Stricker, John L.

Abstract Introduction Remote cognitive assessments offer an accessible, scalable way to support clinical characterization and longitudinal monitoring of cognition. This study evaluates the feasibility of implementing Mayo Test Drive (MTD), a remote, self-administered digital cognitive assessment, within a specialized clinic focused on eligibility assessments for anti-amyloid therapies for Alzheimer’s disease (AD). Method Two-hundred and nine ( N =209) patients (M age =71.2, 53% female) were evaluated through the Mayo Clinic Rochester Alzheimer’s Disease Treatment Clinic (ADTC) and had the opportunity to complete MTD, either as a remote, patient-initiated session through a patient-facing healthcare portal or during an in-clinic neuropsychological evaluation. Results Clinical diagnoses included 55.2% mild cognitive impairment (MCI), 31.0% mild dementia, and 11.6% moderate dementia, in addition to 2.2% cognitively unimpaired (CU), as determined by neurology consensus at a multidisciplinary case conference. Among those who initiated MTD ( n =197/209), 97.5% completed a session (70.6% remotely, 26.9% in clinic). Completion rates were high across diagnostic groups and did not significantly differ by clinical severity ( p =.12): CU (100%), MCI (99.1%), mild dementia (97.0%), and moderate dementia (89.5%). Remote completion rates overall were high (97.9%). Among completers, administration setting (remote vs. in-clinic) did not significantly differ by clinical stage, χ²(3)=2.37, p =.49. Reasons for non-initiation included not receiving or opening portal messages and other/unknown factors. Discussion Remote digital cognitive assessment is feasible within the context of clinic eligibility evaluations for AD therapies. The high completion rates of MTD across clinical impairment levels in this setting support development of paths for broader use in cognitive characterization in AD care.

Abstract Motor speech disorders (MSDs) are early markers of neurological disease, but expert perceptual analysis is rarely available outside specialized centers. Automated speech analysis offers a scalable alternative, yet prior studies have not systematically compared modeling approaches or assessed clinically relevant metrics in independent datasets. This study compared static acoustic features, articulatory informed Phonet features, and self-supervised pretrained models for binary and multi label MSD classification. We trained and evaluated models on 583 speech samples using speaker level splits. Baseline models included logistic regression and Gated Recurrent Units (GRUs) trained on eGeMAPS and MFCCs. We extracted three types of Phonet derived features and evaluated pretrained HuBERT and SSAST models in frozen, partially fine-tuned, and fully fine-tuned configurations. Binary classification distinguished MSDs from controls, while multi label classification identified six MSD subtypes. Models were assessed using validation AUC, and cut points were tested on two independent datasets. Pretrained and Phonet based models substantially outperformed static acoustic features. In binary classification, HuBERT achieved the highest AUC (0.95), while compact Phonet derived GRUs achieved comparable performance (up to 0.94). These models generalized well to independent datasets, maintaining high sensitivity (0.94) and specificity (0.97). In multi label classification, Phonet models achieved the highest macro average AUC (0.86), but threshold-based subtype performance declined on unseen data. Automated MSD detection is feasible and clinically promising. Binary classification generalized well, whereas multi label classification showed limited threshold stability across datasets.

Abstract INTRODUCTION We describe the reliability of remote self‐administered digital cognitive measures completed via the Mayo Test Drive (MTD) web‐based platform. METHODS A total of 1846 participants (mean age = 70, SD = 12, range 31–101; 48% male; 96% White; 99% non‐Hispanic; 97% cognitively unimpaired) with two to four complete MTD sessions at ∼7.5‐month intervals were included. Test‐retest reliability was assessed using single‐rating, absolute‐agreement, and two‐way mixed intraclass correlation coefficients (ICCs) with 95% confidence intervals. ICCs for in‐person‐administered traditional neuropsychological measures were compared to MTD for a subset of 244 participants. Analyses also compared reliability across demographic, residential, and session context subgroups. RESULTS Reliability was good for the MTD Composite (total ICC = 0.79 [0.77, 0.80]), and moderate‐to‐good for the primary outcome variables for each MTD subtest (total ICCs 0.70–0.83 for Stricker Learning Span and Symbols). The reliability of the remote self‐administered MTD was similar to in‐person‐administered cognitive measures. DISCUSSION MTD showed moderate‐to‐good reliability, supporting its use in longitudinal monitoring.

Abstract Background To promote equitable access to neurologic diagnostics, our team developed MASTERMIND—the Mobile Assessment of Speech To Evaluate, Recognize, and Monitor Indicators of Neurologic health and Disease. One subtest, the MASTERMIND Assessment of Naming, includes ten pictures randomly selected from a list of 45 words varying in frequency. This study aimed to 1) evaluate the feasibility of a remotely self‐administered naming test for patients with cognitive, linguistic, and/or communication difficulties and 2) assess the sensitivity and specificity of performance measures (accuracy, error type, and response time) for detecting aphasia and cognitive‐communication difficulties. Method The task was remotely administered to 138 controls, 26 cognitively impaired individuals, 15 with motor speech disorders, 8 with cognitive concerns, and 6 with aphasia. Participants named pictures displayed on‐screen, advancing through the task by clicking “Next.” Responses were processed using natural language processing (NLP). Audio was transcribed with Whisper Large‐v3, and responses were matched to stimuli within a 0.4‐second threshold. Errors were categorized by type (semantic, phonemic, or unclear source) through consensus between two raters. Performance metrics (accuracy, error type, response time) were analyzed using Kruskal‐Wallis and Pearson's Chi‐Square tests, with AUROC analysis evaluating diagnostic performance. Results Differences in accuracy, response time, and error type were observed across groups (Table 1). Controls and cognitively impaired participants performed similarly, with longer response times observed in patients with aphasia and motor speech disorders. Accuracy alone was a poor predictor of all patients versus controls (AUROC = 0.468) but prediction was better in distinguishing aphasia patients from 1) controls (AUROC = 0.698), 2) cognitively impaired participants (AUROC = 0.679), and 3) those with MSDs (AUROC = 0.670). Conclusions The MASTERMIND Assessment of Naming is a feasible, scalable, self‐administered diagnostic tool enhanced by automated NLP. It provides critical insights through response time, accuracy, and error analysis. Limitations include potential bias in error classification and fixed processing thresholds. The study highlights MASTERMIND's potential for patient triaging and screening. Future research will evaluate the sensitivity of combined performance metrics to disease progression, expand sample sizes and diversity, and refine usability based on participant feedback.

Abstract Background Self‐administered and remote digital cognitive assessments provide a scalable method for screening and longitudinal cognitive monitoring. Mayo Test Drive (MTD) is one such assessment with previously demonstrated usability, reliability, validity and multi‐device compatibility. MTD includes a computer‐adaptive word list memory test (Stricker Learning Span) and a measure of processing speed/executive functioning (Symbols Test) that combine to provide an MTD Composite. This study aimed to develop a conditional normative model predicting longitudinal cognitive change for MTD, and to validate the model in two independent samples of cognitively unimpaired (CU) and mild cognitive impairment (MCI)/dementia (DEM) participants. Method 1355 CU participants at initial MTD assessment with 2‐4 consecutive MTD assessments every 7.5 months comprised the development dataset. Linear mixed effects models were used to construct a conditional normative model for MTD assessments 2‐4. The baseline model was chosen a priori , consisting of age, sex, education, baseline score, number of assessments, and time. Device type, race, and baseline score by time interactions were also considered. Variables were only added if a >1% increment in r‐squared was observed. The models were externally validated in 113 CU and 51 MCI/DEM patients. Following model development, the data were updated and additional follow‐up MTD assessments were added for many participants, including the addition of additional CU validation data at test 3 ( N = 126) and 4 ( N = 279). T‐tests compared observed and predicted differences and residual z‐scores between development and validation at each follow‐up assessment. Result Table 1 presents demographic variables. No additional variables increased the r‐squared by >1%. The final models are presented in Table 2. Age, baseline score, sex, and education were significant. Residual z‐scores were not significantly different between CU development and CU validation cohorts, indicating observed scores were in line with model predicted scores. Residual z‐scores were significantly worse among MCI/DEM compared to CU development in 11/12 comparisons (Figure 1), indicating observed scores were below model predicted expectations and implying sensitivity to longitudinal cognitive decline. Conclusion These longitudinal normative models can be used to identify individuals with follow‐up scores outside of expected predicted ranges, signifying abnormal longitudinal cognitive decline.

Abstract Background Neuropathologic evaluation is essential for diagnosing and treating dementia, but definitive diagnoses is only available after post‐mortem brain autopsy. Positron emission tomography (PET) offers a potential in vivo proxy for predicting neuropathology. This study proposes a semi‐supervised learning (SSL) model using a single FDG‐PET image to predict underlying neuropathology. Method The SSL model leverages a self‐organized graph structure based on a patient similarity network (PSN) created from 1,495 FDG‐PET images across the aging and neurodegenerative spectrum. Each patient is represented by a node, with edges indicating image similarity derived from principal component analysis features and cosine similarity weights. The PSN structure connects 20 nearest neighbors. Neuropathologic labels were available for 204 autopsied patients. The Poisson Learning Method was employed to propagate labels from these annotated nodes to unlabeled nodes, guided by the PSN structure. FDG‐PET images were collected from clinical and research participants at the Mayo Clinic and processed using an MRI‐free pipeline. Neuropathologic labels were assigned based on widely accepted diagnostic criteria. Leave‐one‐out cross‐validation (LOOCV) was used to assess model performance. Results The confusion matrix demonstrated high accuracy in predicting Alzheimer's disease, frontotemporal lobar degeneration with tau inclusions, and Lewy body disease pathology. A Sankey diagram and label distribution analysis revealed strong correlations between clinical and predicted neuropathologic labels. Conclusion This study highlights the potential of a graph‐based SSL pipeline to predict neuropathological changes using FDG‐PET imaging data, despite limited pathology labels. The findings underscore the value of SSL in bridging the gap between abundant neuroimaging data and scarce neuropathologic labels, contributing to improved dementia diagnosis and understanding.

e13891 Background: Most patients with central nervous system (CNS) malignancy develop cognitive impairment, fatigue, and mood-related symptoms that are impacted by cancer and treatment, including radiation therapy. We currently have limited therapies to mitigate neurocognitive toxicity. It is important to accurately assess neurocognitive function and symptom burden when working to establish therapeutics. The objective of this trial is to assess the effectiveness of therapies aimed at improving neurocognitive function and quality of life by leveraging a decentralized approach incorporating an established digital neurocognitive assessment tool and development of a remote symptom collection/medication log to improve capture of longitudinal neurocognitive data. Methods: This is an open-label decentralized clinical trial (NCT06377696) that randomizes patients who have received intracranial radiation for primary and secondary brain tumors in a 1:1 allocation to health promotion with a wearable device +/- metformin (targeting radiation-related toxicity). Remote symptom collection is reported through digital patient surveys. Medication adherence is being monitored through a digital medication log developed for the study. Remote neurologic examination is performed by clinical neuro-oncologists at telehealth visits. Participants perform serial remote cognitive assessments with an established digital neurocognitive testing platform – Mayo Test Drive (MTD). Cognitive assessments and symptom surveys were designed to be completed on a smartphone, tablet, or computer device. Results: The study started enrollment on May 31, 2024. As of January 27, 2025, 54 of 100 planned participants have enrolled (27 on each arm). 51 participants remain active on trial; 3 withdrew from trial participation (2 prior to treatment, 1 after treatment). The overall completion rate for a quality-of-life questionnaire (EORTC QLQ-C30) is 85.0%, and each dose of metformin logged is 78.5%. The overall completion rate for virtual neurologic examination and short test of mental status is 94.7%. For MTD, the overall completion rate is 97.9%. The average MTD session duration was 16 minutes. Most participants completed MTD testing using a smartphone (90.5%) at home (89.1%). Conclusions: These results provide preliminary support that a decentralized trial assessing neurocognitive outcomes is feasible in patients with CNS malignancy. The completion rates of questionnaires and objective cognitive evaluation are comparable or higher than expected with historic norms. Final study results will provide real-world cognitive and quality-of-life data for patients with CNS tumors while providing preliminary data on the effectiveness of device-based health promotion +/- metformin for treatment related cognitive decline. Clinical trial information: NCT06377696 .