Abstract:
Food allergies are a growing public health issue and their prevalence is on the increase worldwide as a result of complex, and interaction between genes, environment and lifestyle. Traditional diagnostic tools are still limited in scope and many times cannot discover the hidden biological mechanisms concealed allergic reaction. This research, Beyond the Bite: Unraveling the Mysteries of Food Allergy Through Epigenetics and Machine Learning, examines how factors such as epigenetic indicators and lifestyle elements can be combined with state-of-the-art computational models to better predict risk for allergy. A dataset of 1,830 actual responses were obtained from people using both online surveys and person to person interactions in order to have a wide range of demographics collected. The process involved preprocessing and cleaning the dataset, applying feature selection, training of multiple machine learning model Random Forest, XGBoost, Logistic Regration with validation by using accuracy, precision, recall, and confusion matrix. Of the models tested, Random Forest showed good performance, a good tradeoff between interpretability and predictive power. deep learning approaches show potential for scaling in the future. The results point to important links between diet, environmental exposures and allergy susceptibility and demonstrate the feasibility of data driven, multi-omic approaches to elicit hidden risk factors. These findings suggest that machine learning can complement traditional medical practices and can provide predictive information that can be useful in preventive healthcare, policy development, and public awareness campaigns. Ultimately, this research does play a role in 'bridging the gap between computational biology and clinical practice' with the potential for interdisciplinary approaches in the complex nature of health issue to be addressed.
