In a recent study conducted in Hong Kong, researchers have successfully developed algorithms to predict dementia among Chinese older adults by analyzing cognitive footprints in hospital records. By combining theory-driven and data-driven methods, the study aimed to identify key factors that contribute to the development of dementia in this demographic.
Using electronic medical records from the Clinical Data Analysis and Reporting System, the researchers analyzed data from patients diagnosed with dementia at 65 years or older between 2010 and 2018, as well as matched dementia-free controls. In total, 159,920 individuals were included in the study, with machine learning models such as LASSO, Multilayer perceptron (MLP), XGBoost, and LightGBM utilized to compare predictive accuracy with logistic regression models.
The findings of the study revealed that predictive accuracy significantly improved when considering all factors, as opposed to established modifiable factors only. Antipsychotics, education, antidepressants, head injury, and stroke were identified as the most important predictors in the total sample, while age-specific models highlighted cardiovascular and infectious diseases as prominent factors in older ages.
Overall, the developed algorithms showed satisfactory performance in identifying dementia among Chinese older adults. The researchers believe that these algorithms can be effectively utilized in clinical practice to assist in decision-making and provide timely interventions. The study was funded by the Research Grants Council of Hong Kong under the Early Career Scheme 27110519.
This innovative approach to dementia prediction highlights the potential for employing machine learning models in healthcare to enhance patient care and outcomes. Further research and implementation of these algorithms could lead to cost-effective interventions that benefit older adults at risk of developing dementia.
The study’s results emphasize the importance of considering a wide range of factors when predicting dementia and demonstrate the effectiveness of machine learning models in analyzing complex healthcare data. As the prevalence of dementia continues to rise globally, these advancements in predictive technology offer hope for improving early diagnosis and intervention strategies.
