Automated Self-Administered Scale (AMES)

1. Abstract

The Automated Self-Administered Scale (AMES), developed by Huang, Mei, Ye, and Guo (2023), represents a significant psychometric and technological advancement in computerized neuropsychological assessment. Engineered specifically for early cognitive impairment screening across community-based and primary healthcare settings, the instrument provides an objective, automated assessment pipeline administered via portable touchscreen Android tablets. The AMES quantifies multi-domain neurocognitive functioning across four core structural subtests: Guided Learning, Go/No-Go, Number Tabbing, and Delayed Recall, spanning a total of 16 psychometric items. Conceptually grounded in memory binding architectures and Parallel Distributed Processing (PDP) connectionist paradigms, the scale implements cross-modal semantic cueing and deep encoding methodologies designed to dissociate the distinct cognitive contributions of familiarity from recollection in episodic retrieval.

To eliminate testing barriers associated with low literacy, visual decline, and sensory-motor impairments common in aging cohorts, the AMES minimizes textual overhead by integrating slow-paced, standardized automated audio instructions and streamlined motoric interactions restricted to direct screen tapping. Empirical validation in cohorts comprising normal controls (NC), subjective cognitive decline (SCD), objectively-defined subtle cognitive decline (obj-SCD), and mild cognitive impairment (MCI) establishes robust diagnostic accuracy. The scale demonstrates excellent discriminative power, yielding an area under the receiver operating characteristic curve (AUC) of 0.88 (sensitivity: 86%, specificity: 80%) for differentiating MCI from NC, and an AUC of 0.78 (sensitivity: 89%, specificity: 63%) for distinguishing obj-SCD from NC. Convergent validity is confirmed via significant correlations with gold-standard assessments, including the Mini-Mental State Examination (MMSE; r = 0.436) and the Montreal Cognitive Assessment Basic (MoCA-B; r = 0.551). Completing in approximately 10 minutes, the AMES provides a scalable, standardized, and clinically viable digital biomarker tool for wide-scale preclinical dementia triage.

2. Keywords

Automated Self-Administered Scale, AMES, Cognitive Screening Tool, Mild Cognitive Impairment, Preclinical Dementia, Memory Binding, Parallel Distributed Processing, Executive Function, Guided Learning, Tablet-Based Assessment

3. Authors

The Automated Self-Administered Scale was conceptualized, developed, and clinically validated by a team of neuroscientists and clinical psychometricians affiliated with Shanghai Jiao Tong University School of Medicine:

• Lin Huang — Department of Gerontology, Shanghai Sixth People’s Hospital Affiliated to Shanghai Jiao Tong University School of Medicine, Shanghai, China.
• Zihan Mei — Department of Gerontology, Shanghai Sixth People’s Hospital Affiliated to Shanghai Jiao Tong University School of Medicine, Shanghai, China. (ORCID: 0000-0002-8552-0034).
• Jianfeng Ye — Department of Gerontology, Shanghai Sixth People’s Hospital Affiliated to Shanghai Jiao Tong University School of Medicine, Shanghai, China.
• Qihao Guo (Corresponding Author) — Department of Gerontology, Shanghai Sixth People’s Hospital Affiliated to Shanghai Jiao Tong University School of Medicine, No. 600, Yi Shan Road, Shanghai, 200233, China. Email: [email protected].

4. Purpose

The primary clinical and psychometric objective of the Automated Self-Administered Scale (AMES) is to establish a rapid, standardized, and culturally robust screening instrument capable of detecting incipient cognitive decline within primary care, community health centers, and population-level epidemiological studies. Epidemiological evidence highlights that early pathological hallmarks of neurodegenerative disorders, such as Alzheimer’s disease, emerge decades prior to the manifestation of overt syndromic dementia. Interventions targeted at secondary prevention require precise detection instruments sensitive to subtle neurocognitive deficits, specifically within the transitional stages characterized by Subjective Cognitive Decline (SCD) and Objectively-Defined Subtle Cognitive Decline (obj-SCD).

Traditional bedside screening instruments, including the Mini-Mental State Examination (MMSE) and the Montreal Cognitive Assessment (MoCA), often manifest pronounced psychometric limitations when deployed across decentralized community environments. These conventional tests frequently exhibit substantial ceiling effects among highly educated older adults, susceptibility to examiner-dependent administration and scoring biases, and pronounced vulnerability to demographic confounding factors such as low literacy, educational disparities, and sensory-motor impairments. In addition, standard paper-and-pencil assessments require dedicated administration time by trained clinical neuropsychologists, creating bottlenecks in resource-constrained primary care environments.

The AMES systematically addresses these methodological challenges by deploying fully autonomous, tablet-based digital screening. Designed to eliminate examiner administration variance, the tool standardizes instruction delivery through integrated, slow-cadence spoken audio, removing reliance on reading proficiency or visual acuity. Psychometrically, the purpose of the AMES extends beyond blunt binary screening; it aims to fractionate early cognitive failure into granular neurocognitive subsystems—specifically isolating semantic binding deficits, response inhibition failure, attentional allocation disruptions, and long-term consolidation deficits. By finishing the evaluation protocol within approximately 10 minutes, the AMES equips healthcare systems with an efficient triage mechanism to determine which community-dwelling older adults require advanced molecular, imaging, or comprehensive neuropsychological evaluations.

5. Psychological Construct

The AMES is structured as a multi-component cognitive battery designed to measure distinct yet interacting neuropsychological constructs. The scale synthesizes four functional subtests totaling 16 psychometric items, targeted at capturing early breakdowns in distributed neural networks.

1. Guided Learning (Deep Semantic Encoding and Feature Binding)

This subscale evaluates the capacity of episodic memory to create integrated representational representations via category-guided semantic binding. Rather than assessing unprompted memorization, the Guided Learning task presents target stimuli paired with category or semantic cues. This construct taps into hippocampal and medial temporal lobe networks responsible for conjunction formation, as well as prefrontal mechanisms driving deep semantic elaboration. By examining an individual’s capacity to exploit semantic cues during initial presentation, this domain differentiates between failures of superficial sensory registration and deeper deficits in relational associative binding.

2. Go/No-Go (Inhibitory Control and Motor Executive Regulation)

The Go/No-Go subtest operationalizes behavioral and motor response inhibition, a core facet of executive function mediated by the prefrontal cortex, frontostriatal loops, and the anterior cingulate cortex. Participants are conditioned to execute a rapid motor response (screen tap) to specified target stimuli (“Go”) while withholding responses upon the presentation of non-target lure stimuli (“No-Go”). This measure assesses the suppression of prepotent motor impulses, continuous attentional vigilance, and processing speed, capturing early executive dysregulation that frequently accompanies subcortical microvascular pathology and frontoparietal disconnection.

3. Number Tabbing (Working Memory, Psychomotor Speed, and Visual Scanning)

Operationalizing psychomotor coordination, visuomotor scanning, and active working memory updating, the Number Tabbing subtest requires rapid identification and sequential motor interaction with numerical visual targets. This cognitive construct reflects the integrity of dorsal visual processing pathways and distributed frontoparietal networks. It quantifies how efficiently an individual can maintain temporal sequences within the phonological and visuospatial sketchpad while simultaneously coordinating rapid, fine-motor responses on a digital interface.

4. Delayed Recall (Consolidation and Long-Term Episodic Retrieval)

The Delayed Recall subtest assesses the durability of memory consolidation and retrieval fidelity following an intervening cognitive interference period. Measuring the integrity of hippocampal synaptic plasticity and reciprocal entorhinal-cortical connectivity, this construct determines whether learned associations can be reconstructed without immediate sensory support. The subtest specifically enables the empirical dissociation of recollection-based retrieval (retrieving contextual, associative details) from familiarity-driven recognition, providing sensitive markers for incipient neurodegeneration.

6. Theoretical Framework

The architectural design of the AMES is anchored in two complementary neurocognitive paradigms: the Relational Memory Binding Hypothesis and connectionist Parallel Distributed Processing (PDP) models of dual-process recognition memory.

The Relational Memory Binding Framework

Classical cognitive theory posits that human episodic memory is fundamentally associative; the encoding of an event requires the rapid binding of disparate perceptual, spatial, temporal, and semantic elements into a coherent multimodal representation. Pioneering neuropsychological models demonstrate that the hippocampus and parahippocampal gyrus are critically specialized for this binding computation. In preclinical Alzheimer’s disease and incipient MCI, synaptic disconnection typically manifests first within the transentorhinal and entorhinal cortices, resulting in a breakdown of relational binding mechanisms before generalized item-memory decay becomes apparent.

The AMES operationalizes this framework in its Guided Learning subtest. By systematically pairing target items with categorical and semantic cues, the scale optimizes deep semantic encoding via semantic scaffolding. This paradigm minimizes incidental variations in spontaneous strategic organization, isolating genuine associative binding capacity from organizational or attentional retrieval deficits.

Parallel Distributed Processing (PDP) and Dual-Process Theory

The theoretical architecture of the AMES’s memory assessment is further grounded in Parallel Distributed Processing (PDP) connectionist network theory, as synthesized with dual-process models of recognition memory (formalized by theorists such as Mandler, Yonelinas, and McClelland). Dual-process frameworks argue that recognition judgments are supported by two distinct neurocomputational mechanisms:

• Familiarity: A rapid, continuous, quantitative assessment of perceptual or semantic similarity that operates independently of contextual recall, largely supported by perirhinal and lateral entorhinal cortices.
• Recollection: A thresholded, qualitative, and computationally demanding retrieval process that reconstructs precise contextual and associative elements of the encoding episode, reliant on the integrity of the hippocampus and the prefrontal cortex.

By applying PDP-informed analytical paradigms to response patterns across Guided Learning and Delayed Recall, the AMES systematically models recognition behavior. This mathematical differentiation ensures that individuals who compensate for recollection degradation via intact perceptual familiarity can still be identified early in their neurodegenerative trajectory, elevating the tool’s clinical sensitivity relative to conventional additive memory batteries.

7. Validity

Empirical validation of the AMES has been established through clinical validation trials involving diverse participant cohorts categorized into Normal Controls (NC), Subjective Cognitive Decline (SCD), Objectively-Defined Subtle Cognitive Decline (obj-SCD), and Mild Cognitive Impairment (MCI) in China (Huang et al., 2023).

Convergent and Criterion Validity

Convergent validity was evaluated by correlating AMES total performance scores against established neuropsychological assessment instruments and standardized domain-specific cognitive tests:

• Global Cognition Measures: The AMES exhibited strong, statistically significant positive correlations with the Mini-Mental State Examination (MMSE; Spearman rank correlation coefficient r = 0.436, p < .001) and the Montreal Cognitive Assessment Basic (MoCA-B; r = 0.551, p < .001).
• Memory Function: AMES scores correlated positively with standard Delayed Free Recall tests (r = 0.542, p < .001) and Memory Recognition tasks (r = 0.492, p < .001).
• Language and Fluency: Significant positive associations were observed with Animal Verbal Fluency tests (r = 0.306, p < .001) and standardized language index batteries (r = 0.300, p < .001).
• Attention and Executive Speed: Statistically significant inverse correlations emerged when evaluating timed error scores on attention tasks (r = −0.360, p < .001) and executive function completion matrices (r = −0.484, p < .001), indicating that higher performance on the AMES systematically maps to fewer executive errors and reduced reaction latency.

Diagnostic Sensitivity and Specificity (ROC Analysis)

Receiver Operating Characteristic (ROC) curve analyses demonstrate excellent discriminative efficacy for detecting incipient cognitive decline:

• MCI vs. Normal Controls: In distinguishing clinical patients diagnosed with Mild Cognitive Impairment (MCI) from cognitively intact Normal Controls (NC), the AMES demonstrated an Area Under the Curve (AUC) of 0.88, accompanied by an optimal diagnostic sensitivity of 86% and a specificity of 80%.
• Obj-SCD vs. Normal Controls: In the clinically challenging differentiation between individuals presenting with Objectively-Defined Subtle Cognitive Decline (obj-SCD) and Normal Controls, the AMES maintained an AUC of 0.78, yielding a diagnostic sensitivity of 89% and a specificity of 63%.

8. Reliability

The internal consistency of the AMES subscale modules was evaluated using Cronbach’s alpha coefficient within the clinical validation sample (Huang et al., 2023). Because the scale utilizes brief, functionally distinct, multi-component subtests designed to minimize administrative burden, the calculated internal consistency parameters show variability across individual sub-domains:

• Language Processing Ability (Items 1 and 2): Cronbach’s α = 0.17.
• Immediate Memory (Items 3 and 4): Cronbach’s α = 0.38.
• Executive Function (Items 5, 6, 7, and 8): Cronbach’s α = 0.52.

From a psychometric perspective, low Cronbach’s alpha coefficients across extremely brief subscales (two to four items per subscale) are widely documented in the neuropsychological measurement literature. In multidimensional clinical assessment tools comprising very few items, alpha values underestimate reliability due to its direct mathematical dependence on test length. Furthermore, cognitive screening subscales frequently capture heterogeneous cognitive sub-processes (such as motor execution, perceptual filtering, and cognitive switching within executive function) rather than unifactorial behavioral self-reports. Despite low sub-domain alpha coefficients, the scale’s aggregate diagnostic accuracy (AUC = 0.88), robust criterion validity (r = 0.551 with MoCA-B), and automated scoring protocols support its screening utility.

9. Factor Analysis

Formal exploratory factor analysis (EFA) and confirmatory factor analysis (CFA) structural equation modeling data have not been reported in the primary validation literature for the AMES (Huang et al., 2023). The instrument’s four-subtest architecture (Guided Learning, Go/No-Go, Number Tabbing, Delayed Recall) was constructed on theoretical grounds using established neurofunctional dissociations in clinical neuropsychology rather than derived empirically through latent exploratory factor extraction.

Future structural validity research utilizing larger, multi-center international cohorts is necessary to empirically establish the latent dimensionality of the 16 items. Specifically, confirmatory factor analyses comparing single-factor models of general cognitive ability (g) against hierarchical, multi-factor models (isolating episodic encoding, retrieval consolidation, and fronto-executive inhibition) will help establish structural invariance across differing linguistic, cultural, and educational strata.

10. Instrument / Measurement Tool

The operational specifications, administration protocols, and scoring metrics for the Automated Self-Administered Scale (AMES) are structured as follows:

• Test Type: Original automated cognitive screening inventory / objective digital neuropsychological test battery.
• Delivery Format: Administered electronically via portable Android tablet devices. The testing interface utilizes standardized, slow-paced spoken audio instructions to eliminate reading barriers, paired with minimal on-screen text to accommodate individuals with visual impairments.
• Item Count: 16 psychometric items organized across 4 dedicated subtests.
• Subtests:
  • Guided Learning: Paired-stimulus presentation utilizing semantic and categorical cues for deep encoding.
  • Go/No-Go: Motor response inhibition and continuous vigilance task.
  • Number Tabbing: Sequenced visual scanning, psychomotor speed, and active working memory.
  • Delayed Recall: Interrupted retrieval latency protocol measuring consolidation and familiarity-recollection distinction.
• Response Scale and Interaction Mode: The AMES is administered on Android tablets. User interactions are designed to be simple, primarily involving clicking or tabbing. For each pair of items, the sum of item scores corresponds to the maximum attempts allowed in the respective line, and coefficients are utilized in the calculation of the total score.
• Scoring Mechanism: Direct tablet recording of reaction latency, error frequencies, and successful task completion across repeated attempts. Subtest metrics are weighted via regression coefficients to generate an aggregate composite index indicating cognitive risk.
• Administration Time: Approximately 10 minutes for full autonomous completion.
• Target Population: Adults (18 years and older) and middle-aged/older adults (40–64+ years) evaluated across community, primary care, or memory clinic settings.

11. Permissions & Fee and Test Year

The Automated Self-Administered Scale was introduced and published in 2023 by Lin Huang, Zihan Mei, Jianfeng Ye, and Qihao Guo. The development of the tool was conducted within the Department of Gerontology at Shanghai Sixth People’s Hospital Affiliated to Shanghai Jiao Tong University School of Medicine.

• Commercial Status: Non-commercial instrument developed for clinical and scientific research.
• Access Fee: There is no fee required for academic or non-commercial research utilization.
• Permissions and Software Licensing: The AMES software application, tablet deployment packages, and associated administration algorithms are proprietary. Researchers and healthcare systems seeking to deploy the scale must contact the corresponding author, Dr. Qihao Guo (Department of Gerontology, Shanghai Sixth People’s Hospital, Shanghai Jiao Tong University School of Medicine, No. 600, Yi Shan Road, Shanghai, 200233, China; Email: [email protected]).

12. References

The following peer-reviewed literature provides the empirical foundation, psychometric parameters, and theoretical background for the AMES:

• Huang, L., Mei, Z., Ye, J., & Guo, Q. (2023). AMES: An automated self-administered scale to detect incipient cognitive decline in primary care settings. Assessment, 30(7), 2247–2257. https://doi.org/10.1177/10731911221144774
• Julayanont, P., Tangwongchai, S., Hemrungrojn, S., Tunvirachaisakul, C., Phanthumchinda, K., Hongsawat, J., Suwichanarakul, S., Thanasirorat, S., & Nasreddine, Z. S. (2015). The Montreal Cognitive Assessment-Basic: A screening tool for mild cognitive impairment in illiterate and low-educated elderly adults. Journal of the American Geriatrics Society, 63(12), 2550–2554. https://doi.org/10.1111/jgs.13820
• Nasreddine, Z. S., Phillips, N. A., Bédirian, V., Charbonneau, S., Whitehead, V., Collin, I., Cummings, J. L., & Chertkow, H. (2005). The Montreal Cognitive Assessment, MoCA: A brief screening tool for mild cognitive impairment. Journal of the American Geriatrics Society, 53(4), 695–699. https://doi.org/10.1111/j.1532-5415.2005.53221.x
• Yonelinas, A. P. (2002). The nature of recollection and familiarity: A review of 30 years of research. Journal of Memory and Language, 46(3), 441–517. https://doi.org/10.1006/jmla.2001.2864

13. Items of the Scale