Traditional paper-and-pencil cognitive tests, such as the Mini-Mental State Examination and the Montreal Cognitive Assessment, are widely used in clinical practice. However, these conventional methods exhibit numerous limitations, including the necessity for professional personnel to conduct face-to-face interviews, the potential for human bias in test results, and practice effects resulting from repeated administrations [15,16]. To overcome these limitations and support more scalable approaches to assessment and screening, digital cognitive assessments have emerged, leveraging emerging technologies to achieve automated, standardized, convenient, multidimensional evaluations of cognitive functions. Over the past 2 decades, digital cognitive assessment has experienced rapid development, with extensive research and adoption in clinical settings [15,16]. These digital assessments rely on objective, quantifiable, physiological, and behavioral data collected by digital devices, enabling continuous ecological assessment and long-term monitoring of cognitive health. Many CDTx solutions integrate assessment features, such as digital assessment and screening tests, to track cognitive status over time. Based on functionality, technological characteristics, and application scenarios, digital cognitive assessment can be broadly categorized into three primary types: (1) computerized cognitive assessment, (2) digital biomarker-driven cognitive assessment, and (3) multimodal brain imaging cognitive assessment.

Computerized cognitive assessments include both scale-based assessments and task-based assessments, developed based on traditional pencil-and-paper tests, which have been viewed as the gold standard for diagnosis. For example, digitalized versions of the Montreal Cognitive Assessment, Mini-Mental State Examination, and Alzheimer’s Disease Assessment Scale-Cognitive Subscale have been validated and show good consistency with traditional paper-and-pencil methods [17], offering advantages such as automated scoring and real-time data analysis. Meanwhile, task-based assessments like the Cambridge Neuropsychological Test Automated Battery [18], Brain-Check [19], Neurotrack cognitive Battery [20], and the Beijing Aging Brain Rejuvenation Initiative brain health system evaluate cognitive abilities such as memory, attention, and executive function through interactive tasks and simulated environments [21]. Furthermore, virtual reality (VR) environments and computerized cognitive games constitute significant components in this category. These digital technologies either simulate traditional cognitive assessments or create novel interactive tasks to evaluate a broad range of cognitive functions. For example, the Virtual Supermarket uses immersive environments to simulate shopping tasks, providing an effective method for assessing executive functions and memory capabilities [22]. In addition, computerized cognitive games, such as Episode Gamification, engage user engagement through gamified design while simultaneously evaluating various cognitive domains, including memory, attention, and executive functions [23].

Digital biomarkers are objective, quantifiable, physiological, and behavioral data collected and measured through digital devices, such as embedded environmental sensors, portable devices, and wearable. These data can be gathered in the context of daily activities with minimal intrusion, enabling ecological monitoring. Digital biomarkers are increasingly used for early detection and follow-up of cognitive impairment. For instance, older adults with cognitive impairments show significantly reduced daily activity ranges compared with cognitively normal older adults, with more restricted and stable activity patterns [24]. Therefore, monitoring their daily behavior trajectories can provide early warnings. Furthermore, by integrating machine learning and deep learning algorithms and incorporating speech and natural language processing from both healthy and patient populations, classifiers can be constructed to identify cognitive impairments. It has been found that speech fluency, complexity, and coherence can also serve as markers for early screening [25]. In addition, significant oculomotor impairments have been observed in cognitive impairment patients [26], which monitors eye movements and visual attention through tasks such as reading comprehension and visual search, providing insights into abnormal attention and visual behaviors [27].

Multimodal imaging techniques offer diverse perspectives for observing and analyzing the brain, providing critical imaging evidence for the early diagnosis of AD by assessing structural, functional, and metabolic changes. Currently, widely used multimodal imaging techniques for AD assessment include structural magnetic resonance imaging (sMRI), functional magnetic resonance imaging (fMRI), and positron emission tomography (PET) [28]. Additionally, diffusion tensor imaging (DTI) and other advanced methods provide critical insights into the disease's pathology and progression [28], and no imaging modality can serve all purposes as each has unique strengths and weaknesses [29]. In recent years, machine learning techniques have become increasingly integrated with multimodal imaging data, improving diagnostic efficiency and accuracy by classification and prediction. A widely used machine learning and deep learning approach in this context is ensemble learning, which integrates multiple machine learning models to optimize fusion strategies. For instance, Shukla et al [30] applied ensemble learning to positron emission tomography and T1-weighted MRI data from the Alzheimer Disease Neuroimaging Initiative dataset for both binary and multiclass classification of AD, cognitive impairment, and cognitively normal individuals. Their approach achieved up to 99% accuracy for binary classification and 96% for multiclass classification, demonstrating its potential in early AD diagnosis [30]. Convolutional neural networks have also shown strong performance in AD classification. Farooq et al [31] used deep convolutional neural networks with a 4-way classifier to categorize AD progression stages, achieving 98.8% accuracy on the Alzheimer Disease Neuroimaging Initiative dataset. Furthermore, neuromorphic computing, inspired by human brain mechanisms such as neural connectivity and synaptic plasticity, enhances both machine learning and deep learning models, excelling in cognitive function analysis and medical imaging tasks. In AD research, Turkson et al [32] demonstrated that spiking neural networks outperform traditional machine learning methods in predicting AD classification. In addition, a hybridization of machine learning and deep learning is also used for classifying AD, achieving an accuracy of 91.84% on test data [33].

Digital cognitive assessment methods offer distinct functional strengths across modalities, and their thoughtful integration may provide a more comprehensive understanding and assessment of cognitive performance in real-world environments and clinical settings.

There is currently no definitive cure or intervention for cognitive impairment, with existing management strategies primarily aimed at symptom control and slowing disease progression [22]. In parallel, pharmaceutical approaches face significant challenges, with AD drug development showing a failure rate as high as 99.6% [34]. Given the lack of an intervention for cognitive impairment and dementia, research has increasingly focused on delaying disease progression and developing interventions to enhance and preserve cognitive function [35]. CDTx have emerged as a promising direction in this context, particularly for older adults with cognitive impairment. These interventions can be broadly categorized into 3 types. The first involves strengthening existing cognitive pathways through behavioral exercises to enhance specific cognitive functions; the second targets the direct modulation of brain activity to improve overall brain function or the targeted brain regions; and the third aims to identify and manage risk factors associated with cognitive decline, such as lifestyle behaviors, chronic diseases, and mental health issues, with the goal of mitigating root causes and delaying onset.

Many traditional cognitive interventions have been adapted for use on current technological devices, such as smartphones, tablets, and computers, and corresponding digital interventions have been developed based on identified evaluation metrics. These approaches are not only cost-effective alternatives but also innovative solutions for cognitive enhancement [36]. For instance, CogniFit (CogniFit Ltd) implements personalized and gamified cognitive tasks that participants can perform at home. This intervention effectively enhances cognitive function while promoting engagement and adherence through psychoeducation and behavior modification techniques. It has been shown to significantly improve cognitive impairments in older adults [37]. Furthermore, video games combining cognitive gameplay with physical exercise, referred to as “Exergames,” have been developed. These games have been shown to improve physical function, reduce depression, and enhance cognition and quality of life in older adults [38]. VR also represents a promising intervention, with its immersive and interactive features making it well-suited for simulating complex environments and tasks to improve cognitive training outcomes. Although studies validating the effectiveness of VR remain relatively limited, existing research suggests its potential to improve memory function and social interaction [39]. Further research is needed to optimize the design and broaden the application of these technology-based interventions [40].

This approach primarily uses neuro-modulation technologies to target specific brain regions or neural circuits, enhancing or restoring brain function. Repetitive transcranial magnetic stimulation is a widely used technique that has been shown to improve cognitive function in patients with cognitive impairment [41], and various combinations of repetitive transcranial magnetic stimulation parameters have been found effective in enhancing different cognitive domains–based approaches suggest that cognitive impairments are linked to distributed brain networks rather than isolated regions [42]. The corticohippocampal circuit, crucial for long-term memory, can be targeted through paired associative stimulation (PAS) to enhance memory connections. Studies have shown that PAS improves delayed recall in cognitive impairment by synchronizing neuronal activity within this network [43]. In addition, PAS mimics sleep-related neural oscillations, facilitating memory consolidation and optimizing memory training outcomes [44].

The 2024 Lancet Commission reports that around 45% of dementia cases may be prevented or delayed by addressing 14 modifiable risk factors, including lifestyle, health, environmental, and social factors [45]. In line with this, another approach to improving cognitive abilities involves addressing risk factors that impair cognitive development, such as diabetes, depression, and hypertension [46]. For example, Deprexis, a digital tool based on psychotherapy, helps improve mood and increase positive emotions and behaviors [47]. The HERB Digital Hypertension 1 (HERB-DH1) alleviates hypertension by offering education and lifestyle guidance, including reducing salt intake, controlling weight, exercising, improving sleep quality, and managing stress [48]. In addition, other digital intervention tools targeting diet, smoking, and alcohol consumption can also be effectively used to address these risk factors.

Digital cognitive interventions reflect an evolving understanding that cognitive health is related not only to neural mechanisms, but also to behavioral, psychological, and environmental factors. As such, digital cognitive interventions hold significant potential to complement existing care models and broaden access to timely, personalized support.

Cognitive management and monitoring refer to the ongoing tracking of cognitive functions over time, offering a longitudinal view of an individual’s cognitive health. This approach captures fluctuations in cognitive abilities, supports early detection of decline, and enables timely intervention and cognitive care planning.

Cognitive assistance technologies leverage digital tools, including information and communication devices and decision-support systems, to help individuals with cognitive impairment and their caregivers in daily cognitive care and management. These technologies aim to enhance quality of life, improve caregiving efficiency, and promote independent living. Personal digital assistance programs, such as smartphone apps, assist with daily life management by helping patients manage time, remember appointments, and organize tasks, which have been shown to significantly improve executive function in patients with acquired brain injuries [49]. In addition, many of these therapeutics integrate health monitoring features, such as medication reminders and physiological data tracking, providing valuable feedback loops for clinicians and care teams. Wearable devices such as smartwatches and fitness trackers further expand cognitive assistance by continuously monitoring physical activity, sleep patterns, and cognitive performance, enabling real-time adjustments to interventions as needed [50]. Multimedia-based tools, including video, audio, and internet-based assistive tools also provide rich resources for caregivers, including online training, remote video calls for timely support, and artificial intelligence (AI)–powered decision support systems that generate personalized care plans based on patients’ data [51].

Real-time cognitive monitoring technologies collect objective physiological and behavioral data to continuously assess patients’ cognitive status and functional capacity in daily life. Devices such as actigraphy sensors, GPS, in-home cameras, and electroencephalography (EEG) systems enable and facilitate real-time tracking, analysis, and evaluation of cognitive status and intervention effectiveness. Mobile apps and mobile health solutions are increasingly used to provide task-based feedback and cognitive performance tracking over time, helping patients and clinicians identify patterns in cognitive function [52].

Advanced EEG methods, including quantitative EEG, are now being embedded into digital monitoring devices to detect abnormal brain activity and enable earlier identification of cognitive decline. Furthermore, machine learning algorithms and large language models applied to real-time data can forecast future cognitive decline based on behavioral and neural biomarkers, opening new possibilities for early diagnosis and proactive intervention [53].

Digital cognitive management and monitoring technologies mark a shift toward proactive, personalized, and data-driven cognitive care and management. By supporting both individuals and caregivers through real-time feedback and adaptive support systems, these tools offer meaningful opportunities to enhance quality of life and optimize care delivery across various stages of cognitive impairment.

Figure 1 presents a structured summary of the innovative applications of CDTx for cognitive impairment, categorizing their core functions into 3 domains—digital cognitive assessment, digital cognitive intervention, and digital cognitive management and monitoring. The figure highlights how these technologies are applied across the care continuum to support early detection, personalized treatment, and ongoing disease management for cognitive impairment.

CDTx must undergo rigorous clinical validation to ensure their adoption in health care. Despite growing enthusiasm, the lack of comprehensive clinical validation makes it difficult for stakeholders to assess their true value compared with conventional interventions. One major challenge lies in the heterogeneity of cognitive impairment populations, which complicates the development of standardized protocols and outcomes metrics [54]. This variability hampers the ability to draw consistent, generalizable conclusions from clinical trials and real-world implementations consistently.

Another critical obstacle is the rapidly evolving nature of DTx technologies, which often outpaces the development of clinical guidelines and regulatory frameworks. This technological dynamism can lead to discrepancies in trial designs, end point selection, and outcome measures, making comparisons across studies difficult and limiting the accumulation of robust evidence [55]. In addition, cultural and geographic variations in clinical practice and health care infrastructure further complicate the establishment of unified evidence frameworks that can be applied universally [56].

To address these challenges, innovative strategies emphasize the potential of real-world evidence and home-based interventions. For example, remotely supervised programs that integrate both cognitive and physical training reduce patient and caregiver burden while enabling broader participation and scalable clinical data collection [57]. Likewise, leveraging data from electronic health records and wearable devices allows for the generation of real-world insights that complement traditional clinical trial findings, providing a more comprehensive understanding of DTx outcomes and performance [58].

Cross-sector collaboration is essential to accelerating clinical validation efforts. Consensus building among researchers, clinicians, and patient groups can help define standard outcomes and protocols, as well as the acceptable efficacy threshold [59]. International efforts and coordination on clinical trial methodology and data-sharing practices will be critical to facilitate evidence accumulation, consistency, and comparability of findings.

CDTx face significant regulatory challenges due to the absence of dedicated pathways for their development, validation, market entry, and postmarket oversight. Existing medical device regulations lack the flexibility required to accommodate rapidly evolving digital health technologies and DTx, leading to uncertainties in approval, compliance, and clinical adoption [60,61]. The lack of harmonized regulatory standards results in inconsistent prescribing criteria across jurisdictions, which complicates clinical integration and limits the scalability of DTx solutions [62,63]. In addition, ongoing debates over whether DTx should follow a prescription-based model or be available directly to consumers further fragment the regulatory landscape, influenced by divergent commercial strategies and legal frameworks [64].

Furthermore, current regulatory structures struggle to incorporate dynamic mechanisms that can keep pace with the fast-moving technological innovations [65]. Varying evidence requirements across countries or states often lead to redundant approval processes, increasing costs and delaying the timely entry of products into the market [66]. Furthermore, many current regulatory assessments and policies do not adequately account for the broader impact of DTx on patient education, behavioral change, and system efficiency [67], reflecting underlying policy gaps in recognizing the multifaceted value of DTx. Insufficient or absent postmarket surveillance mechanisms and inadequate policy oversight further limit the ability to monitor the long-term safety, real-world effectiveness, and scalability of CDTx [68].

Several recent policy efforts suggest a shift toward greater alignment and innovation in regulating digital health. At the global level, leading regulatory bodies have also made strides in establishing pathways for the evaluation and approval of DTx. In the United States, the Food and Drug Administration adopted the Software as a Medical Device regulatory framework [9], which aligns with international guidelines developed by the International Medical Device Regulators Forum, this framework provides risk-based classification and guidance for DTx, with several products such as EndeavorRx, which is a digital therapeutic product designed by Akili Interactive, already approved under these standards [69]. In Europe, in its Regulatory Science Strategy to 2025, the European Medicines Agency proposes the creation of an integrated evaluation pathway for assessing medical devices, in vitro diagnostics, and borderline products including digital health technologies and therapeutics [70]. In China, guidelines from the National Health Commission, the National Administration of Traditional Chinese Medicine, and the Chinese Center for Disease Control and Prevention have laid the foundation for promoting AI applications and digital health technologies including DTx, providing a strategic framework and instructions for the application of AI and digital health in health care [71]. These guidelines emphasize the safe and strategic integration of emerging technologies into clinical pathways, providing a regulatory environment that supports both innovation and standardization.

Despite these advances, cross-national variations in evidence thresholds, regulatory timelines, and postmarket surveillance systems persist. This lack of harmonization and consistency often leads to duplicative regulatory burdens, increased costs, and slower global dissemination of DTx. To bridge the gap between technological innovation and outdated regulatory structures, tailored regulatory frameworks are needed to establish safety, efficacy, and usability standards while enabling adaptive to adjust and change that could accommodate ongoing technology advancements [72,73]. One promising model is the use of “regulatory sandboxes,” which provide controlled environments for testing DTx and digital health products under specific safety standards and regulatory oversight [73]. This approach enables more agile evaluation and iterative refinement, allowing regulatory bodies to monitor emerging therapeutics and technologies more closely, thereby accelerating time to market while safeguarding patient safety.

In addition, enhanced collaboration among regulators, policy makers, clinicians, and industry stakeholders will be essential to streamline prescribing guidelines and establish consistent approval criteria [65]. Global harmonization of clinical evidence requirements across jurisdictions can help reduce redundant approval processes, lower market entry barriers, and accelerate the adoption of DTx [66]. Finally, strengthening postmarket surveillance through real-time data analytics, patient registries, and adverse event tracking will enable regulators and policy makers to respond more dynamically, ensuring that DTx remain safe, effective, and sustainable over time [68].

The adoption and implementation of DTx depend significantly on user acceptance and long-term engagement, both of which are influenced by factors such as perceived usefulness, ease of use, trust, and accessibility. One persistent challenge is high dropout rates and nonadherence, often linked to difficulties in maintaining interest and motivation over time. For patients with cognitive impairments or limited digital literacy, these platforms may present a steep learning curve or misunderstanding, leading to user frustration, disengagement, and eventual abandonment of DTx. Another key issue is perceived usefulness, if users do not experience clear benefits such as improved health outcomes and enhanced quality of life compared with traditional treatment methods, they are unlikely to maintain consistent engagement [74]. Similarly, complex user interfaces, technical glitches, or insufficient support can act as further barriers to adoption [75]. These factors are especially important in older populations, where cognitive limitations may intersect with low technology familiarity.

Notably, a qualitative study involving older adults using a cognitive training DTx platform found that although some users experienced difficulties understanding the instructions, many still perceived the platform as both enjoyable and beneficial, particularly when it was recommended by health care providers. Participants expressed strong awareness of the importance of cognitive training and demonstrated greater trust and willingness to engage with digital health technologies and therapeutics that were clinician-endorsed [76].

To address these engagement challenges, user-centered design must be placed at the core of DTx development. Iterative co-design approaches emphasize the importance of involving target users throughout the development cycle and can ensure that DTx products align with their real-world needs, preference, and capabilities [77]. Participatory design frameworks, such as those applied in the iReadMore app (Neurotherapeutics Group), have been shown to enhance usability and promote user motivation and accessibility through personalization and intuitive interfaces [78].

Tailoring DTx products to the cognitive and physical capabilities of target populations is critical. For example, interfaces designed with simplified navigation, voice support, and visual aids may increase accessibility for users with mild cognitive impairment. As Vial et al [79] discuss, human-centered and empathic design processes ensure that digital mental health interventions are both accessible and aligned with user needs, fostering trust and long-term usage. In parallel, trusted intermediaries, such as physicians, therapists, or community health workers, play a crucial role in increasing adoption by legitimizing the use of DTx and encouraging long-term use [79]. Wong et al [80] emphasize that user-centered design not only improves usability but also ensures integration with clinical workflows, enhancing the credibility of digital tools when endorsed by medical professionals. Furthermore, Strauss et al [81] argue that cross-functional frameworks, including input from patients and health care providers, result in solutions that are both effective and engaging. By addressing these barriers and emphasizing user-centered, humanized design, digital therapies can achieve greater acceptance and adherence among diverse patient populations.

The rapid development of DTx brings significant ethical and privacy-related concerns that must be addressed to ensure patient trust, data safety, and equitable access to care. One of the foremost challenges is data privacy and security, as DTx platforms often require the continuous collection and processing of highly sensitive personal health data. This introduces substantial risks of unauthorized access, data breaches, and misuse, particularly when robust data protection measures are currently lacking. These risks are further complicated by regional inconsistencies in data protection legislation, such as varying implementations of General Data Protection Regulation (GDPR)–like frameworks, leading to uneven security levels across jurisdictions [82].

Another critical concern is informed consent. Many patients, particularly those with cognitive impairments, struggle to fully understand the implications of data-sharing policies, especially when consent forms and data practices are opaque or overly complex, undermining the principle of truly informed consent [83].

In addition, digital health technologies frequently require reliable internet access, compatible devices, and basic digital literacy, creating barriers for underprivileged populations and potentially exacerbating health inequities. The lack of robust postmarket surveillance systems further compounds these concerns, as ongoing real-world safety and effectiveness data are often insufficient, making it difficult to oversee and identify emerging ethical risks [65].

A comprehensive, ethically grounded strategy is essential to ensure the safe, ethical, and equitable deployment of CDTx. First, strengthening data protection frameworks is crucial, this includes mandating the use of encryption, anonymization, and secure data storage protocols, as well as enforcing compliance with international standards such as General Data Protection Regulation [84]. Second, simplifying consent processes and improving transparency can empower patients to make truly informed decisions [85], this may involve visual aids, plain language summaries, or interactive digital interfaces that clarify how personal data will be used. To further promote fairness, developers need to actively address algorithmic biases through inclusive design, regular ethical audits, and accountability mechanisms to ensure system decisions do not reinforce existing disparities [86,87].

Furthermore, integrating comprehensive postmarket surveillance on data privacy and safety as a core component of DTx regulation is vital, including real-time monitoring, longitudinal patient registries, and adverse event tracking and reporting systems to continuously monitor and enhance safety and ethical integrity over time [88]. Finally, ethically sound deployment of CDTx requires long-term collaboration among different stakeholders. Regulatory bodies, clinicians, industry developers, and patient advocacy groups must co-develop ethical guidelines and frameworks that reflect diverse values and lived experiences [89]. Such collaboration ensures that DTx are implemented in a way that protects individual data and rights while advancing clinical innovation.

The effective integration of CDTx into existing health care systems is significantly hindered by issues of data interoperability and system integration. Seamless data exchange between CDTx platforms and electronic health records (EHRs) is crucial for clinical utility, yet technical fragmentation and inconsistent standards remain major obstacles [90]. Many health information technology systems operate in silos, and the lack of consistent technical standards across platforms, along with fragmented digital health ecosystems, poses significant barriers to building a unified infrastructure [91].

Furthermore, current health care infrastructures often lack the capacity to accommodate digital health and DTx products, especially in resource-limited settings [91]. Although standards such as HL7 Fast Healthcare Interoperability Resources (FHIR) are gaining momentum, implementation remains inconsistent and uneven across regions and health care organizations, leading to gaps in compatibility and functionality. These problems are compounded by the increasing reliance on internet of things devices, such as wearables and home monitoring tools, which often struggle to interface effectively with cloud-based EHR systems, this lack of compatibility limits the aggregation, accessibility, and clinical utility of patient data in clinical settings [92].

Addressing these challenges requires the creation of an open, interoperable digital ecosystem. Adopting globally recognized interoperability standards will enhance the usability, integration, and scalability of CDTx [93,94]. Among these, HL7 FHIR has emerged as a foundational framework for standardized data exchange across diverse health care systems. Its modular architecture and open application programming interface design make it especially well-suited for supporting DTx in dynamic and data-intensive environments [95].

Beyond interoperability, data security must remain a parallel priority. Integrating blockchain technologies into FHIR-based systems has shown potential for enhancing the trustworthiness and traceability of shared health data while reducing risks of unauthorized access and data breaches [93].

In clinical practice, the successful integration of CDTx with EHR systems could unlock personalized decision support, reduce diagnostic errors, and enable more adaptive, data-driven care pathways [96,97]. As interoperability standards become more widely adopted and system integration capabilities continue to improve, DTx can evolve from standalone tools to fully embedded components within the cognitive care delivery system, significantly enhancing their effectiveness and sustainability.

Challenges and future prospects of CDTx involve 5 key domains, that is, clinical evidence and effectiveness, regulatory uncertainty, patient acceptance, data privacy and ethics, and system interoperability. Figure 2 outlines these challenges and potential development pathways from a multistakeholder perspective. Table 1 complements this by summarizing key issues and proposed strategies across stakeholder groups, offering a practical foundation to support the clinical adoption and broader implementation of CDTx in cognitive care.