An Interpretable Machine Learning Framework for Non-Small Cell Lung Cancer Drug Response Analysis

Ann Rachel; Pranav M Pawar; Mithun Mukharjee; Raja M; Tojo Mathew

cs.CVMar 17, 2026

An Interpretable Machine Learning Framework for Non-Small Cell Lung Cancer Drug Response Analysis

Authors: Ann Rachel, Pranav M Pawar, Mithun Mukharjee, Raja M, Tojo Mathew

Organizations: Department of Computer Science, Birla Institute of Technology and Science Pilani, Dubai Campus, Dubai, UAE

Abstract

Lung cancer is a condition where there is abnormal growth of malignant cells that spread in an uncontrollable fashion in the lungs. Some common treatment strategies are surgery, chemotherapy, and radiation which aren't the best options due to the heterogeneous nature of cancer. In personalized medicine, treatments are tailored according to the individual's genetic information along with lifestyle aspects. In addition, AI-based deep learning methods can analyze large sets of data to find early signs of cancer, types of tumor, and prospects of treatment. The paper focuses on the development of personalized treatment plans using specific patient data focusing primarily on the genetic profile. Multi-Omics data from Genomics of Drug Sensitivity in Cancer have been used to build a predictive model along with machine learning techniques. The value of the target variable, LN-IC50, determines how sensitive or resistive a drug is. An XGBoost regressor is utilized to predict the drug response focusing on molecular and cellular features extracted from cancer datasets. Cross-validation and Randomized Search are performed for hyperparameter tuning to further optimize the model's predictive performance. For explanation purposes, SHAP (SHapley Additive exPlanations) was used. SHAP values measure each feature's impact on an individual prediction. Furthermore, interpreting feature relationships was performed using DeepSeek, a large language model trained to verify the biological validity of the features. Contextual explanations regarding the most important genes or pathways were provided by DeepSeek alongside the top SHAP value constituents, supporting the predictability of the model.

Figures & tables

Figure 1: AI/ML for Personalised Cancer Treatment

Reference	Contribution	ML/DL Method	Cancer Type/Focus	Result
[9] A. Hooshmand (2020)	Proposed a machine learning system to classify genomic profiles for accurate cancer diagnosis.	Naïve Bayes, SVM, Decision Trees, Random Forest, Logistic Regression, K-NN.	22 cancer types using 19,627 genes from WGS data.	Reported precision, sensitivity, and specificity of 1.0 for nearly all cancer types, indicating excellent classification performance.
[10] D. Oniani et al. (2021)	Developed a classification framework combining phenotypic and genetic features from EHRs using multiple GNN models.	Eight different Graph Neural Network architectures explored.	Multiple cancer types, with focus on phenotypic-genotypic interactions.	CNN achieved F1-score of 0.74; Logistic Regression scored 0.90.
[11] B. Bhinder et al. (2021)	Demonstrated AI’s role in risk assessment and personalized care by integrating diverse omics and clinical datasets.	Deep Neural Networks, Random Forest, DeepVariant.	CNS and skin cancers; includes methylation and mutation analyses.	No specific metrics reported; emphasized AI’s utility and lab-level validation.
[12] N. Zong et al. (2021)	Integrated genetic reports and EHRs to predict and classify both primary and unknown cancer types.	Node2vec embeddings, CNNs, Autoencoders.	Multiple cancers incl. prostate, breast, liver, pancreas, and thyroid.	Achieved 95% AUROC for primary classification and 80% for unknown primaries.
[14] S. D’Amico et al.(2022)	Investigated AI-generated synthetic data as a privacy-preserving solution for accelerating hematological cancer research.	Conditional GANs (cGANs) for synthetic clinical-genomic data generation.	Hematological malignancies: Myelodysplastic Syndromes, AML.	Used NNDR (Nearest Neighbor Distance Ratio) between 0.60-0.85 to balance fidelity and privacy.
[15] S. J. MacEachern and N. D. Forkert (2021)	Integrates different types of data, known as multi-modal or multi-omics data, to increase the level of precision medicine.	Deep learning algorithm, such as DeepSEA and DeepVariant, that is applied in genomic data and in estimating the functional effects of genetic variants.	The paper mentions the application of machine learning in various cancer types, not focusing on a certain type.	It highlights that verification of the models through independent test sets is an important factor to take care of, so the model does not do overfitting or underfitting of the data.

Table 1: Summary of studies applying ML/DL to cancer classification and prediction using heterogeneous data sources.

Figure 2: Histogram Showing the Distribution of Log-Transformed IC50 (LN-IC50) Across All Cancer Samples

Figure 3: Boxplot representation of LN-IC50 Distribution in LUAD vs LUSC

Figure 4: Proposed Predictive Model Architecture for Personalized Lung Cancer Drug Response Treatment

Model	MAE	MSE	R 2
XGBoost	0.0851	0.0249	0.9971
Random Forest	0.8228	1.1032	0.8700
Linear Regression	0.2268	0.1298	0.9847

Table 2: Regression Model Performance Comparison

Figure 5: Model Flowchart for Drug Response Prediction System

Figure 6: SHAP Waterfall Plot: Contribution of each Feature

Figure 7: Workflow diagram of the end-user application developed to make use of the proposed model

Metric	Previous Study [13]	Current Work (XGBoost + SHAP)
R 2 Score	0.9900	0.9971
MAE	0.1600	0.0851
MSE	0.0800	0.0249
RMSE	0.2900	0.1578

Table 3: Comparison of Model Performance Metrics

Figure 8: Model Performance with Varying Boosting Rounds

Metric	Fold 1	Fold 2	Fold 3	Fold 4	Fold 5	Average
R² Score	0.9979	0.9967	0.9944	0.9974	0.9963	0.9965
MAE	0.0746	0.0755	0.0884	0.0786	0.0798	0.0794

Table 4: Cross-Validation Performance Metrics

Explore similar work

Jul 1, 2026

Explainable AI for Cancer Drug Response Prediction: Beyond Univariate Feature Attributions

Predicting cancer drug response from transcriptomic profiles is a cornerstone of precision oncology, yet the scientific value of machine learning models hinges not solely on predictive accuracy, but also on their capacity to generate reliable biological insights. Current explainability approaches in this setting are computationally costly, lack robustness, and reduce complex drug response to univariate gene importance scores, overlooking the coordinated gene activity that drives sensitivity and resistance. In this work, we present ILLUME+, a scalable post-hoc explainability framework that moves beyond single-gene assessments to capture multiple, complementary forms of explanation. Integrated into our end-to-end pipeline, ILLUME+ produces more stable gene importance scores than existing baselines, recovers established drug-gene associations and mechanisms of action, and enables AI-assisted hypothesis generation to uncover novel interaction-driven molecular signals in cancer biology.

Martino Ciaperoni, Margherita Lalli, Simone Piaggesi +6

KDD Lab, Scuola Normale Superiore Pisa, Italy · KDD Lab, University of Pisa Pisa, Italy · EMBL-EBI Hinxton, United Kingdom +1

Cancer Genomics Feature Attribution Explainable Artificial Intelligence Clinical Prediction Oncology

Ask PDF

Mar 25, 2026

DeepDTF: Dual-Branch Transformer Fusion for Multi-Omics Anticancer Drug Response Prediction

Cancer drug response varies widely across tumors due to multi-layer molecular heterogeneity, motivating computational decision support for precision oncology. Despite recent progress in deep CDR models, robust alignment between high-dimensional multi-omics and chemically structured drugs remains challenging due to cross-modal misalignment and limited inductive bias. We present DeepDTF, an end-to-end dual-branch Transformer fusion framework for joint log(IC50) regression and drug sensitivity classification. The cell-line branch uses modality-specific encoders for multi-omics profiles with Transformer blocks to capture long-range dependencies, while the drug branch represents compounds as molecular graphs and encodes them with a GNN-Transformer to integrate local topology with global context. Omics and drug representations are fused by a Transformer-based module that models cross-modal interactions and mitigates feature misalignment. On public pharmacogenomic benchmarks under 5-fold cold-start cell-line evaluation, DeepDTF consistently outperforms strong baselines across omics settings, achieving up to RMSE=1.248, R^2=0.875, and AUC=0.987 with full multi-omics inputs, while reducing classification error (1-ACC) by 9.5%. Beyond accuracy, DeepDTF provides biologically grounded explanations via SHAP-based gene attributions and pathway enrichment with pre-ranked GSEA.

Yuhan Zhao, Jacob Tennant, James Yang +3

Department of Computer Science, North Carolina State University, Raleigh, NC, USA · UNC School of Medicine, University of North Carolina at Chapel Hill, Chapel Hill, NC, USA · Department of Molecular and Structural Biochemistry, North Carolina State University, Raleigh, NC, USA

Oncology Clinical Prediction Molecular Representation Learning

Ask PDF

Mar 16, 2026

Learning from Limited and Incomplete Data: A Multimodal Framework for Predicting Pathological Response in NSCLC

Major pathological response (pR) following neoadjuvant therapy is a clinically meaningful endpoint in non-small cell lung cancer, strongly associated with improved survival. However, accurate preoperative prediction of pR remains challenging, particularly in real-world clinical settings characterized by limited data availability and incomplete clinical profiles. In this study, we propose a multimodal deep learning framework designed to address these constraints by integrating foundation model-based CT feature extraction with a missing-aware architecture for clinical variables. This approach enables robust learning from small cohorts while explicitly modeling missing clinical information, without relying on conventional imputation strategies. A weighted fusion mechanism is employed to leverage the complementary contributions of imaging and clinical modalities, yielding a multimodal model that consistently outperforms both unimodal imaging and clinical baselines. These findings underscore the added value of integrating heterogeneous data sources and highlight the potential of multimodal, missing-aware systems to support pR prediction under realistic clinical conditions.

Alice Natalina Caragliano, Giulia Farina, Fatih Aksu +10

Unit of Artificial Intelligence and Computer Systems, Department of Engineering, Università Campus Bio-Medico di Roma, Rome, Italy · Fondazione Policlinico Universitario Campus Bio-Medico, Rome, Italy · Department of Medicine and Surgery, Università Campus Bio-Medico di Roma, Rome, Italy +1

Clinical Outcome Prediction Clinical Prediction Multimodal Learning Learning with Missing Data

Ask PDF