Individualised CCT (MAKSCog) will lead to statistically significantly greater improvements in cognitive capacities over time in people with MCI as compared with basic CCT (active control group).
Exploratory study question
In an open phase of the study, we will investigate long-term cognitive status yearly.
Study design and setting
A prospective double-blind randomised controlled intervention study is being conducted to test the abovementioned research hypothesis. The overall start date of the study was on 01 April 2019. Recruitment began on 16 March 2020 and will continue until 31 January 2021. The study is being carried out in the metropolitan region of Erlangen/Nürnberg, Bavaria, Germany. At baseline, all study participants are randomly assigned to the intervention (individualised CCT; MAKSCog) or active control group (basic CCT). The reasons for choosing an active control group are as follows: First, since the principal usefulness of CCT is well known [11,12,13,14,15,16], it would be unethical to use a control group without CCT. Second, by using an active control group, we will be able to evaluate the additional effect of the learning system. After baseline testing (t0), the study participants receive one of the two computerised training applications for their computer, laptop or tablet. It is recommended that they use the training at least 30 min per day 3 days a week during the 6-month intervention phase. Both computerised training applications contain the same computerised cognition assessment that will be delivered and collected once a month (t0–t6). After the end of the 6-month intervention phase, the study participants in both the intervention and control groups will be free to decide whether or not to continue using the computerised training. Then the open phase of the study will begin in which we will examine the exploratory study question. Trial registration data are presented in Table 1.
Data are being collected by means of psychometric tests and structured interviews using videoconferencing and telephone. The data are being collected by trained student assistants who have no knowledge of group allocation at any time. The procedure is as follows. One of the researchers prepares a folder with the installation application for each study participant on which the software for the assigned computerised training version is located. Shortly before baseline testing (t0), the study participants receive an email with a link to download the software for their version of the computerised application and instructions on how to download and install the software. The participants are not told which of the two applications they are using. At follow-up (t6), after the assessment has been completed, each study participant will also receive an application-specific questionnaire about the usability of the software.
Sample size and effect size estimation
A power analysis was computed with 100 study participants who were distributed to the two groups of the 2×2 factorial variance-analytic experimental design with one repeated measurement (factor 1: two CCTs; factor 2: two time points). With 50 study participants in each group, alpha = 0.05, beta = 0.20 (corresponding to a power of 80%), a correlation between repeated measures of 0.5 and a nonsphericity correction of 1, we will have the power to detect effects with an effect size of f ≥ 0.14 (comparable to Cohen’s d ≥ 0.28).
Participants are being recruited from the general population through individual screening dates. Information channels are media alerts about two newspapers and a magazine for seniors, a regional senior club of retired people, and a health insurance company. The participants are being informed about the project and the planned study in a personal conversation (videoconferencing or phone call).
Eligibility of participants
Individuals who are interested in the study contact the study centre. We offer them an appointment for screening, including a personal conversation about their screening results afterwards. People who fulfil the criteria for inclusion are informed about the study and asked to take part in the project.
Criteria for inclusion are (1) MCI, psychometrically operationalised by a score on the Montreal Cognitive Assessment (MoCA) ≤ 24 (cut-off for cognitive impairment) and at the same time a score on the Mini-Mental State Examination (MMSE) ≥ 24 (cut-off for no dementia); (2) possession of a computer, laptop, or tablet and basic skills in their use; (3) age ≥ 60; and (4) informed consent. Criteria for exclusion are (1) completely blind or deaf; (2) no personal computer, laptop or tablet; (3) normal cognition, operationalised by a score on the MoCA > 24; (4) dementia, operationalised by a score on the MMSE < 24; (5) acute depression, operationalised by a score on the 9-Item Patient Health Questionnaire (PHQ-9) ≥ 12; or (6) other psychiatric or neurological diseases (checklist): psychosis, (schizophrenia, major depression, mania, bipolar psychosis), Parkinson’s disease, multiple sclerosis, several strokes, alcohol abuse/drug abuse (addiction) and other serious brain diseases (especially brain tumour, brain injury, hydrocephalus). The drop out criteria are (1) withdrawal of consent, (2) serious illness or (3) death of the participant.
The MMSE and the MoCA are administered in combination in order to differentiate between normal cognition, MCI, and dementia. The MoCA is administered first in order to differentiate between normal cognition and MCI on the basis of the cut-off score of 24 points . The MMSE is administered in order to differentiate between MCI and dementia on the basis of the cut-off score of 23 points . For these cut-offs, we looked for an optimised ratio of sensitivity and specificity. The criteria for individuals who were positively screened for MCI, normal cognition, or dementia are shown in Table 2.
Our external biostatistics partner is creating computer-generated randomisation lists (Institute of Medical Informatics, Biometry, and Epidemiology, Friedrich-Alexander University Erlangen-Nürnberg, Waldstraße 6, 91054 Erlangen). All individuals meeting the inclusion criteria are randomised into one of the two groups (individualised CCT or basic CCT). Randomisation is software based and stratified by sex. Couples are assigned to the same group.
Participants do not know which treatment condition they are in, and the student assistants who assess the outcomes of the study are blinded to participants’ allocation.
For each intervention group, a set of download links was created and assigned to the participants in the according group. Similarly, any group-specific test material, e.g. the application-specific usability questionnaire at t6, is assigned by the study team in a sealed envelope.
Both computerised applications (intervention and control) are available for Windows PC/laptop and Android tablet and do not require a connection to the Internet. An Internet connection is needed only to install the application. The participants receive a download link for the respective application and the installation instructions one day before baseline testing. After baseline testing, the student assistants are available to help with the installation. If technical support is needed and the application cannot be installed at baseline, then the start date is set to the day of a successful installation.
Individualised CCT for people with MCI (MAKSCog)
The exercises included in this training application have been selected to address the expected level of performance of individuals with MCI. All exercises are available with different levels of difficulty. The playful exercise tasks involve the basic parameters of information processing as well as short-term memory and require different types of decision-making. This application involves ten exercises that focus on different key functions (see Table 3): sustained attention, visuoconstructional reasoning, working memory, recognition memory, visual perception, implicit learning and word finding. These train low to higher cognitive functions. The initial difficulty levels of the exercises are determined by a learning system, which uses (a) a (logistic regression) model that is based on data from people with MCI (individualised by taking into account each participant’s data) and (b) the cognitive status of the participant (i.e. the results of the integrated computerised cognitive assessment) to estimate the likelihood of a participant’s success at a certain difficulty level for a task. The initial model is based on data collected prior to the study: People diagnosed with MCI used the CCT without the learning system. The application chooses the highest level the participant is likely to solve as the entry level. With the learning system, individual (compensation) strategies are nullified, and the ideal level of difficulty for training is generated for each participant.
Basic CCT (active control group)
This training application uses exercise tasks that are oriented towards parlour games and cognitive tasks for everyday life (see Table 3). The exercise tasks, like those of the individualised CCT, are playfully designed and require, among other things, simple strategies, basic arithmetic operations and long-term memory. This application involves five exercises that focus on the following key functions: visuoconstructional reasoning, working memory, semantic and autobiographical long-term memory and planning. Most of the exercises (e.g. the dice game) exist with only a single level of difficulty. The entry-level difficulties of the other exercises are determined solely by the participant’s prior successful results on this exercise. The exercises of the basic CCT aim to provide enjoyable computerised leisure activities with a limited number of cognitive tasks for the active control group.
The data are being collected at baseline and follow-up by student assistants (psychology students) trained to conduct performance tests and interviews. The measures that are being used at the different measurement points are shown in Fig. 1.
Primary outcome measure
Montreal Cognitive Assessment (MoCA) 
The MoCA is a performance test that is used to screen for MCI. It consists of more difficult items than the MMSE and is thus able to better detect MCI [26,27,28,29]. The score ranges from 0 to 30 points, with higher scores indicating better cognitive performance. A score ≤ 24 indicates cognitive impairment . There are three parallel versions of the MoCA currently available. The MoCA has been found to be an appropriate measure for cognitive screening and has good values for validity and reliability .
Secondary outcome measures
Mini-Mental State Examination (MMSE) 
The MMSE is the most frequently employed screening test for dementia . It measures five areas of cognitive functioning: orientation, registration, attention and calculation, recall, and language. The score ranges from 0 to 30 points, with higher scores representing better cognitive performance. Values above 23 points are interpreted as ‘not demented’, whereas scores between 0 and 23 indicate a dementia syndrome . The reliability and validity of the MMSE has been established in numerous studies, e.g. [25, 33, 34].
Computerised cognitive assessment
Both versions of the computerised training application contain a set of exercises for measuring different cognitive abilities monthly, beginning at baseline. Eight tests are used to measure various cognitive abilities (see Table 4).
The PHQ-9 is a short self-assessment tool often used in primary care settings to screen for depression . Its nine items cover the nine DSM-IV criteria by asking patients about their experiences during the last 2 weeks and are rated on a four-point scale ranging from 0 (‘not at all’) to 3 (‘nearly every day’). The total sum score suggests varying levels of depression. A cut-off ≥ 12 was found to show a good balance between sensitivity and specificity . The PHQ-9 was found to be a reliable and valid instrument for screening for depression .
Questionnaire on sociodemographic and health-related data
The following data are being recorded in a standardised questionnaire by the student assistants at baseline: sociodemographic data (age, sex, marital status, highest educational level, employment status, monthly income, household size), modifiable risk factors for MCI (status of general mental activities, physical activities, social participation, sleeping habits, average liquid intake, eating habits, alcohol consumption, nicotine consumption, visual/hearing capacity), non-cognitive symptoms (according to the symptoms of the Neuropsychiatric Inventory ) and health-related data (diseases, medications, body weight, body height, dementia cases in family history).
User Experience Questionnaire (UEQ) 
The UEQ measures attractiveness, perspicuity, efficiency, dependability, stimulation, and novelty of software with 26 bipolar items. The questionnaire consists of pairs of contrasting attributes (e.g. ‘understandable’ vs. ‘not understandable’) that can be rated on a 7-point Likert scale. The UEQ was found to show a satisfactory level of reliability and construct validity .
Application specialised questionnaire
An application specialised questionnaire is used to identify usability issues. For each exercise, subjective difficulty, specific usability issues, and attractiveness are rated on a 5-point Likert scale. The usefulness and efficiency of the task instructions have to be rated as well.
Additional digital data
Both CCTs track usage data. The usage data include the duration of use, difficulty, success, and other parameters for each training task run.
The data are being collected at baseline (t0) and at the end of the 6-month intervention period (t6). As the span between screening and t0 is approximately 2 weeks, we use the MoCA and MMSE of the screening as baseline values. For the exploratory study question, the data are being collected on a yearly basis after the baseline assessment with no specified end point (see Fig. 1). Due to the COVID-19 pandemic, we decided to eliminate the risk of infection related to study participation. The trial will be conducted entirely virtually. All data will be generated via videoconferencing, telephone, or the computerised cognitive assessment that is integrated into the CCTs. The participants were instructed in an invitation mail or letter to prepare an undisturbed environment for the videoconferences. However, participants lacking the hardware for a videoconference or those who are not willing to take part in an assessment by videoconference will be given the option to come into the clinical research centre for their screening visit or follow-up. Testing with the MoCA and MMSE will be conducted via videoconferencing with the student assistants. Videoconferencing assessments with the MoCA and MMSE have very high reliability scores compared with face-to-face testing. The intraclass correlation coefficients (ICC) for the MoCA and the MMSE have been demonstrated in several studies and go up to ICC = 0.99 for the MoCA  and up to ICC = 0.92 for the MMSE . The questionnaire on sociodemographic and health-related data will be sent to the study participants to prepare them for the telephone interview. The data collected by the computerised cognitive assessment during the 6-month intervention period will be obtained from the study participants.
Data quality management
The student assistants involved in the study have been thoroughly trained for their tasks by the study headquarters’ staff. When the study participants have questions concerning the computerised interventions, they can contact the study headquarters by email. The quality of the data is guaranteed by strict data monitoring at the study headquarters for the total study period. Plausibility checks and logical considerations about the relationships between associated variables will be performed. Regular backups will be carried out and saved on an external hard drive.
Patient and public involvement
Study participants or the public are not involved in the development, design, or conducting of the study. The public has been informed about the recruitment process and the study. In order to recruit study participants from the general population, we provided extensive information about our study through local newspapers, a local magazine for older adults, members of a regional health insurance company, and a regional senior club for retired people. The local newspapers will provide information about the study participants’ experiences with the computerised intervention at the end of the 6-month intervention period.
All relevant data, sociodemographic, health-related, primary, and secondary outcome variables will be reported descriptively. In order to be able to assess the quality of the randomisation, the baseline data from the intervention and control groups will be tested for statistically significant differences. For the multivariate analyses, we will impute missing values using the expectation maximum algorithm. The primary hypothesis will be tested by calculating multivariate analyses according to the general linear model. To ensure the robustness of the results, we will perform both analysis strategies ‘intention to treat’ and ‘per protocol’. ‘Intention to treat’ evaluations are carried out with all cases still alive at the end of the intervention or observation period. The significance level is defined as alpha = 0.05. The data analyses will be performed using the ‘IBM SPSS Statistics 24’ software. The members of the data monitoring committee are Prof. Dr. Olaf Gefeller and one of his statistics experts from the Institute of Medical Informatics, Biometry, and Epidemiology, Friedrich-Alexander Universität Erlangen-Nürnberg, Waldstraße 6, 91054 Erlangen.
All legal matters such as voluntariness, right of revocation, and General Data Protection Regulation (European Union) are taken into account. People with MCI are independent and fully capable of conducting business and giving consent. Upon agreement, consent to participate (written informed consent) will be obtained from all participants by the student assistants who are members of the study centre. All participants are informed about the study in a personal videoconference after they are screened for eligibility. A participant information sheet including important information about participation (e.g. randomisation, data protection, data storage) is given to every participant. The opportunity to ask questions will be granted by videoconference, telephone, and email afterwards at any time. Participants are not being given any financial inducement to participate.
The external Reinhard Frank-Stiftung is continuously being informed about the progress of the study. In the case of important protocol modifications, we will inform the Ethics Committee, the funder, and the platform for the trial registry.
The informed consent sheet and other personal data are stored in a locked steel cabinet. Only members of the study team have access to the lists of participants’ names and codes. All data are stored in only a pseudonymised form digitally on a firewall-secured server at the University Hospital Erlangen. In order to be able to clearly assign data from different study participants to the previous data records over the course of a longitudinal study, a key file containing the connection between real names and pseudonyms is necessary. The key file is stored on an offline computer and in printed form in a locked metal cabinet. Access to the room (electronic lock) with the metal cabinet is only possible for a limited and precisely defined number of employees of the University Hospital. Access to the key file stored in the metal cabinet is only available to project employees who are obliged to maintain secrecy towards third parties through their employment contract. A transfer of data to third parties (e.g. other research groups) is not planned because it was not included in the information sheet. Results of the study for scientific or other publications are only published in aggregate form (mean values, etc.). No published material will contain patient-identifying information.
The CCT applications might have an impact on existing excessive computer use. However, both CCT applications that we developed are not based on motivational or emotional components. The CCT applications require cognitive performance, which could instead lead to exhaustion.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.