Non-small cell lung cancer (NSCLC) accounts for approximately 85 % of all lung cancer cases and remains a leading cause of cancer-related mortality worldwide (Sung et al., 2021). Despite advancements in diagnostic and therapeutic interventions, the survival rate for NSCLC remains low, particularly in advanced-stage cases (Le et al., 2025).

Limitations of current standard‑of‑care. Current diagnostic and treatment approaches rely on radiological imaging, histopathology, and molecular profiling; however, these conventional methods may exhibit interobserver variability, delayed diagnoses, and suboptimal treatment personalization (Gandhi et al., 2023). Interobserver variability in manual computed tomography (CT) nodule measurement and in downstream classification can alter management recommendations; semi‑automated and volumetric approaches improve agreement but are not universally available in practice (Gierada et al., 2020, Zacharias and Svahn, 2024). The TNM staging system, while indispensable, is an imperfect prognosticator at the individual level because it cannot fully capture biological heterogeneity; even recent refinements aim primarily to improve anatomic granularity (Rami-Porta et al., 2024). Biomarker assessment, notably PD‑L1 immunohistochemistry, shows laboratory variability and only modest predictive value in many settings, limiting confident patient selection for immunotherapy. Diagnostic pathway delays and fragmented data flows can defer definitive treatment and impede personalized decisions (Vranic and Gatalica, 2023). Collectively, these limitations motivate computational tools that can standardize measurement, integrate multimodal data, and support individualized prognostication and treatment selection.

AI opportunities across the NSCLC pathway. Given these challenges, exploring innovative solutions such as Artificial Intelligence (AI), deep learning using Artificial Neural Networks (ANNs), and Machine Learning (ML) have emerged as promising tools to improve diagnostic accuracy, predict treatment outcomes, and enhance overall survival (OS) (Bekbolatova et al., 2024). AI has shown remarkable progress in lung cancer through automated image analysis, biomarker identification, and survival prediction models (Rabby et al., 2025). AI-driven radiomics, deep learning, and predictive modelling have been applied to diagnostic imaging and histopathological slides, aiding in early detection, tumour characterization, and personalized treatment paradigms (Le et al., 2025, Gandhi et al., 2023). Furthermore, AI-based biomarker identification was crucial in advancing precision medicine, allowing for more accurate prognosis and targeted therapy selection (Rabby et al., 2025). Additionally, AI-integrated decision support systems have shown potential in optimizing chemotherapy regimens, guiding radiotherapy planning, and predicting patient responses to immunotherapy and targeted treatments (Gandhi et al., 2023). Studies suggest that AI-driven models, such as DeepSurv, may outperform conventional statistical models in predicting NSCLC outcomes and optimizing treatment strategies (Le et al., 2025). Moreover, cost-effectiveness analyses suggest that AI may maximize resource allocation, potentially reducing healthcare expenditures by improving diagnostic efficiency and minimizing ineffective treatments (Bekbolatova et al., 2024).

However, despite these advancements, integrating AI into NSCLC clinical workflows remains an ongoing investigation. Concerns regarding data heterogeneity, algorithm transparency, ethical considerations, and real-world applicability pose significant challenges to the widespread adoption of these models (Bekbolatova et al., 2024). While AI has improved diagnostic accuracy and treatment stratification, its impact on patient-reported outcome measures (PROM), OS, and cost-effectiveness remains underexplored (Bekbolatova et al., 2024).

What prior reviews have covered - and what remains missing. Prior syntheses have generally centred on diagnostic or prognostic performance. For example, a recent systematic review and meta‑analysis evaluated AI‑based imaging models to predict lymph‑node metastasis in NSCLC (Chen et al., 2024); others examined radiomics‑based prognostic models (Kothari et al., 2021) or compared ML against conventional statistical approaches for survival prediction in lung cancer (Didier et al., 2024). These reviews characterize algorithmic discrimination and accuracy but do not provide a comprehensive, NSCLC‑specific synthesis of whether AI‑enabled approaches improve PROM, OS when used to guide care, or cost‑effectiveness. Moreover, prospective evaluations of post‑diagnostic AI across oncology remain few, and formal health‑economic evidence for AI in lung‑cancer care is still emerging (Geppert et al., 2025, Wenderott et al., 2024).

Explicit knowledge gap and objective of this review. To our knowledge, there is no prior systematic review that concurrently synthesizes AI’s impact on PROM, OS, and cost‑effectiveness in adult patients with histologically confirmed NSCLC, comparing it to conventional research methodologies and standard‑of‑care approaches. Addressing this gap, this systematic review evaluates whether AI‑driven models confer patient‑centred and economic benefits beyond traditional approaches, and summarizes methodological challenges and implementation considerations for integrating AI into real‑world NSCLC care.