Dietary intervention optimized using machine learning could lower risk of dementia

The term dementia is used to describe various debilitating neurological disorders characterized by a progressive loss of memory and a decline in mental abilities. Estimates suggest that over 55 million people worldwide are currently living with dementia, and this number could rise further over the next decades.

Since effective treatments for dementia remain limited, medical researchers and neuroscientists have been trying to identify factors associated with its emergence. This in turn informed the development of interventions aimed at preventing dementia or lowering the risk of its appearance.

Researchers at Fudan University, Zhejiang University School of Medicine and other institutes recently tried to use machine learning to devise a more effective dietary intervention that could reduce the risk of experiencing dementia in the later stages of life. The newly developed intervention, called MODERN (Machine-learning-assisted Optimizing Dietary intERvention against demeNtia risk), was introduced in a paper in Nature Human Behavior.

“Our study was motivated by the urgent need for effective dementia prevention strategies, particularly in light of limited treatment options currently available,” Prof. Jintai Yu, the corresponding author of the paper, told Medical Xpress.

“While dietary modification represents a highly accessible and modifiable risk factor, existing dietary interventions have often failed to demonstrate significant benefits in randomized controlled trials (RCTs). Our primary objective was to develop a novel, data-driven dietary intervention specifically designed to reduce dementia risk by combining large-scale cohort data with advanced machine learning techniques.”

Using machine learning and data from the large biomedical database known as the UK Biobank, Prof. Yu and his colleagues sought to identify optimal food combinations that could be linked with a reduced risk of dementia and improved brain health. The researchers specifically used a machine learning technique called LightGBM, which was trained on dietary and health-related data collected from 185,012 people in the UK.

“Among the algorithms we tested—including XGBoost and Random Forest—LightGBM demonstrated superior performance, achieving the highest area under the ROC curve (AUC),” explained Prof. Jia You, co-first author of the paper.

“By analyzing the predictive importance of different food groups, the model identified key dietary factors associated with dementia risk. These findings were then translated into a practical dietary scoring system (MODERN), which emphasizes moderate intake of brain-healthy foods (e.g., leafy greens, berries) and restriction of detrimental items like sweetened beverages.”

MODERN, a new dietary intervention developed by Prof. Yu and his colleagues using machine learning, was found to be more strongly associated with a lower risk of developing dementia than other dietary patterns previously linked with improved brain health. In particular, it was found to have more protective effects than MIND (Mediterranean Intervention for Neurodegenerative Delay), a well-established dietary program for the prevention of neurodegeneration.

“Across three independent external validation cohorts, participants in the highest MODERN diet score group exhibited a 36% lower risk of dementia than those in the lowest group,” said Dr. Sijia Chen, the first author of the paper. “Mechanistic analyses revealed plausible neuroprotective pathways, including enhanced brain structural integrity and attenuated neuroinflammation.”

The brain protective effects of the new dietary intervention devised by Prof. Yu and his colleagues could soon be assessed further in real-world settings. Eventually, it could be introduced in public health guidelines and integrated with existing dementia prevention efforts, potentially contributing to a reduction of neurodegenerative diseases worldwide.

“Moving forward, we aim to validate the MODERN diet across diverse populations and conduct randomized controlled trials (RCTs) to establish its causal benefits for dementia prevention,” added Prof. Yu.

“Beyond this, we plan to apply similar data-driven approaches to identify optimal dietary patterns for other brain disorders, such as anxiety and depression. Ultimately, our long-term goal is to develop a unified, evidence-based dietary framework specifically designed to promote brain health and prevent neurological diseases.”

© 2025 Science X Network

The term dementia is used to describe various debilitating neurological disorders characterized by a progressive loss of memory and a decline in mental abilities. Estimates suggest that over 55 million people worldwide are currently living with dementia, and this number could rise further over the next decades.

Since effective treatments for dementia remain limited, medical researchers and neuroscientists have been trying to identify factors associated with its emergence. This in turn informed the development of interventions aimed at preventing dementia or lowering the risk of its appearance.

Researchers at Fudan University, Zhejiang University School of Medicine and other institutes recently tried to use machine learning to devise a more effective dietary intervention that could reduce the risk of experiencing dementia in the later stages of life. The newly developed intervention, called MODERN (Machine-learning-assisted Optimizing Dietary intERvention against demeNtia risk), was introduced in a paper in Nature Human Behavior.

“Our study was motivated by the urgent need for effective dementia prevention strategies, particularly in light of limited treatment options currently available,” Prof. Jintai Yu, the corresponding author of the paper, told Medical Xpress.

“While dietary modification represents a highly accessible and modifiable risk factor, existing dietary interventions have often failed to demonstrate significant benefits in randomized controlled trials (RCTs). Our primary objective was to develop a novel, data-driven dietary intervention specifically designed to reduce dementia risk by combining large-scale cohort data with advanced machine learning techniques.”

Using machine learning and data from the large biomedical database known as the UK Biobank, Prof. Yu and his colleagues sought to identify optimal food combinations that could be linked with a reduced risk of dementia and improved brain health. The researchers specifically used a machine learning technique called LightGBM, which was trained on dietary and health-related data collected from 185,012 people in the UK.

“Among the algorithms we tested—including XGBoost and Random Forest—LightGBM demonstrated superior performance, achieving the highest area under the ROC curve (AUC),” explained Prof. Jia You, co-first author of the paper.

“By analyzing the predictive importance of different food groups, the model identified key dietary factors associated with dementia risk. These findings were then translated into a practical dietary scoring system (MODERN), which emphasizes moderate intake of brain-healthy foods (e.g., leafy greens, berries) and restriction of detrimental items like sweetened beverages.”

MODERN, a new dietary intervention developed by Prof. Yu and his colleagues using machine learning, was found to be more strongly associated with a lower risk of developing dementia than other dietary patterns previously linked with improved brain health. In particular, it was found to have more protective effects than MIND (Mediterranean Intervention for Neurodegenerative Delay), a well-established dietary program for the prevention of neurodegeneration.

“Across three independent external validation cohorts, participants in the highest MODERN diet score group exhibited a 36% lower risk of dementia than those in the lowest group,” said Dr. Sijia Chen, the first author of the paper. “Mechanistic analyses revealed plausible neuroprotective pathways, including enhanced brain structural integrity and attenuated neuroinflammation.”

The brain protective effects of the new dietary intervention devised by Prof. Yu and his colleagues could soon be assessed further in real-world settings. Eventually, it could be introduced in public health guidelines and integrated with existing dementia prevention efforts, potentially contributing to a reduction of neurodegenerative diseases worldwide.

“Moving forward, we aim to validate the MODERN diet across diverse populations and conduct randomized controlled trials (RCTs) to establish its causal benefits for dementia prevention,” added Prof. Yu.

“Beyond this, we plan to apply similar data-driven approaches to identify optimal dietary patterns for other brain disorders, such as anxiety and depression. Ultimately, our long-term goal is to develop a unified, evidence-based dietary framework specifically designed to promote brain health and prevent neurological diseases.”

© 2025 Science X Network