Dementia is a growing global public health challenge, yet early risk screening remains difficult in many community and primary care settings, particularly where trained professionals, cognitive testing and biomarkers are not readily available. A prospective cohort study published in General Psychiatry has developed and validated Cog-Free, a web-based tool that estimates dementia risk using routinely collected health examination data without requiring cognitive tests. The study presents a data-driven framework designed to make initial risk stratification more accessible and scalable in resource-limited settings.
Researchers used the Hubei Memory and Aging Cohort Study, which includes urban and rural older adults from 31 communities in Wuhan and 60 villages in Dawu County, China. The analysis included 2,962 dementia-free adults aged 65 years or older, followed from 2018 to 2024. From 548 candidate variables, the team selected 38 risk-associated variables using LASSO regression, then benchmarked eight machine-learning algorithms. The cohort was divided into a 70% training set (n=2,074) and a 30% internal validation set (n=888), and the optimal model was externally validated in 6,932 participants from the Chinese Longitudinal Healthy Longevity Survey.
Logistic regression provided the best balance of discrimination, calibration and clinical utility. In internal validation, Cog-Free achieved an area under the receiver operating characteristic curve of 0.86 (95% CI 0.82-0.89), with accuracy of 0.81, sensitivity of 0.78, specificity of 0.81 and a negative predictive value of 0.97. It significantly outperformed three established cognitive-testing-free tools: CogDrisk, ANU-ADRI and Modified-LIBRA (all p<0.001). Important variables included education, right-hand grip strength, age, cognitive activity, perceived memory decline, hearing acuity, nighttime awakenings and income satisfaction.
Cog-Free combines variables that can be collected through routine questionnaires and basic physical measurements, such as sociodemographic information, medical history, lifestyle, grip strength, gait and subjective memory concerns. The bilingual online calculator can generate a personalised estimate within about five minutes. The authors suggest that community health services could use the tool as a pre-screening and risk-stratification aid, directing people at higher estimated risk to further cognitive assessment and preventive support. It is not intended to diagnose dementia or replace professional evaluation.
The researchers caution that the mean follow-up was about 2.2 years, participants were all aged 65 years or older, and many variables were self-reported. Differences in available variables and dementia definitions between the development and external validation cohorts may also have influenced performance. Longer follow-up, validation in more diverse populations and simplified versions requiring fewer variables are needed. Nevertheless, the study demonstrates a transparent and reproducible approach for building low-cost dementia risk tools from routine health data, with potential relevance for community screening in low- and middle-income regions.
General Psychiatry
Observational study
Development and validation of a Cog-Free risk predicting tool for dementia in a community setting
1-Jun-2026