Prevention, early diagnosis and management of cognitive disorders are among the greatest global health and social care challenges of our time (1) representing a crucial priority. Cognitive assessment with harmonized neuropsychological testing is the gateway in the diagnostic roadmap for patients suspected of cognitive decline, needed for a correct definition of the syndromic presentation and to correctly address the pathway to biological diagnosis (2–4). Nonetheless, the use of harmonized cognitive testing in clinics is puzzled due to lack of time and resources, and of specific expertise both in primary care and in specialised settings (5, 6).

Digital solutions may offer the opportunity to overcome current limitations in clinical neuropsychology by measuring a broad range of cognitive abilities remotely and with relevant time-saving and reduced resources. Although the pandemic has amplified the offer of telemedicine services, harmonized and clinically validated teleneuropsychology batteries on digital solutions are still lacking. Crucially, there are too few software tools certified for this purpose in Europe [see for example Lesoil et al. (7)]. One of the main obstacles is indeed obtaining the software as medical device (SaMD) certification according to the European regulation 2017/745 (MDR). The procedure requires to prove that the risk-benefit profile of the device is non-inferior to the currently available state of the art. This may be done by performing a comparison with the standard of care. However, this approach is often expensive and time-consuming. Performance simulations are thus increasingly used in digital health to prove SaMD reliability, reproducibility, and safety prior to clinical validation, ensuring that simulated performance does not materially diverge from real-world performance, accelerating digital innovation and providing better resilience to privacy attack and a statistical validity comparable to the traditional approach (8).

In this study, we describe the methodological approach adopted for the analytical/technical and pre-clinical validation of a novel teleneuropsychology platform [i.e., Tenèpsia® (9)], already tested for usability and user-friendless (10), to prove its technical performance and suitability for EU certification under the 2017/745 MDR.

In this study, we describe a simulation approach for the validation of a teleneuropsychology software as SaMD developed for remote cognitive testing in memory clinics [i.e., Tenepsia® (9)]. The use of such approaches is growing in interest in the digital health field (29). As proved by a recent consensus study involving regulators, industry and academic experts (29), this well-recognized step within the MDR 2017/745 pathway is a valid method to provide early-stage evidence of SaMD technical performance and support pre-clinical assessment before the authorization of clinical investigations and real-patient testing. In teleneuropsychology, it may allow the verification of scoring logic, consistency of software performance metrics with real-world use, and functional robustness under controlled conditions, without exposing patients to risks of an uncertified MD (28). Therefore, such preliminary MDR approaches enrich, rather than replace, clinical validation by demonstrating that the system can reproduce and discriminate heterogeneous cognitive profiles.

Our results indicate that the scores simulated in the platform are largely consistent with profiles of potential patients in real-world scenarios. The use of simulations with dummy users provided indeed good accuracy, and coherent and credible simulated cognitive profiles supporting the validation of the platform. Experts confirmed the adequacy of the work done in terms of simulation, through the accurate identification of profiles and their positive evaluation. The digital tests integrated into the platform effectively provide comparable information with their paper-and-pencil counterparts, ensuring equivalent results. Simulated avatars have been equally validated by internal experts from the study group and external clinical experts not involved in the project, demonstrating good consistency and credibility in both groups.

This approach have many advantages in teleneuropsychology for both final users, i.e., patients and clinicians. First, it avoids ethical concerns related to real-life testing of the digital solution on cognitive patients, such as the potential psychological burden, discomfort, or stress that the neuropsychological assessment may cause, especially in vulnerable populations like individuals with dementia or cognitive decline (8). Additionally, it avoids exposing human participants to unproven technologies, thus minimizing the risks associated with early-stage testing. The simulated approach also offers a time-saving and controlled environment for technical testing, free from the unpredictable factors that can arise in clinical settings. This is especially relevant in dementia, where cognitive deficits can impact reliability of collected data (30). One of the main challenges in testing novel digital solutions is the variability in patient responses and clinical settings. Dummy patients ensure a reproducible and standardized setting environment, which enhances the reliability of the data collected. This consistency is crucial for regulatory bodies when assessing the validity and safety of a new device. In addition, synthetic data are generated reducing expenses related to data acquisition, storage, and cleaning and enable the creation of reliable datasets that preserve statistical characteristics without compromising sensitive information. Finally, the controlled environment of simulated approaches also provides a consistent baseline for testing different MDs, allowing to pinpoint and rectify problems more effectively before moving on to real-life trials. This accelerated process can lead to quicker innovations and faster market entry for new digital devices, ultimately benefiting both patients and clinicians. Moreover, the availability of certified digital platforms may also have a significant impact in the neurorehabilitation setting, enabling continuous cognitive monitoring and the personalization of remote cognitive training programs, which are particularly valuable in long-term care pathways.

Some limitations of the current study should be acknowledged. According to the feedback provided by the experts, the lack of qualitative information on medical history and of observation of patient's attitude during the neuropsychological examination may affect the credibility of simulated performances. Previous training on questionnaire evaluation is also to recommend to improve the reliability of experts’ evaluations. Future validation studies are also needed to evaluate the consistency of our anecdotal findings, address diagnostic comparability across different administration settings (hospital- vs. home-based), and test usability in the target population.

In conclusion, simulation approaches represent a promising avenue for ensuring acceleration of digital innovation in clinical neuropsychology saving costs of direct comparison with real-life standard of care. While direct comparison with the standard of care remains a critical component of MD validation, simulated approaches with dummy users provide a complementary pathway addressing ethical, financial, and logistical challenges. This approach may also serve as valuable tool for training and education. Neuropsychologists can use simulations to become familiar with new devices and procedures, ensuring a smoother transition from development to clinical practice. This dual benefit enhances both the validation process and the practical implementation of new technologies in teleneuropsychology. Digital platforms for diagnosing and monitoring cognitive disorders will certainly contribute to a transition towards a more sustainable and equitable future, especially for the most fragile segments of the population.