The escalating global prevalence of dementia, driven by demographic shifts toward an aging popula- tion, accentuates the imperative for accurate early identification of cognitive illnesses[32].Detecting these conditions before overt symptoms manifest is crucial, given that irreversible and profound brain impairment may already be underway by the time symptoms become apparent[8]. Timely identifi- cation is vital to empower at-risk individuals for proactive interventions, potentially slowing disease progression and reducing life-threatening consequences. Machine learning (ML) emerges as a valuable tool, enhancing precision and predictive capabilities through extensive datasets and computational efficiency[33]. Presently, Alzheimer’s disease identification relies heavily on clinical assessments[7], aided by technologies like MRI, PET, and biomarker analysis in cerebrospinal fluid, with genetic markers providing additional diagnostic utility.
Our study introduces a novel early prediction tool, leveraging readily available, cost-effective information for identifying older individuals at risk of Alzheimer’s disease without overt cognitive impairment. It is crucial to acknowledge that different classifiers possess unique advantages and disad- vantages, contributing to diverse outcomes in testing. The performance of ML algorithms, specifically RF and XGBoost, surpasses conventional techniques. Both RF and XGBoost, as integrated learning algorithms, outperform traditional methods, aligning with previous research highlighting the efficacy of integrated methodologies in Alzheimer’s disease categorization[34, 35].
The Random Forest algorithm employs a bagging approach known as Bootstrap aggregation, selecting training samples to independently train multiple decision trees. These trees are then amal- gamated to create the integrated RF model. In contrast, XGBoost utilizes a gradient boosting technique, iteratively training decision trees based on residuals from preceding trees. While RF deter- mines the ideal partition points for feature values through concepts like entropy, Gini index, and standard deviation, XGBoost introduces a regularization term, ”Regularization,” to manage tree complexity. This term considers feature weights during node splitting, enhancing optimal split fea- ture selection. XGBoost leverages second-order derivative information to estimate the loss function, improving feature selection precision and overall model accuracy and generalization capabilities.
Compared to prior research, our study observed comparable sensitivity (0.739) and AUC (0.78) values in XGBoost’s prediction model. Satisfactory specificity (0.677) and accuracy (0.691) were achieved, though a lower precision (0.403) might be attributed to dataset imbalances. Careful charac- teristic selection during model development[36] is crucial, as the observed AUC values falling within the 0.7 to 0.8 range may be indicative of a relationship with feature selection. The findings of a study[37] shown that the model’s AUC was 0.829, which was somewhat lower than the AUC of 0.836 achieved when just using imaging inputs. Furthermore, when including both imaging inputs and patient data inputs, the AUC decreased to 0.754. Conversely, when exclusively considering quantita- tive inputs and patient data, the AUC increased back to 0. 836. This study utilized manual feature selection, potentially excluding variables strongly associated with AD. The predictive classifier shows promising screening capabilities, but caution is warranted in evaluating its success due to its lower precision.
Our study further explores the significance of individual characteristics, highlighting the robust impact of education level on Alzheimer’s disease prediction. Higher education correlates with pre- served white matter microstructure, a potential protective factor against cognitive decline[38, 39]. Past medical history appears to have a smaller impact, emphasizing the influence of demographic variables, mental health, and lifestyle factors on prediction accuracy. Variables such as age, gender, and marital status emerge as significant predictors. Research findings indicate that advanced demen- tia is mostly influenced by age in comparison to other contributing variables[40]. Regarding gender, studies have consistently demonstrated that women not only display an increased vulnerability to Alzheimer’s disease but also experience a more accelerated decline in cognitive function compared to males at equivalent stages of the illness. This observed gender disparity is potentially linked to the endocrine activity of female hormones[41, 42]. However, it’s important to highlight that this differ- ence between sexes may diminish in more advanced age cohorts, particularly among women who have reached the postmenopausal stage[43].Nevertheless, despite potential reductions in the disparity, a higher incidence of AD persists among women compared to males. Anxiety, an understudied aspect, surfaces as a noteworthy predictor of Alzheimer’s disease, echoing previous findings[44]. Marital sta- tus also plays a role, with lifelong singles and the bereaved exhibiting a higher dementia risk[45]. In summary, this research posits that the significance of education level persists as a crucial determi- nant, particularly among women with lower educational attainment. Hence, emphasizing educational enrichment emerges as a beneficial strategy for averting dementia and cognitive decline, particularly given the observed significance of educational levels, especially among women with lower educational attainment. Additionally, placing emphasis on the preservation of optimal sleep quality and mental well-being stands as another valuable avenue in reducing the likelihood of developing dementia.
utilizing models based on easily accessible information presents a more convenient and practical approach for mass screening and early identification when compared to invasive genetic and biochem- ical tests. this simplified method not only proves to be cost-effective but also optimizes healthcare efficiency and resource allocation, offering affordable risk assessment tools tailored to diverse popu- lations. our model streamlines risk assessment, adding early problem identification and interventions for families. healthcare professionals efficiently refer high-risk patients to specialized care, enhancing older adults’ access to dementia care, improving their quality of life, and preventing disease progres- sion. additionally, it serves as an educational tool for early identification of high-risk older individuals, benefiting families pivotal in identifying those susceptible to AD.
However, it is crucial to acknowledge the limitations of this research. Constraints in time and resources prevented the construction of diverse models using various methods, potentially limiting a comprehensive exploration of early risk prediction. The chosen characteristics for this investigation were not exhaustive, emphasizing the need for further research to delve deeper into previous findings. To enhance the precision, applicability, and validation of future models, expanding the research cohort by increasing participant diversity and including various geographical areas is imperative. Addressing these limitations in subsequent research endeavors will contribute to refining the risk prediction model and advancing our understanding of Alzheimer’s disease risk factors.