Pinar Ozel

Background: Clinician-facing artificial intelligence-based clinical decision support systems (AI-CDSSs) are increasingly evaluated with real-world data, but evidence on implementation, human factors, responsible AI, and patient benefit remains unclear. The objective of this research is to map applications, technical characteristics, evaluation approaches, integration, outcomes, and implementation gaps of clinician-facing AI-CDSSs in real-world health care settings. Methods: PubMed/MEDLINE was searched on 15 June 2026 using the same strategy for each publication year from 2020 to 2026, changing only the date range. After duplicate removal and rule-based prescreening, available PMC XML and outside-PMC reports were evaluated against clinical, patient-oriented, practical, and translational criteria. AI-assisted screening and charting received targeted human review and TA200/FT150/DX120 quality control; full-corpus dual-human screening was not performed. Results: Of 14,097 canonical annual records, 14,083 unique records were screened; 5140 reports were sought and 3767 were considered for eligibility. The review included 1181 reports, mapping to 1180 candidate report families in the minimal duplicate-report check. Final stages were clinical validation, 741 (62.7%); prospective/user evaluation, 270 (22.9%); and implementation/impact evaluation, 170 (14.4%). Separate internal, external, multicenter and temporal validation extraction signals were detected in 707 (59.9%), 441 (37.3%), 312 (26.4%) and 27 (2.3%) reports, respectively. Safety, fairness/equity, explainability/interpretability and economic/resource evaluation signals were detected in 98 (8.3%), 61 (5.2%), 155 (13.1%) and 182 (15.4%) reports, respectively. Conclusions: Evidence remains dominated by validation rather than prospective, implementation, and impact studies. More comprehensive evaluation and reporting of temporal validation, technical integration, safety, fairness, explainability, economics, and post-deployment governance across settings, populations, devices, and clinical workflows are needed to demonstrate accountable clinical value beyond technical functionality.

Background/Objectives: This research investigates brain connectivity patterns in reaction to social and non-social stimuli within a virtual reality environment, emphasizing their impact on cognitive functions, specifically working memory. Methods: Employing the LEiDA framework with EEG data from 47 participants, I examined dynamic brain network states elicited by social avatars compared to non-social stick cues during a VR memory task. Through the integration of LEiDA with deep learning and graph theory analyses, unique connectivity patterns associated with cue type were discerned, underscoring the substantial influence of social cues on cognitive processes. LEiDA, conventionally utilized with fMRI, was creatively employed in EEG to detect swift alterations in brain network states, offering insights into cognitive processing dynamics. Results: The findings indicate distinct neural states for social and non-social cues; notably, social cues correlated with a unique brain state characterized by increased connectivity within self-referential and memory-processing networks, implying greater cognitive engagement. Moreover, deep learning attained approximately 99% accuracy in differentiating cue contexts, highlighting the efficacy of prominent eigenvectors from LEiDA in EEG analysis. Analysis of graph theory also uncovered structural network disparities, signifying enhanced integration in contexts involving social cues. Conclusions: This multi-method approach elucidates the dynamic influence of social cues on brain connectivity and cognition, establishing a basis for VR-based cognitive rehabilitation and immersive learning, wherein social signals may significantly enhance cognitive function.

The aim of the present study was to investigate brain volumes in pessimist and optimist participants. Therefore, in the present voxel-based morphometry research, it is investigated whether optimism has a corresponding counterpart in the structure of the brain. Thirty-two participants were screened via a publicly available dataset to test for this. The participants are divided into two groups: low optimists and high optimists, each with sixteen people. In comparison, a significant difference (P

This study aims to conduct a systematic review of studies on neuroimaging measurements used in multimedia learning research. The particular aim of the review is to explore how cognitive processes in multimedia learning are studied with relevant variables through neuroimaging technology. Studies that met the inclusive criteria were selected and analyzed with the data entry tool. Forty articles were reviewed based on the research questions about the research characteristics, the type of learning environments, the variables, the types of cognitive load, the other cognitive load measurements, the types of neuroimaging measures, the techniques that should be known in the field of neuroimaging to study cognitive load in multimedia learning. The results revealed that most of the studies preferred using both subjective and other objective measures to assess cognitive load in addition to neuroimaging measures. The studies examined learning outcomes, cognitive processes, and some other variables besides measuring cognitive load. The most striking observation to emerge from the analysis is that Electroencephalography, Functional Magnetic Resonance Imaging, Functional Near-Infrared Spectroscopy, and Transcranial Doppler Ultrasonography have been found as the most preferred neuroimaging tools utilized in multimedia learning research. Research results were interpreted, and several gaps in research relating to multimedia learning were identified.

Empirical Mode Decomposition (EMD) provides an adaptive signal processing tool, and its multivariate extension is useful to model multichannel signals. Recently, EMD and multivariate EMD have successfully been applied to solve different signal processing problems. Electroencephalogram signals are often employed to explore the emotional concepts for human-machine interaction. In this paper, an emotion recognition model is presented via EEG signal decomposition by utilizing multivariate EMD. Intrinsic Mode Functions extracted by the multivariate EMD algorithm are quasi-orthogonal. Hence the Gram-Schmidt Orthogonalization method is applied to the extracted IMFs. The number of orthogonal components reveals the number of modes used in the second step of the proposed method, where the Empirical Wavelet Transform is used to explore different features of the IMFs. By applying Ensemble and Decision Tree classifiers on the calculated features, the emotional states are classified as high-low arousal, valence, and dominance with 72.7%, 62.0%, and 64.7% highest classification performances using the selected channels, respectively.

The aim of this study is to compare subjective and objective cognitive load measurements in a multimedia learning environment. For this purpose, 20 university students studied in multimedia environments designed by researchers during which eye movements and multichannel electroencephalography (EEG) signals were recorded. Self-report ratings were obtained at the end of the experiment, and retention performances of the students were measured. After the data were collected, Pearson Correlation analysis was applied. According to the results, significant relationship between the number of fixations and EEG frequency band powers was found. In addition, there was a negative relationship between retention performance and number of fixations. Moreover, a negative relationship was found between retention performance and self-reported measurements.

Electrophysiological data processing can take place both in time and in frequency domains as well as in the joint time-frequency domain. Short Time Fourier Transform and Wavelet Transform are commonly used time-frequency analysis methods. The limitations of these methods initiated the use of methods such as synchrosqueezing and multivariate synchrosqueezing methods. In our proposed method 88.9%, 77.8%, 80.6% accuracy rates were obtained respectively for the valence, activation and dominance parameters using and multivariate synchrosqueezing methods and support vector machines(SVM) which yields better results than most of the other methods mentioned in the literature.

In the multimedia learning environment, presenting texts in different formats at the same time revealed the redundancy effect. This study aimed to evaluate cognitive load by recording brain signals and eye movements while texts were presented in different formats. For this purpose, two different multimedia learning environments in which printed text and narration were presented together and solo narration were designed, respectively. An experimental study was conducted by dividing 20 of 40 participants into two different multimedia environments. This study ensured the main perspectives of building up a learning profile structure depending on learners' cognitive profile by providing two different multimedia designs for redundancy effect via participants' brain topographies and eye movements. The findings suggested that the redundancy group demonstrated elevated cognitive load, especially in the frontal and parietal regions, as seen by heightened theta, beta, and gamma wave activity. Conversely, the non-redundancy group exhibited enhanced processing efficiency with less cognitive strain.