This is a companion real-world external validation study of two pre-registered parent randomized controlled trials (RCTs, protocol IDs: FujianUTCM-1 and FujianUTCM-2). We aim to validate the generalizability and clinical applicability of a pre-developed machine learning prediction model (for Tai Chi intervention cognitive benefit population characteristics and individualized dose prescription) in a real-world community-dwelling population.

Mild Cognitive Impairment (MCI) represents a critical early intervention window for delaying the onset and progression of dementia, and early non-pharmacological intervention has become an international consensus in this field. Our research team has completed two multicenter interventional randomized controlled trials (RCTs) pre-registered on ClinicalTrials.gov and their companion multicenter RCTs. Both studies adopted the standardized 24-form simplified Tai Chi as the intervention, with highly unified intervention procedures, movement specifications, and dose settings. These trials confirmed the dose-response relationship of Tai Chi intervention on cognitive improvement in older adults with MCI due to Alzheimer's disease, and preliminarily clarified the intervention effects of different Tai Chi dose combinations on patients' global cognition and cognitive subdomains.

On this basis, we will further identify the core characteristics of the population with significant cognitive benefits from Tai Chi intervention through machine learning algorithms. Based on data mining from 890 RCT participants, we will construct a dual-function machine learning prediction model for "beneficiary population identification + individualized Tai Chi dose prescription formulation". However, the above conclusions are all derived from standardized RCT settings, and their generalizability and clinical applicability in real-world community settings have not yet been validated.

As a sequential validation study, this research will conduct a cross-sectional survey by enrolling 150 older adults with long-term regular Tai Chi practice from communities in Fuzhou, China, to complete two-dimensional external validation of the previous RCT findings: to verify the consistency and feasibility between the beneficiary population characteristics and corresponding Tai Chi dose prescriptions mined from the parent RCTs, and the real Tai Chi dose combinations of the beneficiary population in the real world.

This study will form a closed loop from "causal evidence mining in RCTs" to "clinical application validation in real-world settings", and provide direct, high-level scientific evidence for the clinical promotion and individualized application of Tai Chi intervention in older adults with MCI.

Inclusion Criteria:

• 1. Presence of mild cognitive impairment, not demented; 2. Age ≥ 60 years old; 3.Regular practice of standardized 24-form simplified Tai Chi for at least 6 months (consistent with the form used in the parent randomized controlled trials [RCTs]) ; 4.Informed consent and voluntary participation.

Exclusion Criteria:

1. Geriatric Depression Scale score ≥ 9 points
2. Cognitive impairment caused by other reasons, taking drugs, poisoning, etc;
3. Suffer from severe musculoskeletal system diseases and other contraindications to exercise and are not suitable for Tai Chi training, such as those who suffer from stroke, Parkinson's disease, and have a history of lower limb arthritis, hip and knee joint replacement, etc;
4. Patients with severe heart, liver, kidney failure, malignant tumors, and other major diseases;
5. Individuals with visual/auditory impairments, writing/reading impairments, illiteracy, etc. that affect training and evaluation;
6. Individuals with uncontrolled hypertension (systolic blood pressure greater than 160mmHg or diastolic blood pressure greater than 100mmHg after medication);
7. Participating in other experiments that influence this study. -

This cross-sectional real-world external validation study will enroll 150 community-dwelling older adults aged 60 years and above, who are diagnosed with Mild Cognitive Impairment (MCI) and have regular practice of standardized 24-form simplified Tai Chi for at least 6 months (consistent with the form used in the parent RCTs). Eligible participants will be recruited from communities in Fuzhou, China, to validate the generalizability and clinical applicability of a pre-developed machine learning prediction model for Tai Chi cognitive benefit population identification.