Prediction of Common Diseases Lurking in Smokers and Alcoholics and its Impact on their Health

Smoking and alcohol intake have become very common in modern society, leading to many complications and health problems, including respiratory illnesses and cardiovascular diseases. Early intervention and targeted healthcare interventions are very much needed to mitigate these risks and improve patient outcomes. This research addresses the need by using patient data to predict potential future health risks. By analysing 19 health and lifestyle factors, including smoking, alcohol consumption, diabetes, hypertension, haemoglobin and such factors, the study developed a predictive model to identify diseases that patients may currently have or develop in the near future. After extensive hyperparameter tuning and analysis, the study utilized a deep learning model optimized with the ReduceLROnPlateau technique, Adam as the optimizer and binary cross-entropy as the loss function. With the best parameters, including a learning rate of 0.001, 200 epochs, and a batch size of 64, the model achieved the highest accuracy of 94.69%. Additionally, to assess and compare the efficiency of the deep learning model, the study explored Random Forest algorithm due to its robustness in handling complex datasets. Initially focused on smokers and alcoholics, the study expanded to include non-smokers and non-alcohol consumers, demonstrating the model's broad applicability. These findings highlight the model's strong predictive capability and its potential for widespread use in preventive healthcare. Further, the current study uses dashboard visualization that reveals common disorders that affect smokers and drinkers, providing the valuable insights on targeted treatments to improve patient care and outcomes in population.