Cause adverse outcomes, particularly when erroneous treatment recommendations lead to adverse clinical outcomes

The gap

Evidence profile

Sourced from the future work and recommendations and limitations of the source papers, classified as general, drawn from work published between 2025 and 2026, spanning 3 journals. Those papers have been cited 1 times in total.

Supporting evidence — 4 representative gaps

• Application of artificial intelligence in paediatric oncology imaging (2026) · Pediatric Radiology · doi

  Radiologists, as well as everyone involved with the use of AI in a clinical setting, require familiarity with AI systems to leverage them safely and effectively in clinical practice. AI excels at processing large datasets and handling repeti- tive tasks, while radiologists provide judgment, context, and adaptability for complex cases, as well as the essential task of data curation. Combining these strengths can lead to more accurate, efficient, and reliable medical imaging practices, accelerating the translation of AI models into clinical prac- tice [12, 13, 83, 90]. Overcoming the limitations of paediatric oncology data The challenges of AI development in paediatric oncology imaging stem from limited data availability and the time- consuming, labour-intensive process of producing anno- tations, problems exacerbated by the need for datasets diverse enough to represent the entire paediatric age spec- trum. These challenges can be addressed in several ways: increasing the availability of data, applying (AI) methods that generate synthetic data or learn effectively from limited examples, or combining both strategies in a complementary manner. A primary approach to expanding data availability involves multi-institutional collaboration, which can help create datasets that are more diverse across age, demo- graphics, and clinical characteristics. These initiatives rely on standardised protocols and ethical agreements to ensure privacy, reproducibility, and harmonisation [12]. Further- more, documenting datasets with “datasheets” that detail their composition, collection, and intended uses is critical for responsible implementation [93]. Several dedicated initiatives and platforms now support sharing of paediatric imaging data. Examples of open data- sets and online platforms offering downloadable data across a range of disease types, including oncology, are provided in Table 2 and Table 3. These resources can represent a sub- stantial source of labelled and unlabelled data for diverse AI training purposes. Text-based initiatives like the Pediatric Cancer Data Commons [97], which integrate clinical data across tumour types, are particularly valuable. Their impact could be strengthened by coupling them with dedicated imaging databases, creating larger and more diverse datasets for robust AI development. Such paediatric resources are urgently needed, given that large, dedicated imaging datasets have already become indispensable tools for advancing AI research in adult oncology [64, 89, 100]. Federated infrastructures, furthermore, enable collabo- rative AI development across institutions without sharing sensitive patient data, preserving privacy while leveraging heterogeneous, decentralised datasets. This decentralised approach allows models to be trained locally within each participating institution, with only aggregated model updates exchanged across sites [9, 12]. Platforms like the Europe

  generalfuture work
  Keywords: datasets clinical imaging paediatric across oncology diverse development availability initiatives dedicated platforms radiologists well them
• The translational chasm in machine learning for triple-negative breast cancer: a quantitative landscape assessment (2026) · Frontiers in Oncology · doi

  Based on the aforementioned shortcomings and the key issues identified across the three levels, future actions can be divided into several phases based on urgency, with each phase corresponding to specific clinical benefits. To address the systemic translation barriers ranging from data homogeneity to validation gaps, this study proposes a phased, progressive action framework to bridge the gap between computa- tional outputs and clinical evidence. In the near term (1–2 years), priority should be given to utilizing public data platforms such as TCIA and TCGA to initiate multicenter external validation of retrospective models that have received high citations and are open-source. For example, the 2018 radiomics model developed by Saha A et al. based on 922 breast DCE-MRI images (29)or the 2020 IBSI radiomics standardization initiative by Zwanenburg A (56)can be used to obtain evidence of cross-institutional generaliz- ability at minimal cost. Concurrently, reporting standards such as TRIPOD-AI and MI-CLAIM must be elevated from submission recommendations to mandatory journal requirements, thereby rapidly narrowing the gap between method development and reporting quality by enhancing methodological reproducibility. In the medium term (2–5 years), we need to adopt the governance models of TCIA/TCGA to implement a federated learning frame- work, enabling cross-continental, multi-center model training and validation without transferring raw data, thereby addressing the systemic exclusion of African and Latin American countries from national collaborative networks (Figure 3). In the long term (5–10 years), randomized controlled trials comparing AI-assisted deci- sion-making with conventional decision-making, centered on patient endpoints (such as pCR rates and event-free survival), must be designed. Concurrently, efforts should be made to explore embedding AI models as secondary or exploratory endpoints within existing TNBC drug clinical trial frameworks. Establishing a foun- dation of trust through short-term external validation, expanding geographic and population coverage via mid-term data consortia, and confirming clinical utility through long-term prospective trials may help reduce the current 16:1 retrospective-validation gap and 44:1 prospective-trial gap and shift machine learning research in the TNBC field from being method-driven to truly clinical-driven. These three emerging research areas should also focus on different clinical endpoints. In the field of radiomics, the immediate goal is to conduct prospective pCR efficacy trials to demonstrate that AI predictions can influence treatment decisions; for the multi- omics approach, the near-term goal is to break down the barriers between imaging, pathology, and omics data, enabling models to evolve from predicting known biomarkers to discovering new mechanisms, ultimately yielding biomarker combinations that can guide immunotherapy stratification; and for the drug and spectros- copy approach, the near-term goal is to advance AI-screened lead compounds and Raman metabolic phenotypic features into Phase I or II clinical trials to definitively determine their efficacy in humans. If these three tiers of action can be advanced step by step, the role of machine learning in the field of TNBC will evolve from its current status as an auxiliary diagnostic tool to become the infrastructure for precision treatment decisions. Imagingomics will become the standard assessment method before and after neoadjuvant therapy, multi-omics integration will generate dynam- ically evolving molecular profiles for each patient, and AI-driven drug screening will shorten the timeline from target discovery to clinical validation. However, this vision hinges on validation keep- ing pace; Closing this gap requires that clinical validation efforts keep pace with methodological publication output. Beyond these immediate priorities, emerging technological directions warrant attention. Foundation models trained on spatial proteomics data are beginning to demonstrate zero-shot biomarker discovery in TNBC, with initial evidence of treatment response prediction outperforming established clinical stratification (125) Multimodal architectures that integrate imaging, pathology, and molecular data within unified frameworks are also advancing rapidly (126). These approaches remain retrospective and preclin- ical, but they represent the most likely pathways through which the current evidence gap may begin to narrow. 5 A comparison with previous bibliometric studies Previous bibliometric studies have largely focused on the overall intersection of breast cancer and artificial intelligence, but have rarely addressed triple-negative breast cancer, a highly heterogeneous subtype. An analysis by Wu et al. of 2,701 articles from the Web of

  generalfuture work
  Keywords: clinical validation term models evidence trials tnbc based three near years retrospective radiomics breast learning
• Artificial intelligence in primary aldosteronism: current achievements and future challenges (2025) · Frontiers in Molecular Biosciences · cited 1× · doi

  cause adverse outcomes, particularly when erroneous treatment recommendations lead to adverse clinical outcomes; (5) AI development necessitates large-scale clinical data acquisition, posing substantial privacy and ethical compliance challenges. Future research should prioritize algorithmic innovation and cross-disciplinary cooperation to standardize and facilitate the clinical translation of AI in PA management. Specifically, the following key directions should be prioritized: (1) optimizing model algorithm for primary care settings to enhance screening accessibility and initial diagnostic accuracy; integrating advanced multimodal technologies in tertiary centers to improve the precision of diagnosis and classification; (3) establishing cross-tier data systems to improve treatment algorithms and prognosis models, thereby building a comprehensive AI-assisted diagnostic framework. radiomics and genomics (2)

  generalrecommendations
  Keywords: clinical adverse outcomes treatment cross diagnostic improve cause particularly erroneous recommendations lead development necessitates large
• Applications of artificial intelligence and machine learning models in the prognosis and diagnosis of ovarian cancer (2026) · Frontiers in Oncology · doi

  Data remains the paramount and essential element for the education of AI systems. Utilizing contemporary information processing technologies to exploit radiology report databases may enhance report search and retrieval, thereby assisting radiologists in diagnosis. There is a necessity to advocate for the establishment of interconnected networks that identify patient data globally and facilitate large-scale AI training tailored to diverse patient demographics, geographic regions, and diseases. Furthermore, we underscore the necessity for more diversified imaging libraries for uncommon malignancies, including OC. However, much of the literature is limited by common methodological current weaknesses, including retrospective study designs, small cohort sizes, lack of prospective validation, spectrum bias, variable label quality, segmentation variability, scanner and protocol heterogeneity, risk of data leakage, and inconsistent reporting of model calibration and clinical utility. These limitations highlight the need for more rigorous and standardized study designs to improve the reliability and generalizability of AI applications in oncology. In image-based diagnostic tasks, AI models have demonstrated performance comparable to, or exceeding, that of expert physicians. However, such evaluations often fail to account for the multidimensional information routinely considered by radiologists when interpreting complex examinations. Non-imaging patient attributes, including demographic characteristics, clinical history, and genetic or molecular data, provide critical contextual information that is not fully captured by image-only models and can substantially enhance predictive performance when appropriately integrated (90). The high performance of contemporary AI models is frequently accompanied by substantial algorithmic complexity, involving high-dimensional feature spaces and deep neural network architectures. As a result, the internal decision-making processes underlying image-based predictions are often difficult to interpret or explain, a limitation commonly referred to as the “black-box” problem (91). This lack of transparency presents a major barrier to clinical trust, regulatory approval, and routine implementation in oncological practice. Explainable artificial intelligence (XAI) has emerged as a promising solution to address these limitations by providing model interpretability alongside predictive accuracy. XAI techniques aim to elucidate the relative importance of input features, highlight salient image regions, and identify clinically meaningful variables that drive model outputs (92). Laios et al. (93) demonstrated the clinical utility of XAI by developing ensemble AI models capable of predicting outcomes following cytoreductive surgery for OC, while simultaneously revealing patient- and procedure-specific factors contributing to surgical risk. Subsequent work further extended this framework to predict surgical effort requirements using human-centered and clinically interpretable variables. Despite these advances, many radiomicsbased tools and imaging biomarker models discussed in this review continue to function as black-box systems, limiting their reliability and interpretability in real-world clinical settings.

  generallimitationsevidence 5/5
  Keywords: clinical models image information patient imaging including model based performance systems contemporary report enhance radiologists

Questions about this gap