We propose a machine-learning approach to construct reduced-order models (ROMs) to predict the long-term out-of-sample dynamics of brain activity (and in general, high-dimensional time series), focusing mainly on task-dependent high-dimensional fMRI time series. Our approach is a three stage one. First, we exploit manifold learning and, in particular, diffusion maps (DMs) to discover a set of variables that parametrize the latent space on which the emergent high-dimensional fMRI time series evolve. Then, we construct ROMs on the embedded manifold via two techniques: Feedforward Neural Networks (FNNs) and the Koopman operator. Finally, for predicting the out-of-sample long-term dynamics of brain activity in the ambient fMRI space, we solve the pre-image problem, i.e., the construction of a map from the low-dimensional manifold to the original high-dimensional (ambient) space by coupling DMs with Geometric Harmonics (GH) when using FNNs and the Koopman modes per se. For our illustrations, we have assessed the performance of the two proposed schemes using two benchmark fMRI time series: (i) a simplistic five-dimensional model of stochastic discrete-time equations used just for a “transparent” illustration of the approach, thus knowing a priori what one expects to get, and (ii) a real fMRI dataset with recordings during a visuomotor task. We show that the proposed Koopman operator approach provides, for any practical purposes, equivalent results to the FNN-GH approach, thus bypassing the need to train a non-linear map and to use GH to extrapolate predictions in the ambient space; one can use instead the low-frequency truncation of the DMs function space of L2-integrable functions to predict the entire list of coordinate functions in the ambient space and to solve the pre-image problem.

Early detection of colorectal cancer is crucial for improving outcomes and reducing mortality. While there is strong evidence of effectiveness, currently adopted screening methods present several shortcomings which negatively impact the detection of early stage carcinogenesis, including low uptake due to patient discomfort. As a result, developing novel, non-invasive alternatives is an important research priority. Recent advancements in the field of breathomics, the study of breath composition and analysis, have paved the way for new avenues for non-invasive cancer detection and effective monitoring. Harnessing the utility of Volatile Organic Compounds in exhaled breath, breathomics has the potential to disrupt colorectal cancer screening practices. Our goal is to outline key research efforts in this area focusing on machine learning methods used for the analysis of breathomics data, highlight challenges involved in artificial intelligence application in this context, and suggest possible future directions which are currently considered within the framework of the European project ONCOSCREEN.

Background: There is an emerging need for evidence-based approaches harnessing large amounts of health care data and novel technologies (such as artificial intelligence) to optimize public health policy making. Objective: The aim of this review was to explore the data analytics tools designed specifically for policy making in noncommunicable diseases (NCDs) and their implementation. Methods: A scoping review was conducted after searching the PubMed and IEEE databases for articles published in the last 10 years. Results: Nine articles that presented 7 data analytics tools designed to inform policy making for NCDs were reviewed. The tools incorporated descriptive and predictive analytics. Some tools were designed to include recommendations for decision support, but no pilot studies applying prescriptive analytics have been published. The tools were piloted with various conditions, with cancer being the least studied condition. Implementation of the tools included use cases, pilots, or evaluation workshops that involved policy makers. However, our findings demonstrate very limited real-world use of analytics by policy makers, which is in line with previous studies. Conclusions: Despite the availability of tools designed for different purposes and conditions, data analytics is not widely used to support policy making for NCDs. However, the review demonstrates the value and potential use of data analytics to support policy making. Based on the findings, we make suggestions for researchers developing digital tools to support public health policy making. The findings will also serve as input for the European Union–funded research project ONCODIR developing a policy analytics dashboard for the prevention of colorectal cancer as part of an integrated platform.

The principal objective served by this article is to identify key literature and provide an overview of the breadth of research in the field of machine learning applications on exposomics data with a focus on cardiovascular diseases. Secondarily, this study aimed at identifying common limitations and meaningful directives to be addressed in the future. Most of the identified literature focuses on Disease Understanding, Prevention, and Management and the remaining on healthcare resource planning and management. Linear, non-linear, and ensemble machine learning methods have been applied, with tree-based methods generally being the most popular, and Random Forests specifically reported as being the best performer most of the time. Non-traditional CVD risk factors spanning behavioural/ lifestyle, socio-economic and a wide range of environmental factors have been investigated in the identified literature. According to our findings, the most reported category is the environmental risk factors which are often considered alone rather than in conjunction with the others. Applications of machine Learning have been a substantial accelerating factor on the field, enabling the analysis of high dimensional data, improving accuracy, investigating novel risk factors and expanding our knowledge on the impact of exposome on cardiovascular diseases. However, several challenges persist, including data heterogeneity and the lack of standardized protocols, which introduces bias and hinders reproducibility and comparability across studies, as well as concerns about the validity of inference when applying machine learning methods to identify associations between exposure and health outcomes. Addressing these issues requires collaborative and interdisciplinary effort to integrate standardized exposome data frameworks akin to those of other –omics fields, apply causal inference methods to validate findings, and further expand the use of Explainable Artificial Intelligence to build insights and enable comparability and understanding.

BACKGROUND There is an emerging need for evidence-based approaches harnessing large amounts of health care data and novel technologies (such as artificial intelligence) to optimize public health policy making. OBJECTIVE The aim of this review was to explore the data analytics tools designed specifically for policy making in noncommunicable diseases (NCDs) and their implementation. METHODS A scoping review was conducted after searching the PubMed and IEEE databases for articles published in the last 10 years. RESULTS Nine articles that presented 7 data analytics tools designed to inform policy making for NCDs were reviewed. The tools incorporated descriptive and predictive analytics. Some tools were designed to include recommendations for decision support, but no pilot studies applying prescriptive analytics have been published. The tools were piloted with various conditions, with cancer being the least studied condition. Implementation of the tools included use cases, pilots, or evaluation workshops that involved policy makers. However, our findings demonstrate very limited real-world use of analytics by policy makers, which is in line with previous studies. CONCLUSIONS Despite the availability of tools designed for different purposes and conditions, data analytics is not widely used to support policy making for NCDs. However, the review demonstrates the value and potential use of data analytics to support policy making. Based on the findings, we make suggestions for researchers developing digital tools to support public health policy making. The findings will also serve as input for the European Union–funded research project ONCODIR developing a policy analytics dashboard for the prevention of colorectal cancer as part of an integrated platform.

Ovarian cancer (OC) is the most lethal gynecological malignancy, largely because non-specific symptoms contribute to advanced-stage diagnosis and poor prognosis, underscoring the need for improved non-invasive biomarkers. Circulating microRNAs (c-miRNAs) have emerged as promising candidates for OC detection, risk stratification, prognosis, and treatment monitoring. This systematic review, prospectively registered in PROSPERO (CRD420251230898), evaluated the diagnostic and prognostic performance of c-miRNAs and miRNA-based multivariate signatures. MEDLINE/PubMed, Scopus, and CENTRAL were searched from inception to 30 May 2026. Eligible studies assessed c-miRNAs in serum, plasma, whole blood, or extracellular vesicle fractions from women with confirmed or suspected OC including studies using prediagnostic samples from women subsequently diagnosed with OC. Records were screened, data extracted, and risk of bias evaluated according to the type of result using QUADAS-2 for diagnostic accuracy, QUIPS for prognostic-factor associations, and PROBAST for multivariable prediction models. Of 697 records identified, 91 publications met the inclusion criteria. Most studies used case–control or observational designs and analysed serum or plasma samples. Single c-miRNAs generally showed moderate diagnostic accuracy, whereas multivariate miRNA panels and multimodal models incorporating CA-125 or HE4 generally reported higher AUC values, including in some early-stage cohorts. Frequently reported candidates included members of the miR-200 family, miR-21, miR-1246, miR-125b, miR-145, miR-205, and miR-130a. Prognostic studies associated c-miRNAs with survival, recurrence, treatment response, and chemoresistance, although findings were more heterogeneous than diagnostic findings. Risk-of-bias concerns were common, particularly for patient selection in diagnostic studies, confounding in prognostic-factor analyses, and statistical analysis in prediction-model studies. Overall, c-miRNAs represent promising non-invasive biomarkers for OC, particularly when integrated into multivariate or multimodal models. However, methodological heterogeneity, small cohort sizes, limited external validation, and inconsistent reporting currently restrict clinical translation. Future studies should prioritize standardized pre-analytical and analytical workflows, adequately powered prospective cohorts, and independent multicenter validation.