Investigators will run a three-arm, parallel-group randomized controlled trial at Shanghai University of Sport to test whether exercise timing alters cognitive outcomes in older adults with type 2 diabetes mellitus and mild cognitive impairment. Participants aged 60–75 who meet Petersen’s criteria for MCI and remain functionally independent will be randomized to morning exercise (ME), evening exercise (EE), or a health education (HE) control. The intervention period is 12 weeks with assessments at baseline and post-intervention; post-tests are scheduled within 48–72 hours of the final session. The full assessment battery — cognitive testing, blood sampling, physical fitness measures, and multimodal MRI — will be completed in fixed morning windows to standardize circadian influences on measurement.
A study flowchart and enrolment timeline are provided in the protocol. Recruitment methods include hospital and community outreach. The source reported that, as of August 18, 2026, 56 participants had been enrolled. The trial is registered (ChiCTR2300073737) and received institutional ethical approval.
The ME arm trains from 09:00–10:10. The EE arm trains from 19:00–20:10. Both exercise arms follow an identical 12-week, home-based, remotely supervised program delivered live via Tencent Meeting. Each session contains 5 minutes warm-up, 30 minutes aerobic exercise, 30 minutes resistance exercise, and 5 minutes cool-down, performed three non-consecutive days per week. The HE control receives weekly 30-minute education sessions on diabetes self-care, glucose monitoring, and medication interpretation.
Aerobic sessions use music-guided calisthenics with target intensity prescribed as a percentage of heart-rate reserve (HRR): Stage 1 at 40–50% HRR, Stage 2 at 50–60% HRR. Resistance training employs elastic bands in a full-body circuit of 10 exercises, progressed from two sets of 15RM in Stage 1 to two sets of 12RM in Stage 2. The protocol specifies a standardized grip‑width scale on each band to quantify resistance and uses volitional fatigue to verify RM targets. Heart rate is monitored continuously with Polar® equipment; perceived exertion is recorded via the Borg scale after each session. Participants keep cameras on for real-time supervision and receive pre-session reminders and group messaging for adherence support.
The trial’s primary outcome is global cognitive function measured by the Montreal Cognitive Assessment (MoCA). The MoCA score range is 0–30; higher scores indicate better global cognition. A comprehensive secondary battery will probe cognitive subdomains: the N‑back task (working memory), the Color‑Word Stroop Test (executive control), the Attention Network Test, Trail Making Tests A and B (processing speed and set-shifting), and the Digit Symbol Substitution Test.
Neuroimaging will include T1‑weighted structural imaging, resting‑state functional MRI (rs‑fMRI) to assess functional connectivity, and diffusion tensor imaging (DTI) to evaluate white‑matter integrity. The protocol states that MRI scans will not be performed on participants with contraindications, but those individuals remain eligible for non‑imaging outcomes. Blood biomarkers, including glycemic measures, and physical fitness testing (including oral glucose tolerance) are scheduled at baseline and post-intervention. A 3‑day food diary will be collected before each assessment to control for dietary influences.
The study team outlined two primary hypotheses. First, they hypothesize that evening exercise will produce greater improvements in cognitive performance, brain structure and function, and glycemic control than morning exercise. Second, they propose that improvements in glycemic control will mediate the relationship between exercise and cognitive/brain outcomes. The protocol indicates planned mediation analyses to test whether glycemic indices account for part of the exercise-induced cognitive and neuroimaging changes. The source did not provide statistical details, such as the specific mediation models or power calculations.
Block randomization (block size six) will allocate participants 1:1:1 to ME, EE, or HE using an independent online randomisation system and a sequence generated by an independent statistician. Given the nature of the interventions, participants and fitness coaches cannot be blinded to group assignment. Outcome assessors and the data analysis team will remain blinded; fitness coaches are separated from outcome assessment personnel to protect blinding integrity. The trial will be unblinded only after statistical analyses are complete.
Recruitment began in August 2024 and will continue through October 30, 2026, per the protocol. Ethics approval and trial registration details are documented in the source.
Safety measures are embedded in the remote protocol: continuous heart‑rate monitoring during sessions, real‑time webcam supervision, RPE checks after exercise, and an immediate stop rule for adverse symptoms such as palpitations, hypoglycemia, pain, or dyspnea. Attendance is recorded for every session. The HE arm receives weekly contact, reducing differential attention bias.
The protocol identifies several limitations that will influence interpretation. Blinding of participants and coaches is not feasible. MRI scans are omitted for participants with contraindications, introducing potential differences between imaging and non‑imaging subgroups. The source document does not report sample size calculations or statistical power parameters; those details were not provided. The remote delivery model relies on participant access to and familiarity with digital platforms and continuous camera monitoring, which may affect generalizability. Finally, the protocol describes planned mediation testing by glycemic control but does not specify analytic methods or handling of missing data in the source.
These constraints define the immediate unanswered questions the trial will need to address upon completion: whether the observed cognitive and brain changes differ by imaging availability, how missing data will be managed, and whether the mediation by glycemic control persists after adjustment for behavioral and dietary covariates. The trial’s registered results and subsequent publications will be required to answer these items.