Developing A Multi-Disease Risk Prediction Model Using Machine Learning

Praveen Kumar, Babita Yadav, Sachin Kumar Gupta, Bhoomi Gupta

2025

Closed access Metadata (CSL-JSON)
DOI 10.1109/aist68591.2025.11441609
Data confidence 70 medium
How this score is calculated

It is built from checkable properties of the record itself: whether the article has a DOI, how many independent sources confirmed its data, whether the required fields are filled in, and how recent the latest confirmation is. Disagreement between sources and a missing identifier lower it. A value of 80 or above means the record was published without further checking; 60 to 79 means it is published at reduced confidence, with a manual check advisable. A record verified by an editor by hand is never pushed below 90 by a recalculation. This scores our data about the article, not the article itself.

Abstract

The ability to predict the risk of multiple diseases accurately is a crucial step toward improving early diagnosis and healthcare outcomes. With the potential to improve and automate the diagnostic procedure, machine learning methods have become more effective instruments for this aim in recent years. This study focuses on developing a machine learning model to predict the risk of different diseases by using Transfer Learning, where pre-trained models are used to enhance the prediction accuracy. A comparison was done of three models, namely, Transfer Learning (proposed model), VGG-16, and DenseNet-121, in the multi-disease risk prediction context. The proposed Transfer Learning model reached a maximum accuracy of 98.3%, which is higher than the VGG-16 at 94% and DenseNet-121 at 92% in key evaluation metrics, such as precision, recall, and F1-score. This demonstrated how Transfer Learning could produce more balanced and accurate predictions of a variety of diseases, including influenza, asthma, diabetes, heart attacks, and many others. The findings provide evidence that Transfer Learning can indeed be applied to real-world health care scenarios, and it thus offers a very promising tool in the early prediction of disease. Future work will include fine-tuning the model, dealing with class imbalances in health datasets, and ensemble learning to enhance the model even further. Ultimately, this may then allow for an efficient deployment of machine learning models in clinical settings to improve disease diagnosis and manage risks.