Preliminary results from RADAR-AD presented at AAIC

Functional decline (functioning in daily life) in AD has a major impact on the lives of people living with AD. It is therefore a crucial endpoint in clinical trials. At present, measurement techniques used by clinicians to detect functional decline are questionnaires or direct observation. However, these assessments are only performed periodically and are based on the recall by the participants or their carers. Therefore, these techniques lack the sensitivity to measure subtle symptoms in the early stages of AD. As a result, they are of limited use in clinical trials or intervention studies in preclinical AD. Remote monitoring technologies (RMTs) such as wearables or smartphone applications provide continuous and objective measuring of one’s activity in their own home environment. That is why RADAR-AD researchers expect RMTs to provide more sensitive measures to detect functional decline, especially in the early stages of AD.

RADAR-AD preliminary results

Gait characteristics in preclinical AD

Casper de Boer (Amsterdam UMC) and Jelena Curcic (Novartis) presented preliminary project results from the use of the Physilog (Gait Up) device in the RADAR-AD study. Physilog is a device with sensors which collects gait (a person’s way of walking) parameters by measuring walking pattern and speed. For people living with AD gait is an important functional domain which supports their independent living. Currently, there is little insight into gait impairments in the pre-symptomatic stages of AD. Therefore, one of the objectives of the RADAR-AD study is to compare outcomes of the gait assessments done in clinical settings between healthy “control” participants (with no cognitive symptoms and no AD pathology), participants who do not have any cognitive symptoms but have confirmed pathological hallmarks of AD (preclinical AD), and AD participants with cognitive impairment.

So far, 22 “control” participants, 6 preclinical AD participants and 10 symptomatic AD patients were included. The participants performed a walking task, a dual task – first, walking while counting backwards from 100, and second, a timed up-and-go (TUG) task. While performing these tasks the participants were wearing Physilog sensors on the left and right foot, as well as on the right of their hip. The metrics provided by the sensors included gait speed, steps per minute, time spent on both feet while walking (a measure of gait stability), and TUG task completion time.

As the RADAR-AD study includes more participants and expands this data set, the preliminary findings will be investigated further. The focus of future RADAR-AD research will be placed on the possible pathological (disease related) causality of altered gait patterns. Subsequent analyses will include larger sample sizes and more in-depth analyses of gait behaviour. In addition, these findings will be related to subjective measures of mobility, as reported by the patients themselves.

Augmented reality to detect cognitive decline in early AD

At AAIC 2021 Marijn Muurling, PhD student at the Alzheimer Center Amsterdam (Amsterdam UMC) presented preliminary results from the Altoida app used in the RADAR-AD study.

Marijn introduced Altoida – a tablet- or smartphone-based augmented reality app that combines the virtual and real world. It assesses a person’s spatial navigation and memory, planning skills and use of technology. The app includes a motor task, which collects data on fine motor skills, and 2 hide-and-seek tasks. The latter simulate a complex activity of daily living, where participants are asked to hide virtual objects in a room and then find them. During these tasks the Altoida app assesses the movements of the participant as well as how, when and where they place and find objects. The outcome of over 500 variables of data generated by the app are combined with a machine learning algorithm into 1 outcome: the Neuro Motor Index (NMI). This research included 25 “control” participants, 8 preclinical AD and 8 AD participants with mild cognitive impairment (prodromal AD). RADAR-AD researchers compared the outcomes from the Altoida tasks between “control” groups, people with preclinical AD and those with Mild Cognitive Impairment (MCI) due to AD. Additionally, they compared these results with the outcomes from standard clinical tests – neuropsychological assessments and questionnaires assessing functional decline.

An app to measure functional decline in managing finances in AD

Preliminary data outcomes from the use of the Banking App in the RADAR-AD study were presented by Thanos Stravropoulos, Postdoctoral Research Associate at the Centre for Research and Technology Hellas (CERTH), working on RADAR-AD’s device selection and analytics. The Banking App is used to assess the ability of people living with AD to manage finances by using an ATM simulation. Managing finances is an Instrumental Activity of Daily Living (IADL). It is usually measured via traditional pen-paper methods and interviews. The Banking app offers the possibility to simulate this IADL. In the app, users enter a PIN, an amount to withdraw and confirm all inputs, using a numpad on a tablet’s touchscreen that resembles an ATM. The collected metrics include duration and correctness of each step (i.e. PIN, Amount and confirmation) and total attempt. 36 RADAR-AD participants used the app – 20 “controls”, 6 participants with preclinical AD, 6 cognitively impaired prodromal AD and 4 Mild to Moderate AD-diagnosed participants.

The preliminary findings highlight that managing finances via an app relates to cognitive assessment, further supporting the potential to assess this IADL in people diagnosed with early stages of AD using technology. This early data will be investigated in more depth as the study and data collection progress.

Disease modeling performed on Altoida and ADNI data

Meemansa Sood (Fraunhofer Institute for Algorithms and Scientific Computing (SCAI) presented RADAR-AD’s current results from disease modeling performed on Altoida and ADNI data.

Results

RADAR-AD researchers were able to disentangle non-trivial relationships between different types of digital measure (DMs) reflecting performance across distinct neurocognitive domains, as well as individual tasks obtained from a virtual reality game. The identified associations were statistically significant and predictable. These associations allowed RADAR-AD researchers to simulate DMs for individual ADNI patients. This analysis revealed additional links between DMs, MMSE subitem scores, FAQ subitem scores, as well as indirect links to imaging diagnostics and CSF biomarkers.

From preliminary results to a full data set

The research presented at the conference is promising, but contains preliminary results. All results should therefore be interpreted with caution. Over the months, the consortium aims to enroll more participants in the different studies and perform more and different analyses. Results coming from the study will be published in peer-reviewed scientific papers, and published on the RADAR-AD website.