A Remote Digital Memory Composite to Detect Cognitive Impairment in Memory Clinic Samples in Unsupervised Settings using Mobile Devices

Abstract

INTRODUCTIONMobile app-based unsupervised monitoring of cognition holds the promise to facilitate case-finding in clinical care and the individual detection of cognitive impairment in clinical and research settings. In the context of Alzheimer’s disease, this is particularly relevant for patients who seek medical advice due to memory complaints.OBJECTIVEWe developed a Remote Digital Memory Composite score from an unsupervised remote and mobile cognitive assessment battery focused on episodic memory and long-term recall and assessed its construct validity using a neuropsychological composite score for early cognitive impairment in Alzheimer’s disease, the Preclinical Alzheimer’s Cognitive Composite (PACC5). We also assessed the test-retest reliability of the Remote Digital Memory Composite score across two independent test sessions. Finally, we assessed the diagnostic accuracy of the remote and unsupervised cognitive assessment battery when predicting PACC5-based cognitive impairment in a memory clinic sample and healthy controls.SETTINGThis was an add-on study of the DZNE-Longitudinal Cognitive Impairment and Dementia Study (DELCODE) which was also performed in a separate memory clinic-based sample.PARTICIPANTSA total of 102 study participants were included as healthy controls (HC; n=25), cognitively unimpaired first-degree relatives of AD patients (REL; n=7), individuals with subjective cognitive decline (SCD; n= 48) or patients with mild cognitive impairment (MCI; n=22).MEASUREMENTSWe analyzed results from the objects-in-rooms recall (ORR) test, the mnemonic discrimination for objects and scenes (MDT-OS) test and the complex scene recognition (CSR) test implemented on the neotiv digital platform to derive a Remote Digital Memory Composite. Participants used the neotiv mobile app to complete one unsupervised test session every two weeks on their own mobile device in an environment of their choice. We assessed the relationships of the Remote Digital Memory Composite acquired through the mobile app and in-clinic measures of the PACC5 conducted by trained neuropsychologists in the memory clinics participating in the DELCODE study.RESULTS102 participants provided technically complete data for at least one single session of each of the three test paradigms, of which 87 participants provided data from at least two test sessions of each task. The derived Remote Digital Memory Composite score was highly correlated with the PACC5 score across all participants (r=.75, p<0.001), and also in those without complaints (HC and REL, r=.51, p=0.003) and those with complaints separately (SCD and MCI, r=.76, p<0.001). Good test-retest reliability for the Remote Digital Memory Composite score was observed in those with at least two assessments of the three tests. (r=.74; p<.0001). Diagnostic accuracy for discriminating PACC5-based memory impairment from no impairment was high (AUC = 0.9) with a sensitivity of 0.83 and a specificity of 0.74.CONCLUSIONOur results indicate that unsupervised mobile cognitive assessments in a memory clinic setting using the implementation in the neotiv digital platform has high construct validity and results in a good discrimination between cognitively impaired and unimpaired individuals based on the PACC5 score. Thus, it is feasible to complement neuropsychological assessment of episodic memory with unsupervised, remote assessments on mobile devices. This contributes to recent efforts for implementing remotely performed episodic memory assessment for case-finding and monitoring in large research trials and clinical care.

Unsupervised mobile app-based cognitive testing in a population-based study of older adults born 1944

“…For example, a composite measure of the Boston Remote Assessment for Neurocognitive Health (BRANCH) ( 19 ) was strongly correlated with the PACC5 ( r = 0.62) in cognitively normal participants. Another recent study ( 34 ) used a composite of similar but not identical tasks as used in this study and demonstrated a strong correlation to PACC ( r = 0.51) in cognitively normal individuals without cognitive complaints. A study using the Ambulatory Research in Cognition app ( 15 ) in primarily cognitively normal participants (again using a composite of cognitive measures) reported a strong correlation with a global composite similar to the PACC ( r = −0.53).…”

“…In the course of the follow-up in-clinic examination, 692 of 879 participants have thus far been invited to participate in a two-year study using the neotiv-App ( 12 , 34 ). Of these, 172 participants have signed the informed consent and initiated the mobile app-based cognitive testing to date.…”

“…In ORR-LDR, participants see a computer-generated room with two 3D-rendered objects (see Figure 2 ) ( 12 , 34 ). In the encoding phase, participants learn the position of two objects in a room and are tested for immediate retrieval.…”

Feasibility of Digital Memory Assessments in an Unsupervised and Remote Study Setting

“…Taken together, while age-related memory impairment as well as differences between subjectively stable and declining groups are as expected, future studies combining in-clinic and unsupervised remote assessments need to define convergent and discriminant validity in detail in comparison with established neuropsychological assessment batteries. Recently, we could show that the same memory tests used here, combined to a composite score, are suitable to detect mild cognitive impairment in a memory-clinic population, and show high construct validity with an in-clinic neuropsychological assessment ( 37 ).…”

Feasibility of digital memory assessments in an unsupervised and remote study setting

“…Taken together, while age-related memory impairment as well as differences between subjectively stable and declining groups are as expected, future studies combining in-clinic and unsupervised remote assessments need to define convergent and discriminant validity in detail in comparison with established neuropsychological assessment batteries. Recently, we could show that the same memory tests used here, combined to a composite score, are suitable to detect mild cognitive impairment in a memory-clinic population, and show high construct validity with an in-clinic neuropsychological assessment (Berron et al, 2021).…”