Leader: Giuseppe Riva (UNICATT); Other collaborator(s): UNIFI, CNR-STIMA, Tech4care
The metaverse can be defined as a simulated digital environment that uses augmented reality (AR), virtual reality (VR), and blockchain, along with concepts from social media, to create spaces for rich user interaction mimicking the real world. The metaverse can be seen as a paradigm shift in Smart Living Environments (SLE) because it introduces a new dimension to how we think, create and connect that will have a significant impact on the elderly, too. Specifically, the metaverse can help older adults reduce their social isolation, have more fun and even get healthier. This task will work on the identification of digital biomarkers (objective, quantifiable physiological and behavioral data that are collected and measured by means of digital devices) related to the interaction of the user in the Metaverse that can be used to prevent and monitor the cognitive decline of the Elderly.
Brief description of the activities and of the intermediate results:
Theoretical framework and ethical approval
During the concerned period (November 2023-March 2024), we have made progress in identifying and extracting digital biomarkers to early identify and monitor Mild Cognitive Impairment (MCI) and cognitive decline in the elderly population. In detail, we have identified three crucial digital biomarkers associated with user interactions in a virtual reality environment (VRE), including Head, Hand and Body movements. These biomarkers were chosen to assess diverse cognitive domains such as attention, memory, and executive functions, which are often compromised in MCI and cognitive decline.
We prepared, drafted, and submitted a research protocol to the Pisa Ethics Committee. We obtained the approval in March 2024
Technical development
During the concerned period (November 2023-March 2024) we have drafted the preliminary design of the smart virtual waiting room, defining the functional and no-functional requirements, the story-telling, the specific cognitive tasks that are related to the main compromised cognitive domains in MCI and cognitive decline in the elderly population. For the development of the smart waiting room, we are collaborating with UNIFI and CNR-STIMA.
Moreover, we partnered closely with another member of the project, Tech4care. Tech4chare provided several virtual reality technological solutions that were modified following our recommendation to be able to collect digital biomarkers from the head and hand movements.
Data collection
After the ethical committing approval and the collaboration with Tech4care which provided finalized virtual reality environment able to collect digital biomarker we were able to start a data collection with elderly patients. Specifically, so far data from 10 elderly subjects (over 65) have been collected. The study aims at collecting 60 elderly subjects (20 healthy, 20 with frailty and 20 with MCI) and aims to develop an artificial intelligence model through machine learning techniques to predict MCI and its behavioural patterns through digital biomarkers of the head and hands.
Moreover, we drafted an abstract that we will present at the Age-it general meeting (Venice - May 2024) and a long abstract on the prediction of cognitive health status in older population through digital biomarkers that has been submitted for review at the CyberPsychology, CyberTherapy & Social Networking conference (CYPSY27 Conference) that will take place in September 2024
Main policy, industrial and scientific implications:
The prediction of MCI (Mild Cognitive Impairment) through digital biomarkers has significant implications for policy, industry and clinical practice. Here some of the main implications from the research and points that we have considered as relevants:
Please see the next reporting period.
VR Software Development (completed): In partnership with Tech4Care, we have created four virtual reality (VR) environments. These environments feature tasks designed to engage specific cognitive functions, particularly memory and executive functions. The VR system captures head and hand movements during these tasks, allowing us to extract digital biomarkers for analysis.
Data Collection (ongoing): We are currently collecting data from elderly participants. Our target is 60 subjects, equally divided into three groups: healthy, frail, and those with Mild Cognitive Impairment (MCI). To date, we have gathered data from 40 subjects (19 healthy, 16 frail, and 5 with MCI). We need 20 more participants to reach our goal and complete this phase.
Development of AI Models to Predict MCI (ongoing): Our initial analysis uses a dataset of 20 subjects (10 healthy, 10 frail). This dataset includes:
We've tested various machine learning models to differentiate between healthy and frail subjects. Despite the limited sample size, our results are promising. The Logistic Regression model, in particular, has shown strong performance across several metrics (precision, accuracy, recall, AUC-ROC). This suggests that our digital biomarkers can effectively distinguish between healthy and frail individuals. We expect the performance of our machine learning models, including the Logistic Regression model, to improve further once we complete our full data collection. The larger dataset will allow us to refine and validate our models more robustly.
Main policy, industrial and scientific implications:
Data Collection (ongoing): The goal of Task2.3 is to recruit 60 participants, equally distributed across three categories: healthy, frail, and individuals with Mild Cognitive Impairment (MCI). So far, we have collected data from 44 participants, consisting of 20 healthy, 19 frail, and 5 with MCI. To achieve our target and finalize this phase, we still require 16 additional participants.
Development of AI Models to Predict MCI (ongoing): Upon the completion of data collection, the same Machine Learning (ML) models employed in the preliminary analysis will be trained and tested. Given the promising results observed in the earlier phase, we anticipate improvements in the precision and accuracy of the ML models in predicting outcomes for healthy and frail elderly.
Dissemination (ongoing): We presented the preliminary results of our data collection during an episode of "Tg1 Medicina," the health segment of Italy's Tg1 news program. The broadcast will be aired in early 2025 and will be available for viewing on RaiPlay. "Tg1 Medicina" is a weekly segment that provides in-depth coverage of public and individual health topics, offering therapeutic advice, technological innovations, and expert commentary. The program airs every Sunday at approximately 8:20 AM and is produced by the Tg1 Society editorial team.
The main policy, industrial and scientific implications of the conducted activities coincided with the ones reported in the previous period.
Scientific publications