Chinese scholars develop a new dementia risk prediction model that can predict the onset of dementia ten years in advance

　　China News Service, Shanghai, September 26th (Sun Guogen and Chen Jing) Chinese experts carried out multi-disciplinary joint research, using biomedical big data and artificial intelligence algorithms to develop a new dementia risk prediction model, and named it UKB-DRP.

By entering the relevant information of the individual to be tested on the relevant page, the public can obtain their five-year, ten-year and longer-term dementia risk.

　　This model is a general model that can make prospective intelligent prediction of the risk of all-cause dementia and its main subtype (Alzheimer's disease) at the same time. Precise prediction of onset.

It is reported that there are two main types of dementia: Alzheimer's disease and vascular dementia.

　　The UKB-DRP dementia prediction model has high predictive power for all-cause dementia and Alzheimer's disease in the next five years, ten years or even longer. The predicted AUC value of the disease was higher, reaching 0.86-0.89.

It is understood that the value range of AUC is between 0.5 and 1.

The closer the AUC is to 1.0, the higher the authenticity of the detection method.

　　The research team further evaluated the risk calibration of the UKB-DRP model for predicting dementia, and the model predicted new dementia events and observed dementia events with high consistency.

The research team also compared the UKB-DRP dementia prediction model with internationally published prediction models, and the results showed that the prediction accuracy of the UKB-DRP dementia prediction model was significantly better.

　　The research results carried out by Professor Yu Jintai's clinical research team from Huashan Hospital affiliated to Fudan University and Professor Feng Jianfeng/Cheng Wei's team of young researchers from Fudan University's Brain-inspired Intelligence Science and Technology Research Institute have been published in the latest issue of The Lancet "Electronic Clinical Medicine" ” (“EClinical Medicine”).

　　Yu Jintai said that with the prolongation of human lifespan and the increasing aging of society, the number of dementia patients is increasing year by year. There are currently a large number of dementia patients in China, which brings a heavy care and economic burden to society and families.

The expert pointed out that the course of dementia is long, and pathological changes will occur 20 years before the onset of the disease, and the optimal treatment time window is often missed at the time of diagnosis. Therefore, it is urgent to find an early prediction method for dementia and identify high-risk groups of dementia. Effective prevention and intervention should be carried out in the early stage of the disease, thereby delaying the progression of the disease and reducing the disease burden.

　　According to reports, the clinical research team used the UK Biobank cohort to follow up 425,159 non-demented people aged 40-69 years. During the median follow-up of 11.9 years, 5,287 participants were diagnosed with new-onset dementia.

Multi-dimensional health-related indicators such as cognition, biochemistry, behavior, and genes of the participating population were included in the study, and these indicators were strictly screened and quality-controlled based on clinical experience.

　　Subsequently, the researchers used machine learning algorithms to calculate the importance of each indicator to the dementia prediction model, and finally selected the top ten indicators as dementia predictors, and constructed the UKB-DRP dementia prediction model.

The ten predictors were: age, genes, duration of cognitive paired tests, percent leg fat, number of medications taken, duration of cognitive response tests, peak expiratory flow, maternal age at death, chronic disease, and mean red blood cell volume.

　　Yu Jintai told reporters that the advantage of the UKB-DRP dementia prediction model is that its ten predictors can be quickly obtained from questionnaires, simple physical examinations and routine blood tests.

Compared with other models based on complex predictors such as expensive whole-genome sequencing, invasive lumbar puncture or PET imaging, the UKB-DRP dementia prediction model can be widely used in early screening in medical institutions at all levels.

　　In order to facilitate the one-click operation of the public, the research team has developed a web application of the UKB-DRP dementia prediction model (https://jiayou0907.shinyapps.io/UKB-DRP-Tool/). By measuring information about individuals, it is possible to obtain their five-, ten-, and longer-term dementia risk.