As a behavioral, nonpharmacological intervention aimed at improving cognitive health and well-being in aging, prospective registration was not initially anticipated by the authors, and the study was registered on ClinicalTrials.gov (NCT07108413) only after completion of data collection. Data collection started in January 2024 and finished in April 2025. The protocol registration was submitted in July 2025 and published in August 2025. The aim of this retrospective trial registration and publication is to document this extensive protocol in detail, including a novel intervention. This will allow the protocol to be reproduced, and it will assist with the reporting of the results that will be divided into several publications focused on the analyses of the different outcomes and different aspects of the intervention.

No public involvement is planned for any stage of the study, including in the protocol design.

Our study consists of a 3-arm controlled trial with a parallel study design involving cognitively healthy older people. Participants are allocated into the Activ4Brain group and 2 control groups: an active control group engaged in AE alone and a passive control group, instructed to maintain their daily routines. The passive control group is included to control for the effect of test-retest. The AE group is included to assess if including cognitive training leads to a greater effect on cognition than AE alone and to control for the participants’ motivation to engage in the evaluation sessions, which might be decreased in passive control groups. The group allocation ratio is 1:1:1. The intervention consists of 24 sessions, held twice a week for 3 months. The participants are tested at baseline and after the 3-month training programs (Figure 1).

The study protocol was approved by the Faculty Ethical Committee of the Faculty of Sports Science and Physical Education of the University of Coimbra with the reference number CE/FCDEF-UC/00082023. All participants are asked to read and sign a consent form. This protocol is reported in accordance with the SPIRIT (Standard Protocol Items: Recommendations for Interventional Trials) 2025 guidance (see Checklist 1 for a completed checklist). The results will be reported following the CONSORT (Consolidated Standards of Reporting Trials) extension for nonpharmacological interventions [24].

Participants are recruited from Portugal’s Coimbra District through personal communications, press releases, social media posts, and distribution of informative leaflets and posters. Recruited participants receive comprehensive information regarding data collection procedures, exercise protocols, and measures of protection of personal information procedures. Participants who volunteer to participate in the study are interviewed for eligibility and asked to sign an informed consent form.

Inclusion criteria are cognitively healthy men and women between 55 and 75 years of age, with a physical condition compatible with the practice of moderate AE. Exclusion criteria include any disability or health problem incompatible with moderate AE and objective cognitive impairment. The Physical Activity Readiness Questionnaire is used to screen contraindications for physical exercise. Before the intervention, we ensure participants’ fall risk is low using the Timed Up and Go (TUG) test [25].

Participants are instructed to continue any ongoing concomitant care. Details of the concomitant care received are recorded at baseline and, to assess any changes, at follow-up.

Sample size determination has been made using the G*power (version 3.1.9.7; written by Franz Faul, Universität Kiel, Germany) program. As reported in previous studies [26,27], we expect a small to moderate effect size of the Activ4Brain program on cognitive function (our primary end point). We calculated a sample size of 66 participants by setting Type I error (α) at 0.05, Type II error rate (1−β) at 0.95, and an effect size η² of 0.06 (corresponding to an effect size f of 0.25, using the option for effect size specification as in G*Power 3.0) for repeated-measures ANOVA, within-between interaction, with 3 groups and 2 repeated measures. Accepting a dropout rate of 10%, 75 participants are recruited.

This trial is a quasi-randomized controlled unblinded trial. After recruitment, participants are assigned to 1 of 3 groups based on sex and age, ensuring that each group has similar sex ratios and age ranges. These balanced groups are then randomly assigned, using the random permutation of integers function from MATLAB (The MathWorks Company Ltd), to 1 of the 3 interventions: Activ4Brain, AE, or control.

The AE and Activ4Brain groups engage in 50-minute exercise sessions of moderate intensity (60%‐80% of maximal heart rate), including a 5-minute warm-up, 40 minutes of AEs, and a 5-minute cool-down. The warm-up period includes similar body movements as the main part of the exercise and dynamic stretching, while the cool-down period consists of static stretching. The main part of the exercise session consists of 4 blocks (Table 1). The sessions of the AE group are composed of a sequence of the four 5-minute blocks of AE repeated twice. In the Activ4Brain group sessions, the blocks consist of 5-minute AEs and 5 minutes of cognitive training (Table 2) and are performed only once per session. The Activ4Brain group performs exercise that includes the same rhythmic movements matched in intensity range and duration to the exercise performed by the AE group. The Activ4Brain participants are instructed to keep moving, marching in place, while performing the cognitive training tasks to maintain the increased heart rate. Each group class is composed of 4 participants with a narrow age range to ensure similar physical and cognitive performance.

In both training groups, music with a tempo of 130 to 135 beats per minute-32 count is used to regulate rhythm and enhance motivation [28]. The heart rate of every participant is monitored during the exercise classes with the Polar Team Pro System (Polar Team Pro; Polar Electro), which enables the simultaneous tracking of multiple participants and the monitoring of the exercise intensity in real time, ensuring adequate exercise intensity across all groups.

The Activ4Brain program includes cognitive training tasks targeting 5 cognitive domains: processing speed, inhibitory control, selective attention, working memory, and decision-making. To maintain participants’ motivation and ensure that the tasks are challenging as participants improve their skills, the tasks’ difficulty level is increased if the class average accuracy for the previous week exceeds 80%. The cognitive training tasks were designed using Psychophysics Toolbox Version 3 [29,30] in MATLAB. In each session, 4 cognitive domains are trained in blocks of 5 minutes each (Table 2). At the end of each week, participants are sent by email their average performance metrics for that week, consisting of the accuracy in each cognitive training task, median reaction time, and reaction time variability (SD), with a written explanation of the meaning of each measure.

For the cognitive training, each participant is assigned a set of 4 pedals connected to a computer that records the pedal presses and execution time (Figure 2). The pedals are USB foot switches for PC computers (FS2017 Pcsensor USB Foot Pedal Control Switch Keyboard Adapter). Each pedal is assigned a number (1-4) and a color (white, green, blue, or yellow). These are fixed throughout the 24 program sessions (Figure 2). The participants stand side-by-side at the front of the room facing an LCD TV screen where task instructions and visual stimuli are displayed in adequate size. Participants are instructed to respond by pressing the pedals according to each task instruction. Detailed descriptions of the cognitive training tasks are included in Multimedia Appendix 1

To assess if the Activ4Brain intervention improved performance on the trained cognitive tasks, the Activ4Brain group is tested before and after the 3-month program on the first difficulty level of the following tasks: processing speed, inhibitory control, selective attention, working memory, and decision-making. The tests are performed in the laboratory, 1 participant at a time, using the same pedal setup but without music or any other instructed movements besides pedal presses. In these tests, we assess task accuracy, reaction time (time from stimulus onset to movement initiation), and execution time (time from stimulus onset to pedal press). To calculate reaction time, we use triaxial ankle accelerometers (BIOPAC Systems). For logistical reasons, these tests are only applied to the Activ4Brain group and will be used to determine if improvements in the trained tasks mediate the cognitive changes observed in the cognitive assessment protocol.

Physical evaluation consists of body composition, physical function, and peak oxygen consumption assessment. Tests are performed in the morning between 9 AM and noon, and participants are asked not to participate in vigorous physical activity 24 hours before the tests.

Stature is measured with a stadiometer (Seca Bodymeter, Model 208, Germany) with a precision of 0.1 cm. Body composition (total body mass, fat mass, fat mass percentage, fat-free mass, and phase angle) is measured by bioimpedance (InBody 770; Biospace Co).

To measure the risk of falls and eligibility to perform our exercise programs, TUG, Chair Stand, and Hand Grip Strength tests are applied. TUG and Chair Stand tests are applied according to the Centers for Disease Control and Prevention STEADI guidelines [25,31]. Handgrip strength is measured with a hydraulic hand dynamometer (Lafayette, Model 5030 L1, USA). The strength of the dominant hand is tested in a standing position with straight arms. The test is repeated 3 times, and the best value of the 3 attempts is recorded [32].

Peak oxygen uptake capacity (VO₂ peak) is measured with a submaximal incremental cycle ergometry test (Lode Excalibur, Netherlands). Following a warm-up period, participants start the testing phase with an initial load of 50 watts for women and 75 watts for men [33]. The load is increased by 25 watts at 3-minute intervals. The test is continued at 60 to 70 rpm and is completed once participants achieve 85% of their predicted maximal heart rate, as determined by the Tanaka formula [34].

To examine the effect of the intervention on inflammation and neuroprotection blood markers, blood samples are collected from venous blood (15 mL) in a fasting state in the morning by a registered nurse. A complete blood count is performed using a hematology analyzer (DXH500; Beckman Coulter, USA). Aliquots of plasma and serum samples are stored at −80 °C until further use.

Plasma samples will be quantified by enzyme-linked immunosorbent assay for interleukin (IL)-6, IL-1β, tumor necrosis factor-α, interferon-γ, IL-17, IL-1Ra, IL-10, irisin, and brain-derived neurotrophic factor (Thermo Fisher, UK), while serum samples will be analyzed for insulin growth factor 1, vascular endothelial growth factor, and glial cell line-derived neurotrophic factor (ElabScience, USA) according to the manufacturer’s instructions.

General cognitive function is assessed using standardized neuropsychological tests (Table 3). Participants are screened for the presence of objective cognitive impairment using the Addenbrooke’s Cognitive Examination-Revised, which also incorporates the Mini-Mental State Examination, providing a double score [35], and for the presence of psychopathology using the Geriatric Depression Scale-30 [36,37]. Participants with objective cognitive deficits (according to the available norms) are excluded.

A comprehensive neuropsychological assessment protocol for detailed neurocognitive performance is performed for eligible participants with 4 tests from the Cambridge Neuropsychological Testing Automated Battery Connect Research system: Motor Screening Task, Rapid Visual Information Processing, Multitasking test, and Spatial Working Memory, and the Auditory Verbal Learning Test [38]. This enabled an in-depth characterization of cognitive function, including measures of attention, visual perception, processing speed, verbal learning, memory, and executive functions, particularly working memory and inhibitory control. Lifestyle habits are screened with the Lifestyle Assessment Toolkit [39]. Physical activity habits are assessed with the International Physical Activity Questionnaire-Short Form [40].

The dementia risk score Lifestyle for Brain Health [41] is applied only at baseline to examine the presence of modifiable risk factors for dementia and, together with the measures International Physical Activity Questionnaire-Short Form, Geriatric Depression Scale, and the Lifestyle Assessment Toolkit, is used to examine potential moderators of program efficacy.

To examine the effect of the intervention on brain and autonomic function, we record the EEG, the pupilogram, and the ECG while participants are resting and while they are engaged in a value-based decision-making task (reinforcement learning task, described below). For the resting-state recording, participants sit in front of a computer screen and fixate on a target displayed at the center of the screen for 4 minutes.

Participants’ attendance is recorded for each exercise session. If a participant does not attend 2 consecutive exercise sessions, 1 researcher contacts the participant and encourages them to continue attending the classes. An adherence rate of 60% is necessary for inclusion in the analyses [43].

To ensure maximum safety, the following measures are taken: heart rate monitoring, performing exercise on a soft ground, and exclusion of participants with any health issue not compatible with moderate AE. Any adverse events that occur during the exercise program and study procedures are recorded by the researchers.

At follow-up, the AE and Activ4Brain groups respond to custom-made satisfaction questionnaires and system usability surveys [44-46]. These questionnaires and surveys include questions regarding suggestions for program improvement and scales regarding their perception of the impact of the program on their well-being and their satisfaction with the program.

The study of the effect of the Activ4Brain program on the primary outcome (cognitive function) is confirmatory because we hypothesize, according to previous evidence, that the integrated cognitive-aerobic training program results in larger improvements than AE alone or no intervention (passive control) [47]. All other analyses are exploratory.

Continuous cognitive outcomes will be summarized descriptively with mean (SD) for each group and time. These variables will be analyzed with repeated-measures ANOVA with 1 between-subjects factor (group) and 1 within-subjects factor (time=preintervention vs postintervention). The Activ4Brain intervention will be considered effective in improving cognitive function if the group × time interaction is significant at a level of .05, if the post hoc tests (with CI adjustment using Bonferroni) show a significant effect of time for the Activ4Brain group, and the improvement is higher for this group than for the other control groups. If the data do not meet assumptions of sphericity, the Greenhouse-Geisser correction will be used. Description of the effect sizes by partial η² will follow the following cutoff points: less than 0.06 will be considered a small effect; between 0.06 and 0.14 will be considered a medium effect; and higher than 0.14 will be considered a large effect [48].

Missing data will be assessed prior to analysis. Analyses will be conducted using the available data (complete case analysis), and participants with missing data for a specific outcome variable will be excluded only from the relevant analysis.

A unique code will be attributed to each participant, and the data will be registered in databases separate from any identifiable information. For each stage of data collection, the validity and integrity of the collected data will be checked by at least 2 researchers. The researchers will validate data ranges and cross-check the entries in the databases.

Due to the trial’s short duration and minimal risks, there will be no requirement for a Data Monitoring Committee or interim analyses.

Twelve months after each publication of trial results, the deidentified individual participant data that underlie the published results will be made available indefinitely in a public repository (eg, Open Science Framework, OpenNeuro) to anyone who wishes to use the data for any purpose.