish an evidence-based hierarchy of biomarker utility and provide a quantitative framework for designing cost-effective prediction models and prioritizing biomarker acquisition in clinical research and multimodal AI systems.
Why it matters
This study compared how much each type of biomarker adds to predicting progression from mild cognitive impairment to Alzheimer's over 24 months, and found that removing cognitive assessments caused the largest drop in performance. It is about prediction modeling and does not test any treatment, cure or prevention.
New to this subject? Start with our plain-English guide: Genetics →
Study details
- Participants2,430
- Effect sizeAUC=0.934; AUC=0.883; AUC=0.933
- Evidence strengthLow
How this study scored 16/100
Preprint — not yet peer-reviewed. It scored 28 on the same rubric, but non-peer-reviewed sources are capped at 16, so it shows as 16. Scored zero for study design and randomised / blinded because the record does not state it. We have no data on impact for its field and year, senior author's track record and institution's track record, so those were left out: it earned 24 of the 87 points we could judge it on, which is 28 out of 100.
| Component | Points | Why |
|---|---|---|
| Journal tier | 2 / 10 | Tier 4 — not a peer-reviewed journal (medRxiv (preprint)) |
| Journal's citation record | 0 / 4 | Not a journal article, so the journal's record does not count |
| Study design | 0 / 12 | No study design recorded — scores zero, not guessed (not stated in the record — scored 0) |
| Strength of evidence | 1 / 5 | Rated “low” in our summary |
| Peer-review validation | 4 / 10 | Preprint — not yet peer-reviewed. Hosted on a preprint server. |
| How recent | 5 / 5 | 2026 — within 1 year(s) of the newest study here |
| Sample size | 3 / 4 | 2,430 participants |
| Effect reported | 3 / 3 | Effect reported with a statistic (interval, p-value, ratio or similar) |
| Studied in people | 3 / 3 | Studied in people (participant count stated) |
| Randomised / blinded | 0 / 2 | No randomisation or blinding stated — scores zero (not stated in the record — scored 0) |
| Times cited | 0 / 7 | Cited 0 times (OpenAlex, 2026-10-10) |
| Citations per year | 0 / 5 | 0.0 citations a year over 2 year(s) since 2026 |
| Impact for its field and year | — / 6 | No field-weighted citation impact (OpenAlex has none yet) — left out of this score |
| Cited by studies here | 0 / 6 | No other study in this library cites it |
| Still being cited | 0 / 4 | Cited 0 times in 2024–2026 (2026 still in progress; OpenAlex) |
| Senior author's track record | — / 4 | OpenAlex's profile for last-listed author Sophie Choe holds only 4 work(s), too few to judge a track record — left out of this score |
| Institution's track record | — / 3 | No institution listed in OpenAlex — left out of this score |
| Multi-centre collaboration | 0 / 3 | OpenAlex lists 0 institution(s) in 0 countries — scores zero (affiliation lists can be incomplete) |
| Indexed in PubMed | 1 / 2 | Not in PubMed, but has a DOI |
| Free to read | 2 / 2 | Free to read |
| Points | 24 / 87 | Scored on 17 of 20 parts. We have no data for the rest, so they are left out: 24 of the 87 points we could judge it on is 28 out of 100. |
| Total | 28 / 100 | Capped to 16 because this is not a peer-reviewed article. |
What each of these 20 components measures, and what this score is not →
Read the original
Full text: free to read — OpenAlex lists this copy as free to read: the link above (checked 6 Feb 1020).
Cite this study APA, Vancouver, BibTeX · download .bib or .ris
Built from the record alzhub holds for this paper, not looked up. Volume, issue and page numbers are not held in the record. Check the citation against the original before you use it.
APA 7th
Choe. (2026). Systematic Modality Ablation of Multimodal Machine Learning for Predicting 24-Month Progression from Mild Cognitive Impairment to Alzheimer's Disease. medRxiv (preprint). https://doi.org/10.64898/2026.09.01.26360413
Vancouver
Choe. Systematic Modality Ablation of Multimodal Machine Learning for Predicting 24-Month Progression from Mild Cognitive Impairment to Alzheimer's Disease. medRxiv (preprint). 2026. doi:10.64898/2026.09.01.26360413.
BibTeX
@article{choe2026-c0e87325,
author = {Choe},
title = {{Systematic Modality Ablation of Multimodal Machine Learning for Predicting 24-Month Progression from Mild Cognitive Impairment to Alzheimer's Disease}},
journal = {medRxiv (preprint)},
year = {2026},
doi = {10.64898/2026.09.01.26360413},
url = {https://doi.org/10.64898/2026.09.01.26360413}
}
Studies this one cites
Referenced by this study and also in the alzhub library, so you can read them here. Showing 12 of 18.