Early Prediction of Acute Lymphocyte Leukemia (ALL) Patients Using Convolutional Neural Network
Mansi Sharma, Amit Anil Gudadhe, Chetan Puri · 2025
Acute lymphoblastic leukaemia (ALL) remains a critical wellbeing concern, particularly among young individuals. acute lymphoblastic leukemia (ALL), the larger part of medicines utilized in conventional regimens were made more than 30 a long time prior. Since at that point, a number of novel drugs have been made and included to the treatment of ALL. To advance results for high-risk or backslid ALL, in any case, present day treatment procedures are still required. This consider proposes a novel deep learning approach that leverages various information modalities to improve early diagnosis and survival estimate for ALL. Our proposed model coordinating recurrent neural networks (RNNs) and convolutional neural systems (CNNs) to analyze diverse clinical and biological data. RNNs are utilized to capture temporal patterns in longitudinal patient information, such as laboratory results and treatment history. CNNs are utilized to extricate critical highlights from pictures of bone marrow biopsies and blood smears. By combining these effective deep learning strategies, our model objective to beat traditional procedures in terms of diagnostic accuracy and predictive capabilities. Through the integration of diverse data modalities, our model can learn complex associations and designs inside the data, leading to more exact and informative forecasts. The model's noteworthy 99% accuracy over 10 epochs illustrates its potential to revolutionize the management of ALL by giving clinicians with important insights for early mediation, personalized treatment plans, and improved patient results. our approach can help in hazard stratification by recognizing patients who are more likely to backslide or experience treatment disappointment. This data can direct treatment choices and improve patient care. Our study contributes to the advancement of accuracy pharmaceutical by empowering the advancement of more focused on and viable treatments for ALL. To moderate overfitting and improve generalization performance, we consolidated regularization techniques such as dropout and early halting. These techniques help prevent the model from memorizing the training data and enhance its capacity to form exact forecasts on unseen data.