AI-based differential diagnosis of dementia etiologies on multimodal data.

Differential diagnosis of dementia remains a challenge in neurology due to symptom overlap across etiologies, yet it is crucial for formulating early, personalized management strategies. Here, we present an AI model that harnesses a broad array of data, including demographics, individual and family medical history, medication use, neuropsychological assessments, functional evaluations, and multimodal neuroimaging, to identify the etiologies contributing to dementia in individuals. The study, drawing on 51,269 participants across 9 independent, geographically diverse datasets, facilitated the identification of 10 distinct dementia etiologies. It aligns diagnoses with similar management strategies, ensuring robust predictions even with incomplete data. Our model achieved a micro-averaged area under the receiver operating characteristic curve (AUROC) of 0.94 in classifying individuals with normal cognition, mild cognitive impairment and dementia. Also, the micro-averaged AUROC was 0.96 in differentiating the dementia etiologies. Our model demonstrated proficiency in addressing mixed dementia cases, with a mean AUROC of 0.78 for two co-occurring pathologies. In a randomly selected subset of 100 cases, the AUROC of neurologist assessments augmented by our AI model exceeded neurologist-only evaluations by 26.25%. Furthermore, our model predictions aligned with biomarker evidence and its associations with different proteinopathies were substantiated through postmortem findings. Our framework has the potential to be integrated as a screening tool for dementia in various clinical settings and drug trials, with promising implications for person-level management.

Exploring cognitive progression subtypes in the Framingham Heart Study.

INTRODUCTION: Alzheimer's disease (AD) is a heterogeneous disorder characterized by complex underlying neuropathology that is not fully understood. This study aimed to identify cognitive progression subtypes and examine their correlation with clinical outcomes. METHODS: Participants of this study were recruited from the Framingham Heart Study. The Subtype and Stage Inference (SuStaIn) method was used to identify cognitive progression subtypes based on eight cognitive domains. RESULTS: Three cognitive progression subtypes were identified, including verbal learning (Subtype 1), abstract reasoning (Subtype 2), and visual memory (Subtype 3). These subtypes represent different domains of cognitive decline during the progression of AD. Significant differences in age of onset among the different subtypes were also observed. A higher SuStaIn stage was significantly associated with increased mortality risk. DISCUSSION: This study provides a characterization of AD heterogeneity in cognitive progression, emphasizing the importance of developing personalized approaches for risk stratification and intervention. Highlights: We used the Subtype and Stage Inference (SuStaIn) method to identify three cognitive progression subtypes.Different subtypes have significant variations in age of onset.Higher stages of progression are associated with increased mortality risk.

Speech patterns during memory recall relates to early tau burden across adulthood.

INTRODUCTION: Early cognitive decline may manifest in subtle differences in speech. METHODS: We examined 238 cognitively unimpaired adults from the Framingham Heart Study (32-75 years) who completed amyloid and tau PET imaging. Speech patterns during delayed recall of a story memory task were quantified via five speech markers, and their associations with global amyloid status and regional tau signal were examined. RESULTS: Total utterance time, number of between-utterance pauses, speech rate, and percentage of unique words significantly correlated with delayed recall score although the shared variance was low (2%-15%). Delayed recall score was not significantly different between ß-amyoid-positive (Aß+) and -negative (Aß-) groups and was not associated with regional tau signal. However, longer and more between-utterance pauses, and slower speech rate were associated with increased tau signal across medial temporal and early neocortical regions. DISCUSSION: Subtle speech changes during memory recall may reflect cognitive impairment associated with early Alzheimer's disease pathology. HIGHLIGHTS: Speech during delayed memory recall relates to tau PET signal across adulthood. Delayed memory recall score was not associated with tau PET signal. Speech shows greater sensitivity to detecting subtle cognitive changes associated with early tau accumulation. Our cohort spans adulthood, while most PET imaging studies focus on older adults.

Design and Feasibility Analysis of a Smartphone-Based Digital Cognitive Assessment Study in the Framingham Heart Study.

BACKGROUND: Smartphone-based digital technology is increasingly being recognized as a cost-effective, scalable, and noninvasive method of collecting longitudinal cognitive and behavioral data. Accordingly, a state-of-the-art 3-year longitudinal project focused on collecting multimodal digital data for early detection of cognitive impairment was developed. METHODS AND RESULTS: A smartphone application collected 2 modalities of cognitive data, digital voice and screen-based behaviors, from the FHS (Framingham Heart Study) multigenerational Generation 2 (Gen 2) and Generation 3 (Gen 3) cohorts. To understand the feasibility of conducting a smartphone-based study, participants completed a series of questions about their smartphone and app use, as well as sensory and environmental factors that they encountered while completing the tasks on the app. Baseline data collected to date were from 537 participants (mean age=66.6 years, SD=7.0; 58.47% female). Across the younger participants from the Gen 3 cohort (n=455; mean age=60.8 years, SD=8.2; 59.12% female) and older participants from the Gen 2 cohort (n=82; mean age=74.2 years, SD=5.8; 54.88% female), an average of 76% participants agreed or strongly agreed that they felt confident about using the app, 77% on average agreed or strongly agreed that they were able to use the app on their own, and 81% on average rated the app as easy to use. CONCLUSIONS: Based on participant ratings, the study findings are promising. At baseline, the majority of participants are able to complete the app-related tasks, follow the instructions, and encounter minimal barriers to completing the tasks independently. These data provide evidence that designing and collecting smartphone application data in an unsupervised, remote, and naturalistic setting in a large, community-based population is feasible.

Comparing Cognitive Tests and Smartphone-Based Assessment in 2 US Community-Based Cohorts.

BACKGROUND: Smartphone-based cognitive assessments have emerged as promising tools, bridging gaps in accessibility and reducing bias in Alzheimer disease and related dementia research. However, their congruence with traditional neuropsychological tests and usefulness in diverse cohorts remain underexplored. METHODS AND RESULTS: A total of 406 FHS (Framingham Heart Study) and 59 BHS (Bogalusa Heart Study) participants with traditional neuropsychological tests and digital assessments using the Defense Automated Neurocognitive Assessment (DANA) smartphone protocol were included. Regression models investigated associations between DANA task digital measures and a neuropsychological global cognitive Z score (Global Cognitive Score [GCS]), and neuropsychological domain-specific Z scores. FHS participants' mean age was 57 (SD, 9.75) years, and 44% (179) were men. BHS participants' mean age was 49 (4.4) years, and 28% (16) were men. Participants in both cohorts with the lowest neuropsychological performance (lowest quartile, GCS1) demonstrated lower DANA digital scores. In the FHS, GCS1 participants had slower average response times and decreased cognitive efficiency scores in all DANA tasks (P<0.05). In BHS, participants in GCS1 had slower average response times and decreased cognitive efficiency scores for DANA Code Substitution and Go/No-Go tasks, although this was not statistically significant. In both cohorts, GCS was significantly associated with DANA tasks, such that higher GCS correlated with faster average response times (P<0.05) and increased cognitive efficiency (all P<0.05) in the DANA Code Substitution task. CONCLUSIONS: Our findings demonstrate that smartphone-based cognitive assessments exhibit concurrent validity with a composite measure of traditional neuropsychological tests. This supports the potential of using smartphone-based assessments in cognitive screening across diverse populations and the scalability of digital assessments to community-dwelling individuals.

Fusion of Low-Level Descriptors of Digital Voice Recordings for Dementia Assessment.

Digital voice recordings can offer affordable, accessible ways to evaluate behavior and function. We assessed how combining different low-level voice descriptors can evaluate cognitive status. Using voice recordings from neuropsychological exams at the Framingham Heart Study, we developed a machine learning framework fusing spectral, prosodic, and sound quality measures early in the training cycle. The model's area under the receiver operating characteristic curve was 0.832 (±0.034) in differentiating persons with dementia from those who had normal cognition. This offers a data-driven framework for analyzing minimally processed voice recordings for cognitive assessment, highlighting the value of digital technologies in disease detection and intervention.

Multimodal Machine Learning for 10-Year Dementia Risk Prediction: The Framingham Heart Study.

BACKGROUND: Early prediction of dementia risk is crucial for effective interventions. Given the known etiologic heterogeneity, machine learning methods leveraging multimodal data, such as clinical manifestations, neuroimaging biomarkers, and well-documented risk factors, could predict dementia more accurately than single modal data. OBJECTIVE: This study aims to develop machine learning models that capitalize on neuropsychological (NP) tests, magnetic resonance imaging (MRI) measures, and clinical risk factors for 10-year dementia prediction. METHODS: This study included participants from the Framingham Heart Study, and various data modalities such as NP tests, MRI measures, and demographic variables were collected. CatBoost was used with Optuna hyperparameter optimization to create prediction models for 10-year dementia risk using different combinations of data modalities. The contribution of each modality and feature for the prediction task was also quantified using Shapley values. RESULTS: This study included 1,031 participants with normal cognitive status at baseline (age 75±5 years, 55.3% women), of whom 205 were diagnosed with dementia during the 10-year follow-up. The model built on three modalities demonstrated the best dementia prediction performance (AUC 0.90±0.01) compared to single modality models (AUC range: 0.82-0.84). MRI measures contributed most to dementia prediction (mean absolute Shapley value: 3.19), suggesting the necessity of multimodal inputs. CONCLUSION: This study shows that a multimodal machine learning framework had a superior performance for 10-year dementia risk prediction. The model can be used to increase vigilance for cognitive deterioration and select high-risk individuals for early intervention and risk management.

Association Between Acoustic Features and Neuropsychological Test Performance in the Framingham Heart Study: Observational Study.

BACKGROUND: Human voice has increasingly been recognized as an effective indicator for the detection of cognitive disorders. However, the association of acoustic features with specific cognitive functions and mild cognitive impairment (MCI) has yet to be evaluated in a large community-based population. OBJECTIVE: This study aimed to investigate the association between acoustic features and neuropsychological (NP) tests across multiple cognitive domains and evaluate the added predictive power of acoustic composite scores for the classification of MCI. METHODS: This study included participants without dementia from the Framingham Heart Study, a large community-based cohort with longitudinal surveillance for incident dementia. For each participant, 65 low-level acoustic descriptors were derived from voice recordings of NP test administration. The associations between individual acoustic descriptors and 18 NP tests were assessed with linear mixed-effect models adjusted for age, sex, and education. Acoustic composite scores were then built by combining acoustic features significantly associated with NP tests. The added prediction power of acoustic composite scores for prevalent and incident MCI was also evaluated. RESULTS: The study included 7874 voice recordings from 4950 participants (age: mean 62, SD 14 years; 4336/7874, 55.07% women), of whom 453 were diagnosed with MCI. In all, 8 NP tests were associated with more than 15 acoustic features after adjusting for multiple testing. Additionally, 4 of the acoustic composite scores were significantly associated with prevalent MCI and 7 were associated with incident MCI. The acoustic composite scores can increase the area under the curve of the baseline model for MCI prediction from 0.712 to 0.755. CONCLUSIONS: Multiple acoustic features are significantly associated with NP test performance and MCI, which can potentially be used as digital biomarkers for early cognitive impairment monitoring.

Automated detection of mild cognitive impairment and dementia from voice recordings: A natural language processing approach.

INTRODUCTION: Automated computational assessment of neuropsychological tests would enable widespread, cost-effective screening for dementia. METHODS: A novel natural language processing approach is developed and validated to identify different stages of dementia based on automated transcription of digital voice recordings of subjects' neuropsychological tests conducted by the Framingham Heart Study (n = 1084). Transcribed sentences from the test were encoded into quantitative data and several models were trained and tested using these data and the participants' demographic characteristics. RESULTS: Average area under the curve (AUC) on the held-out test data reached 92.6%, 88.0%, and 74.4% for differentiating Normal cognition from Dementia, Normal or Mild Cognitive Impairment (MCI) from Dementia, and Normal from MCI, respectively. DISCUSSION: The proposed approach offers a fully automated identification of MCI and dementia based on a recorded neuropsychological test, providing an opportunity to develop a remote screening tool that could be adapted easily to any language.

Associations Between the Digital Clock Drawing Test and Brain Volume: Large Community-Based Prospective Cohort (Framingham Heart Study).

BACKGROUND: The digital Clock Drawing Test (dCDT) has been recently used as a more objective tool to assess cognition. However, the association between digitally obtained clock drawing features and structural neuroimaging measures has not been assessed in large population-based studies. OBJECTIVE: We aimed to investigate the association between dCDT features and brain volume. METHODS: This study included participants from the Framingham Heart Study who had both a dCDT and magnetic resonance imaging (MRI) scan, and were free of dementia or stroke. Linear regression models were used to assess the association between 18 dCDT composite scores (derived from 105 dCDT raw features) and brain MRI measures, including total cerebral brain volume (TCBV), cerebral white matter volume, cerebral gray matter volume, hippocampal volume, and white matter hyperintensity (WMH) volume. Classification models were also built from clinical risk factors, dCDT composite scores, and MRI measures to distinguish people with mild cognitive impairment (MCI) from those whose cognition was intact. RESULTS: A total of 1656 participants were included in this study (mean age 61 years, SD 13 years; 50.9% women), with 23 participants diagnosed with MCI. All dCDT composite scores were associated with TCBV after adjusting for multiple testing (P value <.05/18). Eleven dCDT composite scores were associated with cerebral white matter volume, but only 1 dCDT composite score was associated with cerebral gray matter volume. None of the dCDT composite scores was associated with hippocampal volume or WMH volume. The classification model for differentiating MCI and normal cognition participants, which incorporated age, sex, education, MRI measures, and dCDT composite scores, showed an area under the curve of 0.897. CONCLUSIONS: dCDT composite scores were significantly associated with multiple brain MRI measures in a large community-based cohort. The dCDT has the potential to be used as a cognitive assessment tool in the clinical diagnosis of MCI.

Digital sleep measures and white matter health in the Framingham Heart Study.

AIM: Impaired sleep quality and sleep oxygenation are common sleep pathologies. This study assessed the impact of these abnormalities on white matter (WM) integrity in an epidemiological cohort. METHODS: The target population was the Framingham Heart Study Generation-2/Omni-1 Cohorts. Magnetic resonance imaging (diffusion tensor imaging) was used to assess WM integrity. Wearable digital devices were used to assess sleep quality: the (M1-SleepImage™ system) and the Nonin WristOx for nocturnal oxygenation. The M1 device collects trunk actigraphy and the electrocardiogram (ECG); sleep stability indices were computed using cardiopulmonary coupling using the ECG. Two nights of recording were averaged. RESULTS: Stable sleep was positively associated with WM health. Actigraphic periods of wake during the sleep period were associated with increased mean diffusivity. One marker of sleep fragmentation which covaries with respiratory chemoreflex activation was associated with reduced fractional anisotropy and increased mean diffusivity. Both oxygen desaturation index and oxygen saturation time under 90% were associated with pathological directions of diffusion tensor imaging signals. Gender differences were noted across most variables, with female sex showing the larger and significant impact. CONCLUSIONS: Sleep quality assessed by a novel digital analysis and sleep hypoxia was associated with WM injury, especially in women.

Detection of dementia on voice recordings using deep learning: a Framingham Heart Study.

BACKGROUND: Identification of reliable, affordable, and easy-to-use strategies for detection of dementia is sorely needed. Digital technologies, such as individual voice recordings, offer an attractive modality to assess cognition but methods that could automatically analyze such data are not readily available. METHODS AND FINDINGS: We used 1264 voice recordings of neuropsychological examinations administered to participants from the Framingham Heart Study (FHS), a community-based longitudinal observational study. The recordings were 73 min in duration, on average, and contained at least two speakers (participant and examiner). Of the total voice recordings, 483 were of participants with normal cognition (NC), 451 recordings were of participants with mild cognitive impairment (MCI), and 330 were of participants with dementia (DE). We developed two deep learning models (a two-level long short-term memory (LSTM) network and a convolutional neural network (CNN)), which used the audio recordings to classify if the recording included a participant with only NC or only DE and to differentiate between recordings corresponding to those that had DE from those who did not have DE (i.e., NDE (NC+MCI)). Based on 5-fold cross-validation, the LSTM model achieved a mean (±std) area under the receiver operating characteristic curve (AUC) of 0.740 ± 0.017, mean balanced accuracy of 0.647 ± 0.027, and mean weighted F1 score of 0.596 ± 0.047 in classifying cases with DE from those with NC. The CNN model achieved a mean AUC of 0.805 ± 0.027, mean balanced accuracy of 0.743 ± 0.015, and mean weighted F1 score of 0.742 ± 0.033 in classifying cases with DE from those with NC. For the task related to the classification of participants with DE from NDE, the LSTM model achieved a mean AUC of 0.734 ± 0.014, mean balanced accuracy of 0.675 ± 0.013, and mean weighted F1 score of 0.671 ± 0.015. The CNN model achieved a mean AUC of 0.746 ± 0.021, mean balanced accuracy of 0.652 ± 0.020, and mean weighted F1 score of 0.635 ± 0.031 in classifying cases with DE from those who were NDE. CONCLUSION: This proof-of-concept study demonstrates that automated deep learning-driven processing of audio recordings of neuropsychological testing performed on individuals recruited within a community cohort setting can facilitate dementia screening.

An Artificial Intelligence-Assisted Method for Dementia Detection Using Images from the Clock Drawing Test.

BACKGROUND: Widespread dementia detection could increase clinical trial candidates and enable appropriate interventions. Since the Clock Drawing Test (CDT) can be potentially used for diagnosing dementia-related disorders, it can be leveraged to develop a computer-aided screening tool. OBJECTIVE: To evaluate if a machine learning model that uses images from the CDT can predict mild cognitive impairment or dementia. METHODS: Images of an analog clock drawn by 3,263 cognitively intact and 160 impaired subjects were collected during in-person dementia evaluations by the Framingham Heart Study. We processed the CDT images, participant's age, and education level using a deep learning algorithm to predict dementia status. RESULTS: When only the CDT images were used, the deep learning model predicted dementia status with an area under the receiver operating characteristic curve (AUC) of 81.3% ± 4.3%. A composite logistic regression model using age, level of education, and the predictions from the CDT-only model, yielded an average AUC and average F1 score of 91.9% ±1.1% and 94.6% ±0.4%, respectively. CONCLUSION: Our modeling framework establishes a proof-of-principle that deep learning can be applied on images derived from the CDT to predict dementia status. When fully validated, this approach can offer a cost-effective and easily deployable mechanism for detecting cognitive impairment.

Association Between the Digital Clock Drawing Test and Neuropsychological Test Performance: Large Community-Based Prospective Cohort (Framingham Heart Study).

BACKGROUND: The Clock Drawing Test (CDT) has been widely used in clinic for cognitive assessment. Recently, a digital Clock Drawing Text (dCDT) that is able to capture the entire sequence of clock drawing behaviors was introduced. While a variety of domain-specific features can be derived from the dCDT, it has not yet been evaluated in a large community-based population whether the features derived from the dCDT correlate with cognitive function. OBJECTIVE: We aimed to investigate the association between dCDT features and cognitive performance across multiple domains. METHODS: Participants from the Framingham Heart Study, a large community-based cohort with longitudinal cognitive surveillance, who did not have dementia were included. Participants were administered both the dCDT and a standard protocol of neuropsychological tests that measured a wide range of cognitive functions. A total of 105 features were derived from the dCDT, and their associations with 18 neuropsychological tests were assessed with linear regression models adjusted for age and sex. Associations between a composite score from dCDT features were also assessed for associations with each neuropsychological test and cognitive status (clinically diagnosed mild cognitive impairment compared to normal cognition). RESULTS: The study included 2062 participants (age: mean 62, SD 13 years, 51.6% women), among whom 36 were diagnosed with mild cognitive impairment. Each neuropsychological test was associated with an average of 50 dCDT features. The composite scores derived from dCDT features were significantly associated with both neuropsychological tests and mild cognitive impairment. CONCLUSIONS: The dCDT can potentially be used as a tool for cognitive assessment in large community-based populations.

Development and validation of an interpretable deep learning framework for Alzheimer's disease classification.

Alzheimer's disease is the primary cause of dementia worldwide, with an increasing morbidity burden that may outstrip diagnosis and management capacity as the population ages. Current methods integrate patient history, neuropsychological testing and MRI to identify likely cases, yet effective practices remain variably applied and lacking in sensitivity and specificity. Here we report an interpretable deep learning strategy that delineates unique Alzheimer's disease signatures from multimodal inputs of MRI, age, gender, and Mini-Mental State Examination score. Our framework linked a fully convolutional network, which constructs high resolution maps of disease probability from local brain structure to a multilayer perceptron and generates precise, intuitive visualization of individual Alzheimer's disease risk en route to accurate diagnosis. The model was trained using clinically diagnosed Alzheimer's disease and cognitively normal subjects from the Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset (n = 417) and validated on three independent cohorts: the Australian Imaging, Biomarker and Lifestyle Flagship Study of Ageing (AIBL) (n = 382), the Framingham Heart Study (n = 102), and the National Alzheimer's Coordinating Center (NACC) (n = 582). Performance of the model that used the multimodal inputs was consistent across datasets, with mean area under curve values of 0.996, 0.974, 0.876 and 0.954 for the ADNI study, AIBL, Framingham Heart Study and NACC datasets, respectively. Moreover, our approach exceeded the diagnostic performance of a multi-institutional team of practicing neurologists (n = 11), and high-risk cerebral regions predicted by the model closely tracked post-mortem histopathological findings. This framework provides a clinically adaptable strategy for using routinely available imaging techniques such as MRI to generate nuanced neuroimaging signatures for Alzheimer's disease diagnosis, as well as a generalizable approach for linking deep learning to pathophysiological processes in human disease.

Identification of digital voice biomarkers for cognitive health.

AIM: Human voice contains rich information. Few longitudinal studies have been conducted to investigate the potential of voice to monitor cognitive health. The objective of this study is to identify voice biomarkers that are predictive of future dementia. METHODS: Participants were recruited from the Framingham Heart Study. The vocal responses to neuropsychological tests were recorded, which were then diarized to identify participant voice segments. Acoustic features were extracted with the OpenSMILE toolkit (v2.1). The association of each acoustic feature with incident dementia was assessed by Cox proportional hazards models. RESULTS: Our study included 6, 528 voice recordings from 4, 849 participants (mean age 63 ± 15 years old, 54.6% women). The majority of participants (71.2%) had one voice recording, 23.9% had two voice recordings, and the remaining participants (4.9%) had three or more voice recordings. Although all asymptomatic at the time of examination, participants who developed dementia tended to have shorter segments than those who were dementia free (P < 0.001). Additionally, 14 acoustic features were significantly associated with dementia after adjusting for multiple testing (P < 0.05/48 = 1 × 10-3). The most significant acoustic feature was jitterDDP_sma_de (P = 7.9 × 10-7), which represents the differential frame-to-frame Jitter. A voice based linear classifier was also built that was capable of predicting incident dementia with area under curve of 0.812. CONCLUSIONS: Multiple acoustic and linguistic features are identified that are associated with incident dementia among asymptomatic participants, which could be used to build better prediction models for passive cognitive health monitoring.