Biomarkers for the Preclinical Detection of Cognitive Decline

24 concepts23 papers1 proposal

The Challenge of the Invisible Transition

Alzheimer’s disease (AD) and Mild Cognitive Impairment (MCI) are traditionally diagnosed only after substantial, irreversible cognitive decline has manifested. However, as Yanhong Zhou et al. note in The Biomarkers for Identifying Preclinical Alzheimer's Disease via Structural and Functional Magnetic Resonance Imaging, pathological brain changes begin a decade or more before clinical symptoms emerge. During this preclinical window, individuals appear completely asymptomatic on standard tests, yet underlying neural networks are already degrading.

The core diagnostic challenge, as framed by Rahila Umer et al. in Prediction of Possible conversion from MCI to AD using Machine learning, is that the subtle transition from healthy aging to early-stage pathology is easily missed by standard clinical screening tools. To intervene effectively before irreversible structural damage occurs, researchers are moving away from late-stage symptoms like severe memory loss, focusing instead on identifying hidden behavioral, physiological, and digital micro-signatures of early neurodegeneration.

Rethinking the Cognitive Timeline

The popular conception of Alzheimer's centers on episodic memory loss—such as forgetting recent events or names. However, longitudinal data reveals that other cognitive domains degrade first. In a study mapping the temporal order of decline, Murat Bilgel et al. demonstrated that subtle deficits in attention and working memory actually precede episodic memory failures (Temporal order of Alzheimer’s disease-related cognitive marker changes in BLSA and WRAP longitudinal studies). These shifts occur while individuals are still clinically classified as "cognitively normal."

This shifting timeline is supported by Thomas Adi Kurnia Susanto et al. in Cognition, Brain Atrophy, and Cerebrospinal Fluid Biomarkers Changes from Preclinical to Dementia Stage of Alzheimer's Disease and the Influence of Apolipoprotein E, who found that executive function decline and regional brain atrophy often surface before overt memory impairment. By shifting focus to these early-failing cognitive systems, researchers can design highly sensitive behavioral assays.

Behavioral Micro-Signatures and Passive Sensing

Because attention, executive control, and spatial processing degrade early, they manifest as subtle "micro-behaviors" during routine activities. Rather than relying on intensive, subjective clinical interviews, modern approaches capture these signatures passively or through targeted behavioral tasks.

The Dynamic Brain: Functional Network Shifts

Why do these behavioral shifts happen? Because the brain's underlying communication networks are losing their flexibility. Traditional neuroimaging looks at static structural snapshots, but the earliest signs of neurodegeneration are dynamic; the brain's ability to coordinate timing and switch signals smoothly between networks breaks down long before major structural atrophy occurs.

Mohammad S.E. Sendi analyzed dynamic functional network connectivity (dFNC) and discovered that sensory and motor networks experience temporal disruptions much earlier than the Default Mode Network (DMN) during the transition to mild dementia (Alzheimer’s Disease Projection From Normal to Mild Dementia Reflected in Functional Network Connectivity: A Longitudinal Study). However, within the DMN subnetwork itself, subtle phase alterations do occur early on. Chunting Cai et al. used the Hilbert Transform on resting-state fMRI data to uncover distinct "Phase Position Connectivity" (PPC) disruptions that separate subjective cognitive decline from objective amnestic MCI (Altered Patterns of Phase Position Connectivity in Default Mode Subnetwork of Subjective Cognitive Decline and Amnestic Mild Cognitive Impairment).

These dynamic communication breakdowns can also be seen in electrophysiological oscillations. In A novel biomarker of amnestic MCI based on dynamic cross-frequency coupling patterns during cognitive brain responses, the authors showed that modeling how different brainwave frequencies interact (cross-frequency coupling) during cognitive tasks yields a highly accurate biomarker for early impairment. To move these insights from expensive labs to pervasive care, Ajay M. Cheriyan et al. designed low-cost embedded hardware for real-time, continuous monitoring of EEG and SpO2 outside the clinic (Pervasive embedded real time monitoring of EEG & SpO2).

Extracting Signal from High-Dimensional Noise

The challenge with functional neuroimaging and molecular data is the "curse of dimensionality"—there are thousands of features but relatively few patient samples. Advanced machine learning and statistical pipelines are required to isolate true predictive signals from biological noise.

For instance, Jaehwan Jung et al. paired Common Component Analysis with penalized logistic regression to successfully classify MCI using high-dimensional fMRI functional connectivity data (Penalized logistic regression using functional connectivity as covariates with an application to mild cognitive impairment). On the structural side, Qi Feng et al. extracted 385 quantitative radiomic features from hippocampus MRI scans, utilizing machine learning to narrow them down to just four highly predictive structural biomarkers for amnestic MCI (Hippocampus Radiomic Biomarkers for the Diagnosis of Amnestic Mild Cognitive Impairment: A Machine Learning Method).

To capture the full picture, modern frameworks rely on multi-source fusion. Yanxian He et al. developed a deep learning framework designed to fuse multimodal neuroimaging data to objectively quantify cognitive function during the critical transition stage (Quantification of Cognitive Function in Alzheimer’s Disease Based on Deep Learning). Finally, these imaging techniques are complemented by scalable molecular assays: Yuek Ling Chai et al. identified plasma osteopontin (OPN) as a robust blood-based biomarker that correlates with early neuroinflammatory pathways and structural brain damage before full-blown dementia sets in (Plasma osteopontin as a biomarker of Alzheimer’s disease and vascular cognitive impairment).

Key Foundational Work

Prediction of Possible conversion from MCI to AD using Machine learning (2012) — contributed 1 concept elements to this direction.
Pervasive embedded real time monitoring of EEG & SpO2 (2009) — contributed 1 concept elements to this direction.
History & citationVersion 1 · 2026-10-04

Current · Permanent link

View frozen version 1

Cite this direction

Latest version (a living page):

@misc{lacuna_preclinical_disease_identification_10136f4d,
  title = {preclinical disease identification},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/preclinical-disease-identification/rdr_10136f4d52df48c49a4f43d64e8c91ce},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-21},
}

This exact version (version 1, frozen text):

@misc{lacuna_preclinical_disease_identification_10136f4d_v1,
  title = {preclinical disease identification},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/preclinical-disease-identification/rdr_10136f4d52df48c49a4f43d64e8c91ce/v/1},
  note = {Version 1, frozen 2026-10-04. Accessed 2026-11-21},
}
About

These directions overlap with this one through shared papers.

See something wrong? Report a mistake