Alzheimer's Disease, Alzheimer's Disease Dementia, Mild Cognitive Impairment
Conditions
Keywords
Wearable Devices, Mild Cognitive Impairment, Alzheimer's disease, Alzheimer's Disease Dementia, Machine Learning
Brief summary
Cognitive decline affects millions of older adults worldwide and has a profound impact on individuals, families, and healthcare systems. Mild Cognitive Impairment (MCI) is often an early stage of Alzheimer's disease (AD), a condition for which there is currently no cure. Identifying individuals at risk at the earliest possible stage remains a major challenge. Traditional diagnostic approaches, such as laboratory biomarkers, neuroimaging, and neuropsychological testing, are usually performed at a single point in time and may fail to detect subtle or early changes in brain function and daily behavior. Recent advances in wearable technology, such as smartwatches and smart rings, allow continuous and noninvasive monitoring of physiological and behavioral patterns in daily life. These devices can capture data related to physical activity, sleep, heart rate, and other parameters that may change before clear cognitive symptoms become evident. When combined with clinical, laboratory, neuropsychological, neuroimaging, and electroencephalographic (EEG) information, these data may help identify early signs of cognitive decline. The objective of this study is to develop and validate models capable of detecting early indicators of MCI and early-stage Alzheimer's disease by integrating multiple sources of data, including clinical assessments, blood tests, neuropsychological evaluations, brain imaging, EEG recordings, and continuous data obtained from wearable devices. This is an observational, analytical, single-center, prospective cohort study that will include 150 participants of both sexes, aged 65 years or older. Participants will be recruited from the Dementia Outpatient Clinic of Getúlio Vargas University Hospital (HUGV), through referrals from external neurologists, or via study dissemination on social media. To achieve the target sample size, up to 250 individuals may be approached using a non-probabilistic, convenience-based recruitment strategy. After providing informed consent, participants will undergo a comprehensive medical evaluation, standardized and validated neuropsychological testing, laboratory and imaging examinations, and EEG recording. Participants will also receive training to use wearable devices for continuous monitoring in their daily routines. A control group of older adults without cognitive impairment will be included for comparison. All collected data will be securely stored in a centralized database and used to develop and validate analytical models aimed at identifying patterns associated with cognitive decline. The results of this study may support earlier identification of individuals at risk for MCI and Alzheimer's disease, help guide timely interventions, and potentially delay disease progression and early institutionalization, contributing to improved quality of life for older adults and their families.
Detailed description
Mild Cognitive Impairment (MCI) and early-stage Alzheimer's disease (AD) represent critical stages along the continuum of neurodegenerative cognitive disorders. Understanding these conditions is essential for enabling early diagnosis, implementing targeted therapeutic strategies, and developing comprehensive care plans aimed at improving patient and caregiver quality of life and, when possible, delaying neurodegenerative progression. The global burden of dementia is rapidly increasing and poses an urgent public health challenge. Projections estimate that approximately 115 million people worldwide will be living with dementia by 2050. In the United States alone, the prevalence of MCI is expected to exceed 21 million individuals, with nearly 14 million cases of Alzheimer's disease by 2060. These figures highlight the pressing need for advances in translational neuroscience, particularly in early diagnostic strategies and preventive, personalized approaches to care. In this context, wearable technologies have emerged as promising tools for continuous, noninvasive monitoring of physiological and behavioral parameters in real-world settings. Devices such as smartwatches and smart rings enable longitudinal collection of data related to heart rate variability, sleep architecture, physical activity, and circadian rhythms-factors increasingly associated with cognitive decline. Emerging evidence suggests that changes in these parameters may precede or accompany early cognitive impairment. When integrated with advanced artificial intelligence (AI) methods, these data may reveal subtle patterns indicative of early neurodegenerative processes that precede overt clinical symptoms. Combining wearable-derived data with neuroimaging and electroencephalography (EEG) has the potential to generate more robust diagnostic models. While wearables capture behavioral and physiological dynamics, magnetic resonance imaging (MRI) provides detailed information on brain structure and function, and EEG enables analysis of neural oscillations and connectivity patterns linked to early cognitive impairment. The integration of these multimodal data streams represents a complex methodological challenge, requiring advanced computational frameworks capable of handling heterogeneous, high-dimensional datasets. Advances in artificial intelligence and machine learning enable the integration of multimodal data and the identification of complex patterns not detectable through traditional analytical approaches. Multimodal data fusion strategies that combine wearable-derived physiological and behavioral features with neuropsychological, neuroimaging, and EEG-derived variables may enhance diagnostic performance and support individualized risk stratification. This is an observational, analytical, single-center, prospective cohort study designed to integrate multimodal clinical and digital data for the development and validation of AI-based models aimed at early detection of MCI and early-stage Alzheimer's disease. The study is conducted at Getúlio Vargas University Hospital (HUGV), a tertiary academic center and regional reference for high-complexity care, in collaboration with the Center for Research and Development in Electronic and Information Technology (CETELI/UFAM), which provides expertise in intelligent systems, data infrastructure, and AI model development. Machine learning and deep learning approaches are applied to identify patterns associated with cognitive decline, support early detection of MCI and early AD, and enable risk stratification. Model development and validation prioritize robustness, interpretability, and potential clinical applicability. This study aims to support the development of scalable, accessible, and noninvasive AI-based tools for early detection of cognitive impairment. By leveraging continuous wearable monitoring and multimodal data integration, the proposed approach may contribute to earlier diagnosis, improved risk stratification, and more timely intervention strategies for individuals at risk of Alzheimer's disease.
Interventions
None listed
Sponsors
Getúlio Vargas University Hospital
Universidade Federal do Amazonas
Samsung Eletrônica da Amazônia Ltda
Study design
Observational model
COHORT
Time perspective
PROSPECTIVE
Eligibility
Sex/Gender
ALL
Age
65 Years to No maximum
Healthy volunteers
Yes
Inclusion criteria
Age 65 years or older; Clinical suspicion of MCI, early AD Patients who are conscious, oriented, and able to respond to questionnaires; Ability and willingness to use wearable devices for 30 days; Signature of the Informed Consent Form (ICF) Non-inclusion criteria Advanced dementia Severe uncontrolled psychiatric disorder Severe visual or hearing impairment that prevents communication Severe physical or cognitive inability to use wearables Pacemaker incompatible with the devices
Exclusion criteria
Unconfirmed diagnosis of MCI or early AD; Voluntary withdrawal at any stage of the study; Death during the research period; Inappropriate use or non-adherence to the use of wearable devices Refusal or inability to fully perform the requested tests and exams.
Design outcomes
Primary
| Measure | Time frame | Description |
|---|---|---|
| Accuracy of a Predictive Model for Identification of Mild Cognitive Impairment and Early Alzheimer's Disease | Up to 30 days of continuous wearable monitoring | Evaluation of the accuracy, defined as the proportion of correctly classified participant, of a predictive model developed to identify Mild Cognitive Impairment and early-stage Alzheimer's Disease based on data collected from wearable technologies and digital questionnaires, using established clinical and standardized neuropsychological diagnoses as the reference standard. |
Secondary
| Measure | Time frame | Description |
|---|---|---|
| Association Between Speech and Voice Features From Smartphone Audio and Clinical Group | Up to 30 days | Evaluation of the relationship between speech and voice features extracted from smartphone-based audio recordings including speech rate articulation pause patterns pitch variability and voice quality measures and clinical group classification control, mild cognitive impairment, and early Alzheimer's disease, including correlations with cognitive and functional scores. |
| Association Between Wearable-Derived Sleep Metrics and Clinical Group | Up to 30 days | Evaluation of the association between wearable-derived sleep parameters including total sleep time, sleep efficiency, sleep latency, wake after sleep onset, and sleep stage distribution and clinical group classification cognitively unimpaired control, mild cognitive impairment, and early Alzheimer's disease, as well as correlations between sleep metrics and standardized cognitive and functional assessment scores. |
| Association Between Wearable-Derived Physical Activity Metrics and Clinical Group | Up to 30 days | Assessment of the relationship between wearable-derived physical activity metrics including daily step count, activity intensity levels, sedentary time, and energy expenditure and clinical group classification control, mild cognitive impairment, and early Alzheimer's disease, including correlations with cognitive and functional performance measures. |
| Association Between Wearable-Derived Heart Rate and Heart Rate Variability Metrics and Clinical Group | Up to 30 days | Analysis of the association between wearable-derived heart rate and heart rate variability parameters including resting heart rate and time-domain and frequency-domain HRV indices and clinical group classification control, mild cognitive impairment, and early Alzheimer's disease, as well as correlations with cognitive and functional scores |
| Difference in Timed Up and Go Performance Across Clinical Groups | Baseline assessment Day 1 (up to 30 days) | Comparison of Timed Up and Go test performance across clinical groups control, mild cognitive impairment, and early Alzheimer's disease and evaluation of correlations between TUG performance and wearable-derived physical activity and sleep metrics. |
| Association Between Facial Expression Features From Smartphone Video and Clinical Group | Up to 30 days | Assessment of the association between facial expression features extracted from smartphone-based video recordings including facial action units emotional expressivity and movement dynamics and clinical group classification control, mild cognitive impairment, and early Alzheimer's disease. |
| Association Between Smartphone-Derived Eye Movement Features and Clinical Group | Up to 30 days | Assessment of the association between eye movement and ocular tracking features derived from smartphone-based assessments including saccade metrics fixation stability and pursuit performance and clinical group classification control, mild cognitive impairment, and early Alzheimer's disease. |
| Association Between Electroencephalography (EEG) Classification and Clinical Group | Up to 30 days | Evaluation of the association between electroencephalography classification normal versus abnormal and type of abnormality and clinical group classification control, mild cognitive impairment, and early Alzheimer's disease, as well as correlations between EEG findings and wearable-derived sleep and physical activity metrics. |
| Association Between Non-Invasive Intracranial Compliance Metrics and Clinical Group | Up to 30 days | Assessment of the relationship between non-invasive intracranial compliance metrics obtained using the Brain4Care device including P2 P1 ratio and waveform morphology and clinical group classification control, mild cognitive impairment, and early Alzheimer's disease, including correlations with wearable-derived sleep and physical activity metrics. |
| Association Between Carotid and Vertebral Doppler Findings and Clinical Group | Up to 30 days | Evaluation of the association between carotid and vertebral Doppler ultrasound findings including intima-media thickness and presence or degree of arterial stenosis and clinical group classification control, mild cognitive impairment, and early Alzheimer's disease. |
| Feasibility and Adherence to Continuous Wearable Monitoring | Up to 30 days | Assessment of feasibility and participant adherence to continuous wearable monitoring, including the percentage of days with valid wearable data, average daily wear time, and study dropout rate. |
| Association Between Cognitive and Neuropsychological Performance and Clinical Group | Up to 30 days | Evaluation of the relationship between global and domain-specific cognitive and neuropsychological performance, including screening measures and standardized neuropsychological tests assessing memory, executive function, attention, language, processing speed, and visuospatial abilities, and clinical group classification control, mild cognitive impairment, and early Alzheimer's disease. |
| Association Between Functional, Neuropsychiatric, and Behavioral Measures and Clinical Group | Up to 30 days | Assessment of the association between functional capacity, neuropsychiatric symptoms, and behavioral measures, including instrumental activities of daily living, anxiety, and depressive symptoms, and clinical group classification control, mild cognitive impairment, and early Alzheimer's disease, including correlations with cognitive and digital biomarkers. |
| Association Between Blood-Based Biomarkers and Clinical Group | Up to 30 days | Evaluation of the relationship between blood-based biomarkers, including routine laboratory parameters and plasma biomarkers of Alzheimer's disease pathology, and clinical group classification control, mild cognitive impairment, and early Alzheimer's disease, including correlations with cognitive, functional, and neuroimaging measures. |
| Association Between Neurophysiological Measures and Clinical Group | Up to 30 days | Assessment of the association between neurophysiological features, including electroencephalographic patterns and noninvasive intracranial compliance waveform metrics, and clinical group classification control, mild cognitive impairment, and early Alzheimer's disease, including correlations with cognitive and neuroimaging outcomes. |
| Association Between Neuroimaging Features and Clinical Group | Up to 30 days | Evaluation of the relationship between structural neuroimaging and cerebrovascular features, including cortical and medial temporal atrophy, white matter disease, microbleeds, infarcts, volumetric and cortical thickness measures, and clinical group classification control, mild cognitive impairment, and early Alzheimer's disease. |
Countries
Brazil
Contacts
CONTACTRobson Luís Oliveira de Amorim, PhD
CONTACTEliana Brasil Alves, Master of Health Sciences
PRINCIPAL_INVESTIGATORRobson Luís Oliveira de Amorim
Getúlio Vargas University Hospital
STUDY_DIRECTOREliana Brasil Alves
Getúlio Vargas University Hospital
STUDY_DIRECTORMarly Guimarães Fernandes Costa
Federal University of Amazonas
STUDY_DIRECTORCícero Ferreira Fernandes Costa Filho
Federal University of Amazonas
STUDY_CHAIRCaio Eduardo Rodrigues Falcão
Getúlio Vargas University Hospital
Outcome results
None listed