Participants
A sample of 276 individuals older than 50 years participated in this study: 154 CH individuals and 122 subjects diagnosed with MCI, of whom 61 were non-amnestic (naMCI) and 61 amnestic (aMCI). All of them were evaluated at Fundació ACE, Institut Català de Neurociències Aplicades (Barcelona, Spain), a non-profit Alzheimer’s institution that provides diagnostic, treatment, and patient management services to the Catalan Public Health Service [17]. All participants underwent a complete neuropsychological assessment using the Neuropsychological Battery of Fundació ACE (NBACE), whose normative data and cutoff scores have been reported elsewhere [18]; a neurological history and examination; a semi-structured psychosocial interview, including a functionality assessment by the Blessed Dementia Rating Scale (BDRS) [19, 20]; and a diagnosis of their cognitive status by the clinical team from the Memory Unit at a daily consensus conference [17]. The sample comprised 64 participants enrolled in the Fundació ACE Healthy Brain Initiative (FACEHBI) study [21], 58 participants in the BIOFACE cohort, 42 individuals recruited from the Open House Initiative (OHI) [22], and 112 subjects evaluated at Fundació ACE’s Memory Unit.
The inclusion criteria for the whole sample were the following: age over 50, an educational level of at least elementary school (in order to ensure the correct understanding of the FACEmemory® instructions), a clinical diagnosis of CH or MCI [17], and completion of FACEmemory®.
The clinical diagnoses for the CH group were as follows: absence of objective cognitive impairment, with average or above-average scores on NBACE [18, 23]; normal general cognition (Mini-Mental State Examination (MMSE) score ≥ 27) [24, 25]; a Clinical Dementia Rating (CDR) [26] of 0; and no history of functional impairment due to declining cognition, with a score below 4 on the BDRS [19, 20].
The clinical diagnosis for the MCI group were as follows: subjective cognitive complaints, essentially preserved general cognitive function (MMSE score ≥ 24) [24, 25], preserved performance in activities of daily living (BDRS < 4) [19, 20], absence of dementia; a CDR [26] of 0.5, an objectively measurable impairment in memory or another cognitive function (aMCI or naMCI) [12, 27], and the absence of prescribed symptomatic treatment for dementia (i.e., acetylcholinesterase inhibitors or memantine).
Participants were not required to have previous knowledge of tablet computer use. To ensure the correct understanding of the FACEmemory® instructions and the completion of the test, exclusion criteria included an educational level below elementary school and significant auditory or visual abnormalities, such as glaucoma, cataracts, or severe aphasia.
As mentioned above, a subsample of 65 individuals was also administered the S-FNAME (paper-and-pencil version with 16 items). That is, they completed FACEmemory® and S-FNAME on different days (no more than 2 months apart).
Finally, 65 subjects underwent lumbar puncture (LP) to measure AD biomarkers in CSF.
In collaboration with Dr. Rentz’s team, we transformed the original paper-and-pencil FNAME-12 [9] into a self-administered computerized version (named FACEmemory®) with images, names, and occupations representative of the Spanish population. The test was administered using a tablet computer with voice recognition and touchscreen, enabling us to immediately score and register the results anonymously in a database. The scores of all variables ranged from 1 to 12, and the total FACEmemory® scores ranged from 1 to 96.
The FACEmemory® procedure
All subjects completed FACEmemory® on a tablet with voice recognition, a touchscreen, and scoring registration in Fundació ACE. The temporal sequence of FACEmemory® was the following: two learning trials, a short-term memory task, and a long-term memory task that included face, name, and occupation recognition. The total test duration was approximately 30 min.
After the study, participants completed a satisfaction survey to indicate their level of satisfaction with the test. The survey consisted of 5 questions and required the participants to score their satisfaction from 1 to 5 (1 being the worst score), to choose what aspect of the test they disliked the most, to indicate whether they would recommend the test to family or friends, to select what aspect they liked the most, and to mention whether they would like to repeat the test in the future.
FACEmemory® was self-administered with minimal supervision from a psychologist (who only addressed technological issues). The psychologist simultaneously scored the test manually to determine whether the automatic scoring matched the hand scoring.
As shown in Fig. 1, the first learning trial consisted of a total of 12 faces, each one associated with a name and an occupation that appeared beneath it for 8 s. The second learning trial was identical, except that the faces appeared in a different order. The participants were instructed to read the name and occupation appearing beneath each face aloud and to try to remember it. Then, the application asked the participants to press the red microphone button and to say the name (LN1/LN2) and occupation (LO1/LO2) they remembered as being associated with each face. If they did not remember anything, they were allowed to say “I don’t know.”
For the evaluation of short-term memory, 2 min after the second learning trial, the application again asked the participants to say the name (RSN) and occupation (RSO) they remembered as being associated with each face, but if they did not remember anything, they could say “I don’t know.” They had to press the red microphone button and to answer.
Finally, 20 min after the second learning trial (LN2/LO2), the application started the long-term memory assessment, which involved free recall and recognition tasks. First, the participants were instructed to recognize, out of 3 faces, the face that had appeared in the first learning trial and to touch it (FR). Then, the correct face appeared, and the application asked the participants to say the name (RLN) and occupation (RLO) they remembered for each face. After each answer, a screen appeared, showing the right face and 2 rows beneath it. Each row had 3 name and 3 occupation options. The participants were instructed to touch the name (REN) and occupation (REO) they remembered as being associated with that face.
The application registered the touched responses and the voice of the participants automatically and entered the scores into a database anonymously.
Application development
The application was developed with Android language and using Google Books for voice recognition and interactions with the device. Google voice recognition libraries were used, as they are the most complete and because they allow the use of different languages.
During development, screens were constantly checked to ensure they were clear, so that the user’s attention would not be distracted while they were learning the data presented and so that the test would be as reliable as possible. The application has been designed so that when the user interacts with it using voice commands, the elements that appear on the screen are minimal. The FACEmemory® application is linked to a control panel or back end, from which the saved data from all the tests performed can be viewed and extracted, and filtered by date or test. Access to the back end requires a username and password.
One of the most important features of the FACEmemory® application is that it keeps the user’s data completely anonymous for better protection. Despite this anonymity, the application is configured so that a user can perform multiple tests at different times. When a participant finishes FACEmemory®, the application generates a code Fundació ACE saves on a separate system that complies with the General Data Protection Regulation (GDPR) and that connects the user to the test carried out. Therefore, there are data saved from individual tests or tests related to others saved in the system, but no personal data, thus complying with the GDPR at all times. Security and data encryption measures have been also applied to communications between system elements, to make both the tablet application and the back office safe and robust.
In summary, the FACEmemory® application allows the user to focus their attention on the test, as well as allowing Fundació ACE to manage the data in a simple and anonymous way.
Lumbar puncture and cerebrospinal fluid collection
This protocol followed the consensus recommendations established by the Alzheimer’s Biomarkers Standardization Initiative [28]. Briefly, those participants who agreed to undergo a LP in Fundació ACE had the procedure performed by an experienced neurologist, with the patients in a seated position and under fasting conditions. After applying local anesthesia (1% mepivacaine) subcutaneously, the neurologist obtained CSF by LP in the intervertebral space of L3–L4. The fluid was collected passively in two 10-ml polypropylene tubes (Stardest Ref. 62610018). The first tube was analyzed externally for basic biochemistry (glucose, total proteins, proteinogram, and cell type and number). The second tube was centrifuged (2000×g 10 min at 4 °C), and the fluid was aliquoted into polypropylene tubes (Stardest Ref. 72694007) and stored at − 80 °C until analysis. The time delay between CSF collection and storage was less than 2 h.
On the day of the analysis, the aliquots were thawed at room temperature and vortexed for 5–10 s to determine AD biomarkers in CSF [29]. One aliquot/patient was used to determine the concentrations of Aβ1-42, t-tau, and p181-tau using the commercially available enzyme-linked immunosorbent assays (Innotest, Fujirebio Europe) at the Research Laboratory of Fundació ACE.
Statistical analysis
Statistical analyses were performed using SPSS (version 20.0; SPSS Inc., Chicago, IL). Univariate analysis of variance (ANOVA) with post hoc comparisons (Bonferroni) was used to compare sociodemographic data between the CH, naMCI, and aMCI groups. An ANOVA adjusted by years of formal education, age, and sex (ANCOVA) was carried out to compare FACEmemory® scores between the CH, naMCI, and aMCI groups.
Logistic regression analyses were carried out to search for discrimination indexes, in which the diagnostic comparisons between CH/MCI, CH/aMCI, and CH/naMCI individually were the dependent variables, and the total FACEmemory® score was the independent variable. Moreover, diagnostic sensitivities, specificities, and total FACEmemory® cutoff, as measured by the analysis of the receiver operating characteristic (ROC) curve, were obtained for the purpose of discriminating between CH/MCI and CH/aMCI. Cutoff points for the whole sample were established by calculating the sensitivity and specificity for the total FACEmemory® score. ROC analysis was used to calculate the optimal cutoff value between CH and the MCI and aMCI groups. Moreover, to calculate the sensitivity and specificity of the cutoff score, an area under the curve (AUC) analysis was also carried out and yielded a confidence interval of 95%. Our goal was to obtain an AUC greater than 0.75 for the total FACEmemory® variable and for the sensitivity and specificity values.
To ensure the reliability of the automatic scoring, Pearson’s correlation analyses were carried out between hand and automatic (obtained from voice recognition/touchscreen) scorings of the FACEmemory® test. We also correlated performances on FACEmemory® with those on the S-FNAME (paper-and-pencil version with 16 items) to examine convergent validity. Finally, partial correlation analyses were performed between the FACEmemory® scores and CSF AD biomarkers, adjusting for age, sex, and education.