AI can explore the relationship between gum health and the risk of developing Alzheimer's, a rather innovative idea that is based on research suggesting connections between oral health and cognitive health. To develop an AI model to investigate this potential correlation, we will need data that includes several variables related to both gum health and cognitive and general health risk factors.
In the field of medical research, early detection of neurodegenerative diseases such as Alzheimer's is crucial for improving patient prognoses and quality of life. Recently, the correlation between periodontal health and the risk of developing Alzheimer's has been explored, suggesting that inflammatory markers present in gum diseases could be linked to the onset of this brain pathology.
In this context, an artificial intelligence (AI) model has been developed with the objective of predicting the likelihood of an individual developing Alzheimer's within five years, based on various indicators of oral health and other risk factors. This model was constructed using a synthetic dataset, specifically designed to simulate the relevant characteristics of an at-risk population.
The dataset includes variables such as age, gender, plaque index, periodontal pocket depth, number of teeth lost, medical history of diabetes and hypertension, educational level, among others. A Random Forest algorithm was used to train the model, given its effectiveness in handling multiple input variables and its ability to model complex interactions between them.
The training process revealed that the plaque index, initial cognitive score, and periodontal pocket depth are the variables that most influence the prediction of Alzheimer's development.
These results not only validate the hypothesis that periodontal health could be an early indicator of Alzheimer's but also provide a basis for future research that could focus on preventive interventions in oral health to reduce the risk of developing this disease.
Despite the significant advancements represented by this study, it is crucial to mention that the model has limitations, especially in its ability to correctly detect positive cases, which may be attributed to the class imbalance in the dataset. Future improvements in the model could include class balancing techniques, the incorporation of more clinical data, and the validation of the model with real data to confirm its applicability and efficacy in clinical settings.
This study not only underscores the importance of technological innovation in preventive medicine but also highlights how artificial intelligence can transform the management and early diagnosis of complex health conditions, potentially improving interventions and long-term clinical outcomes.
According to the training and reporting carried out, the model can correctly identify individuals who will not develop Alzheimer's in 84% of cases, but has difficulty identifying those who will develop the disease (the positive class). This may be due to the imbalance in the dataset classes.
The variables that most influence the prediction, based on their relative importance in the model, are:
Plaque Index: This suggests that the condition of plaque in the gums has a strong correlation with the risk of developing Alzheimer's.
Initial Cognitive Score: A lower score could indicate a higher risk.
Periodontal Pocket Depth: Indicates the severity of periodontal disease, which also appears to be strongly related.
Age: Well-known risk factor for Alzheimer's.
Number of teeth lost: Possibly related to general health and oral hygiene.
These results can help understand which factors are most predictive according to the model and could guide future research or adjustments to the model to improve its ability to correctly predict positive cases.
Disclaimer: This is only a demo for this hackathon, it does not currently offer any results, services or medical consultations.
Top comments (0)