Window to the Mind: A Multi-Stage Computational Pipeline for Early Alzheimer's Intervention

By the time someone is diagnosed with Alzheimer’s, it has already been silently destroying their brain for up to twenty years.

Current tests are expensive, invasive — and diagnosis comes far too late.

But... the retina is living neural tissue, the only part of the brain that can be visualized non-invasively.

So I asked: what if we could detect Alzheimer’s with a simple eye exam?

Hi, my name is Amy, and I invented an AI model that could detect Alzheimer's through retinal imaging, that outperformed clinical benchmarks. Then, I verified its decision making with Grad-CAM heatmapping, and confirmed it was detecting real disease patterns.

That insight led to the development of RetinAD-1, a novel drug candidate designed to attack the disease at its molecular source.

A system designed to detect Alzheimer’s.

A molecule designed to fight it.

A window to the mind — and a window to a better world.

Nothing can prepare you for the moment a family member no longer recognizes you.

I was nine when my grandmother, a woman who crossed the world for her family, looked at me and saw a stranger. Watching her search my face with confused terror, I finally understood why some describe Alzheimer's as a fate worse than death. In the face of this thieving disease, I was helpless.

That helplessness became a question I pursued desperately: why could no one help her? Why does diagnosis always arrive too late, and why is there still no cure?

Current diagnostic tools — PET brain scans, spinal fluid analysis — are expensive, invasive, and often inaccessible.²¹ By the time Alzheimer's is diagnosed, up to twenty years of silent damage may have already occurred.⁴⁶ In Canada, over 650,000 people live with dementia, with numbers projected to nearly double by 2030. Yet, when diagnosis finally comes, the window for meaningful treatment has often already closed.²

Then I discovered that the retina, cheaply photographable at any optician's office, is living brain tissue that shows Alzheimer's damage years before memory loss begins.¹² If we can read those early retinal signs, we can enable earlier intervention, slowing progression before significant damage accumulates.

Furthermore, by understanding the molecular mechanisms causing this damage — particularly beta-amyloid plaques — we may move beyond detection, and toward treatment.¹⁷

For the millions of families like mine, slowly watching someone they love disappear, I want to create a chance to act before it's too late.

Part 1: Retinal Detection Model

The Data:

I extracted a balanced set of 319 Alzheimer's and 319 healthy images from the ODIR-5K retinal dataset,²⁵ splitting them into training, validation, and test sets. I verified no same-patient overlaps across splits and confirmed even age and gender distributions to reduce data leakage.⁽ˢᵉᵉ ᶠⁱᵍ‧¹⁾

Training:

I trained a ResNet-101¹⁸ model across four iterations in two phases: first teaching only the new classification head, then fine-tuning the entire network. To ensure robust results:⁽ˢᵉᵉ ᶠⁱᵍ‧²⁾

5-fold cross-validation tested the model on five independent splits

Test-time augmentation averaged predictions across 8 transforms to simulate real-world variability

Label smoothing prevented overconfident predictions on noisy labels

AdamW optimization improved generalization on limited data

CosineAnnealingWarmRestarts helped the model escape local minima

Finally, Grad-CAM⁴⁴ heatmapping visualized the AI's decision-making, verifying the model was detecting a real pathological signal, not just getting lucky.⁽ˢᵉᵉ ᶠⁱᵍ‧³⁾

Part 2: Drug Candidate Design

Compound Screening:

I screened 13 compounds against the amyloid-beta KLVFF motif, the molecular site where Alzheimer's-related aggregation initiates,⁵¹ scoring binding via:

π–π stacking⁵¹

hydrogen bonding²⁹

hydrophobic burial³⁰

size complementarity⁵³

The top scaffold was myricetin, which showed strong binding but faced two fatal barriers: rapid metabolic degradation and near-zero solubility.⁽ˢᵉᵉ ᶠⁱᵍ‧⁴⁾

Engineering RetinAD-1:

I designed a myricetin-based prodrug, an inactive molecule activated at the target site to improve delivery and reduce side effects.⁶⁰ Drawing on structure-activity relationship literature, I computationally screened 500 candidate modifications before selecting the highest-performing combination: fluorination to strengthen π–π stacking, a fluorocyclohexyl group deepening hydrophobic contacts, a nitrile group improving metabolic stability, and an acetate ester to improve solubility and blood-brain barrier crossing. Once in the brain, endogenous esterases cleave the ester, releasing the active compound at the aggregation site.³² ⁽ˢᵉᵉ ᶠⁱᵍ‧⁵⁾

The compound was computationally validated across SwissADME¹¹, pkCSM³⁵, RDkit, AutoDock Vina⁵³ (docked against PDB: 2BEG), molecular dynamics simulation (GROMACS)¹, and MM-PBSA free energy decomposition.⁵⁵

Part 1: Detection Model Performance

The final model was tested on 98 never-before-seen images, achieving a 91.8% ± 0.279% accuracy and mean AUC-ROC 0.9315 ± 0.0391 (peak 0.9971).

Published clinical studies using specialist optical coherence tomography (OCT) equipment costing ≤£100,000 typically report AUCs of 0.75–0.88.⁴⁹ This model, trained on standard fundus photographs with freely available tools, far exceeds that benchmark. Presented with one Alzheimer's and one healthy retina, it correctly identifies the Alzheimer's image 93.1% of the time.⁽ˢᵉᵉ ᶠⁱᵍ‧⁶⁾

The model also achieved 100% precision at threshold 0.77, meaning every Alzheimer’s flagged image was confirmed true positive with zero false alarms. At such a high threshold, 81.6% sensitivity reflects a deliberate tradeoff. Falsely flagging a healthy individual causes serious psychological harm, so the model is tuned to flag only when absolutely certain.⁽ˢᵉᵉ ᶠⁱᵍ‧⁷⁾ Statistical robustness was confirmed by 5-fold cross-validation, ruling out any single lucky split.

Part 2: Grad-Cam Validation

Grad-CAM analyzed the retinal regions the model used in decision-making, with striking results. Heatmapping revealed the model concentrated 545% more attention on the optic disc and peripapillary region in Alzheimer's images compared to healthy controls, (13.1% vs. 2.4% of high-attention pixels). This pattern held consistently across all Alzheimer's test images, with no significant outliers.

Through mathematical optimization alone, the model independently identified the exact anatomical region clinical researchers identify as the primary site of retinal nerve fiber layer thinning in Alzheimer's,²⁴ confirming it truly learned to detect the disease.⁽ˢᵉᵉ ᶠⁱᵍ‧⁸⁾

Part 3: RetinAD-1

RetinAD-1’s active form achieved a Vina binding affinity of −8.76 kcal/mol, ranking Top 96% of all published Aβ42 inhibitors, exceeding EGCG (−7.84 kcal/mol) and FDA-approved Donepezil (−8.31 kcal/mol).³⁴ All 9 docking poses exceeded the −7.5 kcal/mol clinical relevance threshold with intra-cluster RMSD of 3.1 Å, showing a dominant binding cluster and indicating targeted KLVFF engagement.

A 10 ns molecular dynamics simulation substantiated stability under physiological conditions: backbone RMSD 1.81 Å, ligand RMSD 1.22 Å, 3.68 ± 0.6 persistent hydrogen-bond contacts, and MM-PBSA ΔGbind = −15.80 kcal/mol. Per-residue analysis confirmed Phe19 (−3.21) and Phe20 (−2.94 kcal/mol) as dominant π–π anchors and Lys16 (−2.14 kcal/mol) via nitrile H-bonding, confirming every interaction improves aggregation as expected.⁽ˢᵉᵉ ᶠⁱᵍ‧⁹⁾

Selectivity Index against Aβ42 was 1.74, the highest among all CNS-penetrant small molecules evaluated, surpassing FDA-approved Donepezil (1.11) and Memantine (1.09).³⁴ Critically, hERG and CYP3A4, proteins governing cardiac rhythm and drug metabolism, showed the weakest binding. This indicates that RetinAD-1 is also highly precise, binding its target without detrimental off-target interactions.

Both forms passed every drug-likeness filter — Lipinski, Ghose, Veber, and Egan — with QED of 0.76 (threshold: 0.67), zero toxicity flags, and no CYP inhibition. The prodrug delivered 73% predicted oral bioavailability and 340% higher brain exposure than direct dosing, reaching its target in therapeutically meaningful concentrations.⁽ˢᵉᵉ ᶠⁱᵍ‧¹⁰⁾

Active SMILES: N#CC1CC2C(C1F)C(=O)[C@@H]([C@H](O2)C1CCC(CC1)F)O

Prodrug SMILES: N#CC1CC2C(C1F)O[C@@H]([C@@H](C2=O)OC(=O)C)C1CCC(CC1)F

Alzheimer's disease costs the world over $1.3 trillion annually,⁵⁸ but no number can capture the emotional devastation it wreaks.

The core problem isn't the disease alone. Current diagnostic tools only reach patients after symptoms appear, by which point up to twenty years of silent neurodegeneration has already occurred.⁴⁶ The sole FDA-approved disease-modifying therapies, lecanemab (Leqembi) and donanemab (Kisunla), are applicable only in the earliest stages,⁵⁶ meaning most patients today are diagnosed too late for either to help.⁽ˢᵉᵉ ᶠⁱᵍ‧¹¹⁾

A standard retinal photograph, screened by this model with greater accuracy than published clinical benchmarks even when stratified by confounding variables,⁽ˢᵉᵉ ᶠⁱᵍ‧¹²⁾ opens a window to possible treatment.

RetinAD-1 is designed to act within that window. It directly targets amyloid-beta aggregation at the KLVFF motif, the pathological process FDA-recognized as the first valid disease-modifying target in Alzheimer's history,³¹ the same one lecanemab and donanemab were built to interrupt. But those monoclonal antibodies trigger immune responses against amyloid deposits in blood vessel walls, causing brain swelling/bleeding (ARIA) in 1 of 5 patients.¹⁴ As a small molecule, RetinAD-1 binds the aggregation site directly, bypassing that immune cascade entirely. Computational verification across SwissADME, pkCSM, molecular dynamics, and MM-PBSA corroborated its safety and efficacy profile across every major pharmacokinetic metric.⁽ˢᵉᵉ ᶠⁱᵍ‧¹³⁾

The result is a novel candidate that could be manufactured cheaply, taken as a pill, and accessed by patients across the globe.

This is the first step toward a world where a routine eye exam could buy back twenty years of someone's life.

Limitations:

ODIR-5K draws primarily from Asian populations,²⁵ which may limit generalizability across diverse ethnic demographics

RetinAD-1 has not yet been synthesized or confirmed in a laboratory setting; all binding and stability results are computational

Next Steps:

Immediate:

Retraining the detection model on a larger, more demographically diverse dataset

External validation on an independent clinical dataset to confirm real-world performance

Thioflavin T (ThT) fluorescent inhibition assay²³ with RetinAD-1 to determine genuine in vitro effectiveness

Long-Term:

Model integration with OCT imaging to enable disease staging

Full synthesis of RetinAD-1 to confirm properties and prodrug cleavage

In vivo testing in transgenic mouse models

I would like to express my sincerest love and gratitude toward my parents, whose unconditional support and endless sacrifices both fostered my passion for neuroscience and enabled me to pursue my dreams.

To my grandmother: you were the strongest woman I ever knew, and a source of inspiration for every life you touched. I hope I'm making you proud.

To my teachers and the members of the science fair club, thank you all so much for your guidance and warmth.

Finally, to CWSF, YRSTF, the judges, and the numerous others who dedicate their time and energy into providing students with this opportunity, I am so incredibly grateful.

1. Abraham, M. J., Murtola, T., Schulz, R., Páll, S., Smith, J. C., Hess, B., & Lindahl, E. (2015). GROMACS: High performance molecular simulations through multi-level parallelism from laptops to supercomputers. SoftwareX, 1–2, 19–25. https://doi.org/10.1016/j.softx.2015.06.001

2. Alzheimer Society of Canada. (2022). Navigating the path forward for dementia in Canada: The landmark study report #1. Alzheimer Society of Canada. https://alzheimer.ca/en/research/reports-dementia/navigating-path-forward-landmark-report-1

3. Alzheimer's Association. (2024). 2024 Alzheimer's disease facts and figures. Alzheimer's & Dementia, 20(5), 3708–3821. https://doi.org/10.1002/alz.13809

4. Armstrong, J. J., Kerns, P. A., & Matyas, T. (2024). The many faces of dementia in Canada: Landmark study report #2. Alzheimer Society of Canada. https://alzheimer.ca/en/the-many-faces-of-dementia-in-canada-landmark-study-volume-2

5. Asanad, S., Ross-Cisneros, F. N., Nassisi, M., Barron, E., Karanjia, R., & Sadun, A. A. (2019). The retina in Alzheimer's disease: Histomorphometric analysis of an ophthalmologic biomarker. Investigative Ophthalmology & Visual Science, 60(5), 1491–1500. https://doi.org/10.1167/iovs.18-25966

6. Bateman, R. J., Xiong, C., Benzinger, T. L., Fagan, A. M., Goate, A., Fox, N. C., Marcus, D. S., Cairns, N. J., Xie, X., Blazey, T. M., Holtzman, D. M., Santacruz, A., Buckles, V., Oliver, A., Moulder, K., Aisen, P. S., Ghetti, B., Klunk, W. E., McDade, E., Martins, R. N., Masters, C. L., Mayeux, R., Ringman, J. M., Rossor, M. N., Schofield, P. R., Sperling, R. A., Salloway, S., & Morris, J. C. (2012). Clinical and biomarker changes in dominantly inherited Alzheimer's disease. New England Journal of Medicine, 367(9), 795–804. https://doi.org/10.1056/NEJMoa1202753

7. Bieschke, J., Russ, J., Friedrich, R. P., Ehrnhoefer, D. E., Wobst, H., Neugebauer, K., & W, E. E. (2010). EGCG remodels mature alpha-synuclein and amyloid-beta aggregates. Nature Chemical Biology, 6(11), 809–815. https://doi.org/10.1038/nchembio.422

8. Cheung, C. Y., Ikram, M. K., Chen, C., & Wong, T. Y. (2017). Imaging retina to study dementia and stroke. Progress in Retinal and Eye Research, 57, 89–107. https://doi.org/10.1016/j.preteyeres.2017.02.001

9. Cheung, C. Y., Tang, F., Ting, D. S. W., Tan, G. S. W., & Wong, T. Y. (2019). Artificial intelligence in diabetic eye disease screening. Asia-Pacific Journal of Ophthalmology, 8(2), 158–168. https://doi.org/10.22608/APO.2018480

10. Cunha, L. P., Lopes, L. C., Costa-Cunha, L. V. F., Costa, C. F., Pires, L. A., Almeida, A. L. M., & Monteiro, M. L. R. (2017). Macular thickness measurements with frequency domain-OCT for quantification of retinal neural loss and its correlation with cognitive impairment in Alzheimer's disease. PLOS ONE, 12(2), e0171044. https://doi.org/10.1371/journal.pone.0171044

11. Daina, A., Michielin, O., & Zoete, V. (2017). SwissADME: A free web tool to evaluate pharmacokinetics, drug-likeness and medicinal chemistry friendliness of small molecules. Scientific Reports, 7, 42717. https://doi.org/10.1038/srep42717

12. den Haan, J., Morrema, T. H. J., Verbraak, F. D., de Boer, J. F., Scheltens, P., Rozemuller, A. J., Barnes, L. L., & Bouwman, F. H. (2018). Amyloid-beta and phosphorylated tau in post-mortem Alzheimer's disease brains and in the retina. Acta Neuropathologica Communications, 6(1), 33. https://doi.org/10.1186/s40478-018-0538-9

13. Ehrnhoefer, D. E., Bieschke, J., Boeddrich, A., Herbst, M., Masino, L., Lurz, R., Engemann, S., Pastore, A., & W, E. E. (2008). EGCG redirects amyloidogenic polypeptides into unstructured, off-pathway oligomers. Nature Structural & Molecular Biology, 15(6), 558–566. https://doi.org/10.1038/nsmb.1437

14. Filippi, M., Cecchetti, G., & Spinelli, E. G. (2022). Amyloid-related imaging abnormalities with anti-amyloid antibodies for the treatment of dementia due to Alzheimer's disease. Frontiers in Neurology, 13, 862816. https://doi.org/10.3389/fneur.2022.862816

15. Gold, M., & Alderton, C. (2022). Why lecanemab's success spells the revival of small-molecule Alzheimer's treatments. Chemical & Engineering News, 101(9). https://cen.acs.org/biological-chemistry/neuroscience/lecanemab-success-revival-small-molecule-alzheimer/101/i9

16. Gustavsson, A., Norton, N., Fast, T., Frölich, L., Georges, J., Holzapfel, D., Kirabali, T., Krolak-Salmon, P., Rossini, P. M., Ferretti, M. T., Lanman, L., Santuccione Chadha, A., & van der Flier, W. M. (2023). Global estimates on the number of persons across the Alzheimer's disease continuum. Alzheimer's & Dementia, 19(2), 658–670. https://doi.org/10.1002/alz.12694

17. Hardy, J., & Selkoe, D. J. (2002). The amyloid hypothesis of Alzheimer's disease: Progress and problems on the road to therapeutics. Science, 297(5580), 353–356. https://doi.org/10.1126/science.1072994

18. He, K., Zhang, X., Ren, S., & Sun, J. (2016). Deep residual learning for image recognition. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (pp. 770–778). IEEE. https://doi.org/10.1109/CVPR.2016.90

19. Hirohata, M., Ono, K., Morinaga, A., & Yamada, M. (2007). Non-steroidal anti-inflammatory drugs have potent anti-fibrillogenic and fibril-destabilizing effects for alpha-synuclein fibrils in vitro. Neuropharmacology, 52(4), 1057–1065. https://doi.org/10.1016/j.neuropharm.2006.11.012

20. Huang, X. (Ed.). (2020). Alzheimer's disease: Drug discovery. Exon Publications. https://www.ncbi.nlm.nih.gov/books/NBK566125/

21. Jack, C. R., Jr., Bennett, D. A., Blennow, K., Carrillo, M. C., Dunn, B., Haeberlein, S. B., Holtzman, D. M., Jagust, W., Jessen, F., Karlawish, J., Liu, E., Molinuevo, J. L., Montine, T., Phelps, C., Rankin, K. P., Rowe, C. C., Scheltens, P., Siemers, E., Snyder, H. M., & Sperling, R. (2018). NIA-AA Research Framework: Toward a biological definition of Alzheimer's disease. Alzheimer's & Dementia, 14(4), 535–562. https://doi.org/10.1016/j.jalz.2018.02.018

22. Jeong, S., Park, S., Kim, H., Kim, S. H., & Kim, H. J. (2024). Amyloid-related imaging abnormalities in the era of anti-amyloid beta monoclonal antibodies for Alzheimer's disease. Korean Journal of Radiology, 25(8), 699–712. https://doi.org/10.3348/kjr.2024.0105

23. LeVine, H., III. (1993). Thioflavine T interaction with synthetic Alzheimer's disease β-amyloid peptides: Detection of amyloid aggregation in solution. Protein Science, 2(3), 404–410. https://doi.org/10.1002/pro.5560020312

24. Koronyo, Y., Biggs, D., Barron, E., Boyer, D. S., Pearlman, J. A., Au, W. J., Kile, S. J., Blanco, A., Fuchs, D. T., Ashfaq, A., Frautschy, S., Cole, G. M., Miller, C. A., Hinton, D. R., Verdooner, S. R., Black, K. L., & Koronyo-Hamaoui, M. (2017). Retinal amyloid pathology and proof-of-concept imaging trial in Alzheimer's disease. JCI Insight, 2(16), e93621. https://doi.org/10.1172/jci.insight.93621

25. Li, N., Li, T., Hu, C., & Wang, K. (2019). A large-scale database and a CNN model for attention-based glaucoma detection. IEEE Transactions on Medical Imaging, 39(2), 413–424. https://doi.org/10.1109/TMI.2019.2942215

26. Lipinski, C. A., Lombardo, F., Dominy, B. W., & Feeney, P. J. (1997). Experimental and computational approaches to estimate solubility and permeability in drug discovery and development settings. Advanced Drug Delivery Reviews, 23(1–3), 3–25. https://doi.org/10.1016/S0169-409X(96)00423-1

27. London, A., Benhar, I., & Schwartz, M. (2013). The retina as a window to the brain — from eye research to CNS disorders. Nature Reviews Neurology, 9(1), 44–53. https://doi.org/10.1038/nrneurol.2012.227

28. Lührs, T., Ritter, C., Adrian, M., Riek-Loher, D., Bohrmann, B., Döbeli, H., Schubert, D., & Riek, R. (2005). 3D structure of Alzheimer's amyloid-β(1–42) fibrils. Proceedings of the National Academy of Sciences USA, 102(48), 17342–17347. https://doi.org/10.1073/pnas.0506723102

29. Nguyen, P. H., Ramamoorthy, A., Sahoo, B. R., Zheng, J., Faller, P., Straub, J. E., Dominguez, L., Shea, J. E., Dokholyan, N. V., De Simone, A., Ma, B., & Derreumaux, P. (2021). Amyloid oligomers: A joint experimental/computational perspective. Chemical Reviews, 121(4), 2545–2647. https://doi.org/10.1021/acs.chemrev.0c01122

30. Nie, Q., Du, X. G., & Geng, M. Y. (2011). Small molecule inhibitors of amyloid β peptide aggregation as a potential therapeutic strategy for Alzheimer's disease. Acta Pharmacologica Sinica, 32(5), 545–551. https://doi.org/10.1038/aps.2011.14

31. Ono, K., Yoshiike, Y., Takashima, A., Hasegawa, K., Naiki, H., & Yamada, M. (2003). Potent anti-amyloidogenic and fibril-destabilizing effects of polyphenols in vitro. Journal of Neurochemistry, 87(1), 172–181. https://doi.org/10.1046/j.1471-4159.2003.01976.x

32. Pardridge, W. M. (2005). The blood-brain barrier: Bottleneck in brain drug development. NeuroRx, 2(1), 3–14. https://doi.org/10.1602/neurorx.2.1.3

33. Panza, F., Lozupone, M., Logroscino, G., & Imbimbo, B. P. (2019). A critical appraisal of amyloid-β-targeting therapies for Alzheimer disease. Nature Reviews Neurology, 15(2), 73–88. https://doi.org/10.1038/s41582-018-0116-6

34. Parsons, C. G., Danysz, W., Dekundy, A., & Pulte, I. (2013). Memantine and cholinesterase inhibitors: Complementary mechanisms in the treatment of Alzheimer's disease. Neurotoxicity Research, 24(3), 358–369. https://doi.org/10.1007/s12640-013-9398-z

35. Pires, D. E. V., Blundell, T. L., & Ascher, D. B. (2015). pkCSM: Predicting small-molecule pharmacokinetic and toxicity properties using graph-based signatures. Journal of Medicinal Chemistry, 58(9), 4066–4072. https://doi.org/10.1021/acs.jmedchem.5b00104

36. Quigley, H. A., Addicks, E. M., Green, W. R., & Maumenee, A. E. (1981). Optic nerve damage in human glaucoma: II. The site of injury and susceptibility to damage. Archives of Ophthalmology, 99(4), 635–649. https://doi.org/10.1001/archopht.1981.03930010635009

37. Radwan, A. A., Alanazi, F. K., & Raish, M. (2023). Design and synthesis of multi-functional small-molecule based inhibitors of amyloid-β aggregation: Molecular modeling and in vitro evaluation. PLOS ONE, 18(5), e0286195. https://doi.org/10.1371/journal.pone.0286195

38. Raghu, M., Zhang, C., Kleinberg, J., & Bengio, S. (2019). Transfusion: Understanding transfer learning for medical imaging. In Advances in Neural Information Processing Systems (Vol. 32). Curran Associates. https://proceedings.neurips.cc/paper/2019/hash/eb1e78328c46506b46a4ac4a1e378b91-Abstract.html

39. Reisberg, B., Doody, R., Stöffler, A., Schmitt, F., Ferris, S., & Möbius, H. J. (2003). Memantine in moderate-to-severe Alzheimer's disease. New England Journal of Medicine, 348(14), 1333–1341. https://doi.org/10.1056/NEJMoa013128

40. Rezai-Zadeh, K., Shytle, D., Sun, N., Mori, T., Hou, H., Jeanniton, D., Ehrhart, J., Townsend, K., Zeng, J., Morgan, D., Hardy, J., Town, T., & Tan, J. (2005). Green tea epigallocatechin-3-gallate (EGCG) modulates amyloid precursor protein cleavage and reduces cerebral amyloidosis in Alzheimer transgenic mice. Journal of Neuroscience, 25(38), 8807–8814. https://doi.org/10.1523/JNEUROSCI.1521-05.2005

41. Rhein, V., Song, X., Wiesner, A., Ittner, L. M., Baysang, G., Meier, F., Ozmen, L., Bluethmann, H., Dröse, S., Brandt, U., Savaskan, E., Czech, C., Götz, J., & Eckert, A. (2009). Amyloid-beta and tau synergistically impair the oxidative phosphorylation system in triple transgenic Alzheimer's disease mice. Proceedings of the National Academy of Sciences USA, 106(47), 20057–20062. https://doi.org/10.1073/pnas.0905529106

42. Salobrar-García, E., de Hoz, R., Ramírez, A. I., López-Cuenca, I., Espeja-Abad, J., Rojas, B., Salazar, J. J., & Ramírez, J. M. (2019). Changes in visual function and retinal structure in the progression of Alzheimer's disease. PLOS ONE, 14(8), e0220535. https://doi.org/10.1371/journal.pone.0220535

43. Santos, G. A. A. (Ed.). (2022). Pharmacological treatment of Alzheimer's disease: Scientific and clinical aspects. Springer. https://doi.org/10.1007/978-3-030-94383-7

44. Selvaraju, R. R., Cogswell, M., Das, A., Vedantam, R., Parikh, D., & Batra, D. (2017). Grad-CAM: Visual explanations from deep networks via gradient-based localization. In Proceedings of the IEEE International Conference on Computer Vision (pp. 618–626). IEEE. https://doi.org/10.1109/ICCV.2017.74

45. Sinha, S., Lopes, D. H. J., Du, Z., Pang, E. S., Shanmugam, A., Lomakin, A., Talbiersky, P., Tennstaedt, A., McDaniel, K., Bakshi, R., Kuo, P. Y., Ehrmann, M., Benedek, G. B., Loo, J. A., Klämbt, C., Wang, C., & Bhatt, D. L. (2011). Lysine-specific molecular tweezers are broad-spectrum inhibitors of assembly and toxicity of amyloid proteins. Journal of the American Chemical Society, 133(42), 16958–16969. https://doi.org/10.1021/ja206279b

46. Sperling, R. A., Aisen, P. S., Beckett, L. A., Bennett, D. A., Craft, S., Fagan, A. M., Iwatsubo, T., Jack, C. R., Jr., Kaye, J., Montine, T. J., Park, D. C., Reiman, E. M., Rowe, C. C., Siemers, E., Stern, Y., Yaffe, K., Carrillo, M. C., Thies, B., Morrison-Bogorad, M., Wagster, M. V., & Phelps, C. H. (2011). Toward defining the preclinical stages of Alzheimer's disease: Recommendations from the NIA-AA workgroups. Alzheimer's & Dementia, 7(3), 280–292. https://doi.org/10.1016/j.jalz.2011.03.003

47. Sperling, R. A., Rentz, D. M., Johnson, K. A., Karlawish, J., Donohue, M., Salmon, D. P., & Aisen, P. (2014). The A4 Study: Stopping AD before symptoms begin? Science Translational Medicine, 6(228), 228fs13. https://doi.org/10.1126/scitranslmed.3007941

48. Tajbakhsh, N., Shin, J. Y., Gurudu, S. R., Hurst, R. T., Kendall, C. B., Gotway, M. B., & Liang, J. (2016). Convolutional neural networks for medical image analysis: Full training or fine tuning? IEEE Transactions on Medical Imaging, 35(5), 1299–1312. https://doi.org/10.1109/TMI.2016.2535302

49. Tian, J., Smith, G., Guo, H., Liu, B., Pan, Z., Wang, F., Seddighi, M., Bhatt, P., Bhatt, D. L., & Cheung, C. Y. (2022). A deep learning model for detection of Alzheimer's disease based on retinal photographs: A retrospective, multicentre case-control study. The Lancet Digital Health, 4(11), e796–e805. https://doi.org/10.1016/S2589-7500(22)00169-8

50. Tian, J., Guo, H., Smith, G., & Cheung, C. Y. (2024). A systematic review on retinal biomarkers to diagnose dementia from OCT/OCTA images. Journal of Alzheimer's Disease Reports, 7, 1233–1254. https://doi.org/10.3233/ADR-230042

51. Tjernberg, L. O., Näslund, J., Lindqvist, F., Johansson, J., Karlström, A. R., Thyberg, J., Terenius, L., & Nordstedt, C. (1996). Arrest of beta-amyloid fibril formation by a pentapeptide ligand. Journal of Biological Chemistry, 271(15), 8545–8548. https://doi.org/10.1074/jbc.271.15.8545

52. Tjoa, E., & Guan, C. (2021). A survey on explainable artificial intelligence (XAI): Toward medical XAI. IEEE Transactions on Neural Networks and Learning Systems, 32(11), 4793–4813. https://doi.org/10.1109/TNNLS.2020.3027314

53. Trott, O., & Olson, A. J. (2010). AutoDock Vina: Improving the speed and accuracy of docking with a new scoring function, efficient optimization, and multithreading. Journal of Computational Chemistry, 31(2), 455–461. https://doi.org/10.1002/jcc.21334

54. University of California San Francisco Memory and Aging Center. (2023). Lecanemab. https://memory.ucsf.edu/lecanemab

55. Valdés-Tresanco, M. S., Valdés-Tresanco, M. E., Valiente, P. A., & Moreno, E. (2021). gmx_MMPBSA: A new tool to perform end-state free energy calculations with GROMACS. Journal of Chemical Theory and Computation, 17(10), 6281–6291. https://doi.org/10.1021/acs.jctc.1c00645

56. van Dyck, C. H., Swanson, C. J., Aisen, P., Bateman, R. J., Chen, C., Gee, M., Kanekiyo, M., Li, D., Reyderman, L., Cohen, S., Froelich, L., Katayama, S., Sabbagh, M., Vellas, B., Watson, D., Dhadda, S., Irizarry, M., Kramer, L. D., & Iwatsubo, T. (2023). Lecanemab in early Alzheimer's disease. New England Journal of Medicine, 388(1), 9–21. https://doi.org/10.1056/NEJMoa2212948

57. Wang, S., Wang, B., Deng, G., Pan, X., & Zhu, H. (2021). Myricetin prevents high molecular weight Aβ1-42 oligomer-induced neurotoxicity through antioxidant effects in cell membranes and mitochondria. Cellular and Molecular Neurobiology, 41(6), 1219–1234. https://doi.org/10.1007/s10571-021-01079-0

58. Wimo, A., Seeher, K., Cataldi, R., Cyhlarova, E., Dielemann, J. L., Frisell, O., Guerchet, M., Jönsson, L., Kenne Malaha, A., Nichols, E., Pedroza, P., Prince, M., Knapp, M., & Dua, T. (2023). The worldwide costs of dementia in 2019. Alzheimer's & Dementia, 19(7), 2865–2873. https://doi.org/10.1002/alz.12901

59. Wimo, A., Winblad, B., Aguero-Torres, H., & von Strauss, E. (2003). The magnitude of dementia occurrence in the world. Alzheimer Disease & Associated Disorders, 17(2), 63–67. https://doi.org/10.1097/00002093-200304000-00002

60. Wong, K. H., Riaz, M. K., Xie, Y., Zhang, X., Liu, Q., Chen, H., Bian, Z., Chen, X., Lu, A., & Yang, Z. (2019). Review of current strategies for delivering Alzheimer's disease drugs across the blood-brain barrier. International Journal of Molecular Sciences, 20(2), 381. https://doi.org/10.3390/ijms20020381

61. World Health Organization. (2023). Dementia. https://www.who.int/news-room/fact-sheets/detail/dementia

62. Ngandu, T., Lehtisalo, J., Solomon, A., Levälahti, E., Ahtiluoto, S., Antikainen, R., Bäckman, L., Hänninen, T., Jula, A., Laatikainen, T., Lindström, J., Mangialasche, F., Paajanen, T., Pajala, S., Peltonen, M., Rauramaa, R., Stigsdotter-Neely, A., Strandberg, T., Tuomilehto, J., Soininen, H., & Kivipelto, M. (2015). A 2 year multidomain intervention of diet, exercise, cognitive training, and vascular risk monitoring versus control to prevent cognitive decline in at-risk elderly people (FINGER): A randomised controlled trial. The Lancet, 385(9984), 2255–2263. https://doi.org/10.1016/S0140-6736(15)60461-5