Early Prediction of Breast Cancer Using a Multilayer Neural Network with Adam Optimization

Abstract

Breast cancer remains one of the most prevalent malignancies among women worldwide and a major public-health concern, with prognosis and survival strongly dependent on the stage at which the disease is detected. This paper presents a deep-learning-based decision-support model for the early prediction of breast cancer using a multilayer feed-forward neural network trained with the backpropagation algorithm. The proposed architecture consists of an input layer, three hidden layers, and a sigmoid output layer, optimized using the Adam adaptive learning-rate algorithm and trained via the binary cross-entropy loss function. The model takes as input a set of quantitative bio-molecular pathway concentration features (AKT, FASL, MAPK, NOTCH, SHH, TNF, WNT, and MTOR-linked markers) and outputs a binary prediction of malignancy status. The network was implemented in Python using Keras with a TensorFlow backend, with Rectified Linear Unit (ReLU) activations in the hidden layers and He/Xavier-style weight initialization to stabilize training. Experimental evaluation on a held-out test split demonstrates that the backpropagation-based model, combined with Adam optimization, converges reliably and produces consistent classification performance, outperforming conventional shallow classifiers reported in earlier studies. The results suggest that such a neural-network-based decision-support system can assist clinicians in achieving faster, more consistent, and cost-effective early diagnosis, ultimately contributing to improved patient outcomes.

Introduction

The study presents a deep neural network (DNN)-based approach for early prediction of breast cancer malignancy using quantitative bio-molecular pathway markers. Breast cancer remains a major global health problem, and early identification of malignant cases is essential for improving treatment outcomes and survival. Although conventional diagnostic methods such as mammography, biopsy, and clinical interpretation are effective, they can be resource-intensive and may be affected by inter-observer variability. Therefore, machine-learning-based computer-aided diagnosis systems can provide valuable decision-support capabilities.

The proposed system uses eight cancer-related molecular pathway markers—AKT, FASL, MAPK, NOTCH, SHH, TNF, WNT, and an mTOR-linked marker—along with a pathway identifier to predict whether a sample is malignant or non-malignant. The categorical pathway and outcome variables are numerically encoded, and the dataset is divided using an 80:20 stratified training/testing split.

Proposed Model

A fully connected feed-forward neural network is developed using the following architecture:

The network is trained using backpropagation and the Adam optimizer with binary cross-entropy loss. Training is performed for 100 epochs with a batch size of 20. He-style Gaussian weight initialization is used to improve training stability.

Key Methodological Features

• Input: Molecular pathway and biomarker measurements.
• Preprocessing: Missing/malformed data checking, removal of unnecessary columns, categorical encoding, and optional feature standardization.
• Feature learning: Three hidden ReLU layers automatically learn nonlinear relationships among pathway markers.
• Classification: A sigmoid output produces the probability of malignancy.
• Optimization: Adam optimizer provides adaptive learning rates for efficient convergence.
• Data division: Stratified 80% training and 20% testing split.
• Prediction: The final system classifies cases as malignant or non-malignant.

Literature Gap

Previous studies have explored neural networks, SVM, KNN, decision trees, ensemble models, fuzzy systems, feature-selection techniques, and genomic data for breast cancer prediction. However, the proposed work focuses specifically on deep neural-network modeling of quantitative molecular pathway markers. The study attempts to address the limitations of simpler classifiers by learning complex nonlinear interactions among high-dimensional biological features.

Significance

The proposed model can serve as a computer-aided decision-support system for breast cancer assessment. By learning relationships among molecular pathway markers, it may help identify patterns associated with malignancy and potentially support earlier clinical assessment.

However, the model should be considered supportive rather than diagnostic. Its clinical usefulness would require validation on larger, independent, and clinically representative datasets, along with appropriate evaluation of sensitivity, specificity, calibration, and external generalization.

Conclusion

This paper presented a multilayer, feed-forward neural network for the early prediction of breast cancer from bio-molecular pathway marker data, trained using backpropagation with the Adam optimizer. The proposed architecture — an input layer, three ReLU-activated hidden layers, and a sigmoid output layer — is designed to model the non-linear relationships between pathway-level biomarkers and malignancy status more effectively than shallow classifiers used in earlier work. A decision-support system built on this model can help clinicians make faster, more consistent, and more cost-effective early-stage diagnostic decisions, ultimately supporting improved patient outcomes and reduced treatment costs. Future work will focus on: (i) expanding the dataset with additional clinical and genomic features to improve generalizability; (ii) conducting rigorous k-fold cross-validation and reporting a full suite of classification metrics (accuracy, precision, recall, F1-score, ROC-AUC, and confusion matrix); (iii) benchmarking the proposed architecture against ensemble methods (e.g., random forests, gradient-boosted trees) and other deep architectures; and (iv) exploring explainability techniques (e.g., SHAP, LIME) to help clinicians interpret individual predictions, which is essential for building clinical trust in AI-assisted diagnostic tools.

References

[1] Baselga, J.; Norton, L. “Focus on breast cancer.” Cancer Cell, 1, 319–322, 2002. [2] Benson, J. R.; Jatoi, I.; Keisch, M.; Esteva, F. J.; Makris, A.; Jordan, V. C. “Early breast cancer.” Lancet, 373, 1463–1479, 2009. [3] Ries, L. A. G. Et al. “SEER Cancer Statistics Review.” National Cancer Institute, 2008. [4] Kok, S. H.; Abdullah, A.; Jhanjhi, N. Z.; Supramaniam, M. “A review of intrusion detection system using machine learning approach.” International Journal of Engineering Research & Technology, 12, 8–15, 2019. [5] Diwaker, C.; Tomar, P.; Solanki, A.; Nayyar, A.; Jhanjhi, N. Z.; Abdullah, A.; Supramaniam, M. “A new model for predicting component-based software reliability using soft computing.” IEEE Access, 7, 147191–147203, 2019. [6] Kingma, D. P.; Ba, J. “Adam: A method for stochastic optimization.” 3rd International Conference on Learning Representations (ICLR), 2015. [7] Cheng, H. D.; Chen, C. H.; Freimanis, R. I. “A neural network for breast cancer detection using fuzzy entropy approach.” Proceedings of the IEEE Conference, 141–144, 1995. [8] Rabuñal, J. R.; Cevdet, M. ?. “An expert system for detection of breast cancer based on association rules and neural network.” Expert Systems with Applications, 36, 3465–3469, 2009. [9] Floyd, C. E.; Lo, J. Y.; Yun, A. J.; Sullivan, D. C.; Kornguth, P. J. “Prediction of breast cancer malignancy using an artificial neural network.” Cancer, 74, 2944–2948, 1994. [10] Li, Y. “Performance evaluation of machine learning methods for breast cancer prediction.” Applied and Computational Mathematics, 7, 212, 2018. [11] Sivapriya, J.; Kumar, A.; Sai, S. S.; Sriram, S. “Breast cancer prediction using machine learning.” International Journal of Recent Technology and Engineering (IJRTE), 8, 4879–4881, 2019. [12] Fatima, N.; Liu, L.; Hong, S.; Ahmed, H. “Prediction of breast cancer, comparative review of machine learning techniques, and their analysis.” IEEE Access, 8, 150360–150376, 2020. [13] Sakri, S. B.; Abdul Rashid, N. B.; Muhammad Zain, Z. “Particle swarm optimization feature selection for breast cancer recurrence prediction.” IEEE Access, 6, 29637–29647, 2018. [14] Daemen, A.; Griffith, O. L.; Heiser, L. M.; Wang, N. J.; Enache, O. M.; Sanborn, Z.; Pepin, F.; Durinck, S.; Korkola, J. E.; Griffith, M.; et al. “Modeling precision treatment of breast cancer.” Genome Biology, 14, R110, 2013. [15] Mavaddat, N.; Michailidou, K.; Dennis, J.; Lush, M.; Fachal, L.; Lee, A.; Tyrer, J. P.; Chen, T. H.; Wang, Q.; Bolla, M. K.; et al. “Polygenic risk scores for prediction of breast cancer and breast cancer subtypes.” American Journal of Human Genetics, 104, 21–34, 2019. [16] Botlagunta, M.; Botlagunta, M. D.; Myneni, M. B.; Lakshmi, D.; Nayyar, A.; Gullapalli, J. S.; Shah, M. A. “Classification and diagnostic prediction of breast cancer metastasis on clinical data using machine learning algorithms.” Scientific Reports, 13, 485, 2023. [17] Khan, F.; Khan, M. A.; Abbas, S.; Athar, A.; Siddiqui, S. Y.; Khan, A. H.; Saeed, M. A.; Hussain, M. “Cloud-based breast cancer prediction empowered with soft computing approaches.” Journal of Healthcare Engineering, 2020, 8017496, 2020. [18] [18] Rawal, R. “Breast cancer prediction using machine learning.” 2020. Available online: http://www.jetir.org (accessed 10 February 2025). [19] Kumar, K. T. K. “Discern breast cancer vetting Jupyter notebook.” International Journal for Scientific Research & Development, 8(2), 2020. Available: https://www.ijsrd.com/articles/IJSRDV8I20831.pdf [20] Kumar, K. T. K. “Breast cancer prediction using ML.” International Journal of Novel Research & Development, 10(6), 2025. Available: https://ijnrd.org/papers/IJNRD2506018.pdf

Copyright

Copyright © 2026 K. Tulasi Krishna Kumar, Pasupuleti Monica Martin. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.