搜索结果：Mistake

Background/Objectives: This study aimed to evaluate the effect of adherence to the DASH and Mediterranean diets on cognitive performance in adults. Methods: In this study, the Mediterranean Diet Adherence Screener (MEDAS), DASH Diet Quality Scale (DASH-Q), Oktem Verbal Memory Processes Test (Oktem-VMPT), and Trail Making Test (TMT) were administered face-to-face to adult individuals living in Afyonkarahisar, T&#xfc;rkiye, together with a form assessing sociodemographic characteristics and dietary habits. The collected data were analyzed using SPSS v27 software. Results: As participants' ages increased, DASH scores decreased (p < 0.05). As participants' BMI and waist/hip width increased, a decrease in DASH and MEDAS scores was observed (p < 0.05). As participants' ages increased, the IST-A, IST-B, and IST-Total scores increased (p < 0.05), but as their education level increased, the IST-A, IST-B, and IST-Total scores decreased (p < 0.05). As participants' education level increased, the total recall score on the Oktem-SBST scale tended to increase (p < 0.05). As participants' DASH scores increased, the "Immediate Memory" and "Spontaneous Recall" sub-components of the Oktem-SBST increased, while the "Learning Mistake Score," "USB Mistake Score," and "IST-A," "IST-B," and "IST Total" scores decreased (p < 0.05). As participants' MEDAS scores increased, the sub-components of Oktem-SBST, namely "Criteria Achievement," "Maximum Learning," "Spontaneous Recall," "Recognition," and "Total Recall," also increased, while the "Learning Mistake Score" decreased. (p < 0.05). Conclusions: Age, educational status, DASH, and MEDAS scores are associated with cognitive performance. The DASH and MEDAS diets have a positive impact on cognitive performance, highlighting the importance of healthy eating in public health strategies for maintaining cognitive health.

Objectives: Sleep-wake state discrepancy, the discrepancy between self-reported and objective sleep measures, is commonly experienced in poor sleep and insomnia. While perfectionism is implicated in insomnia, its relationship to sleep-wake state discrepancy has not been investigated. This study aimed to assess the association between sleep-wake state discrepancy and perfectionism and explore whether dysfunctional sleep beliefs and pre-sleep arousal mediate that relationship. Methods: Sixty adult participants from community and clinical populations were conveniently sampled (85% females, mean age 30.28 &#xb1; 11.13 years, 38% with insomnia symptoms). Sleep-wake state discrepancy measures were calculated using data from actigraphy and sleep diary collected over 14 days. The Frost Multidimensional Perfectionism Scale (FMPS), Hewitt-Flett Multidimensional Perfectionism Scale (HFMPS), Dysfunctional Beliefs about Sleep (DBAS), and Pre-sleep Arousal Scale (PSAS) were also collected. Results: High perfectionism levels were associated with high levels of sleep-wake state discrepancy. Concern over Mistakes and Doubts about Actions correlated with sleep onset latency discrepancy with small effects (r = 0.26 and 0.29, respectively). Doubts about Actions was associated with sleep onset latency discrepancy. Furthermore, pre-sleep arousal and cognitive pre-sleep arousal mediated relationships between sleep onset latency discrepancy and Concern over Mistakes and Doubts about Actions. Conclusions: Concern over Mistakes and Doubts about Actions relate to a poorer perception of sleep relative to objective sleep measures. During sleep onset, cognitive pre-sleep arousal appears to mediate relationships between perfectionism and sleep-wake state discrepancy. Therefore, perfectionism may be an important cognitive-emotional factor to consider when assessing and treating sleep-wake state discrepancy that commonly accompanies insomnia.

We investigated the impact of robot politeness and error-prone behavior on user perceptions through two user studies involving non-humanoid robots. Politeness was operationalized at two levels based on Lakoff's, (1973) politeness rules-one condition implemented all three of Lakoff's rules, demonstrating the highest level of politeness, while the other omitted them, resulting in a behavior that was strict but not impolite. The correctness was manipulated by comparing an error-free robot behavior to a behavior that included intentional errors. The studies were conducted using two different tasks with two robot types-a mobile robot and a manipulator robot-and involved 59 young adult participants (ages 24-28) with engineering backgrounds. Participants consistently rated the correct and polite robots most favorably. However, politeness did not offset the negative effects of erroneous behavior. In both studies, the robot that was correct but strict was rated more positively than the one that was polite yet made mistakes. These findings suggest that, at least in utilitarian task settings involving technically proficient users, politeness alone cannot compensate for performance failures. Moreover, a polite robot that makes errors may even frustrate users more than a straightforward but accurate one. The findings also emphasize the importance of evaluating HRI performance across different robot types and tasks, as these factors significantly shape user perceptions.

The integration of artificial intelligence in medical image classification for screening has the potential to enhance efficiency, diagnostic accuracy and accessibility. However, ethical concerns such as accountability, bias, transparency and the impact on healthcare professionals remain critical. This review synthesises qualitative evidence on the ethical considerations surrounding artificial intelligence adoption in screening programmes. A systematic search of qualitative studies, from June 2020 to September 2024, was conducted across multiple databases: MEDLINE, EMBASE, PsycInfo&#xae; (American Psychological Association, Washington, DC, USA) and Cumulative Index to Nursing and Allied Health Literature. Primary qualitative studies exploring healthcare professionals', patients' and other stakeholders' perspectives on artificial intelligence in screening were included. Thematic analysis was performed, and findings were assessed using the Grading of Recommendations Assessment, Development and Evaluation-Confidence in the Evidence from Reviews of Qualitative Research approach to evaluate confidence in the evidence. Fourteen qualitative studies were included, covering perspectives from clinicians, radiologists, artificial intelligence developers, policy-makers and patients. Key ethical concerns identified included: (1) the necessity of human oversight to ensure that artificial intelligences diagnostic recommendations are appropriate; (2) challenges in assigning liability when artificial intelligence errors occur; (3) risks of algorithmic bias due to discrepancies between training data sets and real-world populations; (4) concerns over data privacy, cybersecurity and informed consent in artificial intelligence-driven decision-making; (5) the need for transparency in artificial intelligence decision-making processes to build trust and (6) potential deskilling of healthcare professionals and shifts in professional responsibilities. While artificial intelligence was seen as a valuable tool to augment clinical decision-making, stakeholders emphasised that ethical frameworks must guide its implementation to maintain public trust and patient safety. This review highlights the critical considerations that must be addressed to ensure the responsible integration of artificial intelligence in medical screening. Policy-makers, healthcare institutions and developers should prioritise human oversight, robust regulatory frameworks and strategies to mitigate bias and ensure transparency. Future research should focus on disease-specific artificial intelligence applications and long-term ethical implications. The protocol for this study is registered on PROSPERO as CRD42024599536. This award was funded by the National Institute for Health and Care Research (NIHR) Evidence Synthesis programme (NIHR award ref: NIHR172233) and is published in full in Health Technology Assessment; Vol. 30, No. 51. See the NIHR Funding and Awards website for further award information. Research is exploring if artificial Intelligence could help doctors find cancer by looking at medical images like X-rays and scans. Artificial intelligence could spot tiny signs of cancer that people might miss. This could help detect cancer and other diseases earlier and more accurately, for example in breast cancer and diabetic eye screening. Artificial intelligence can also speed up the process, so patients get results faster. However, ethical questions arise with using artificial intelligence in this way. While there are not yet specific national or international guidelines for artificial intelligence in screening, general healthcare guidance highlights the following key issues: transparency: being clear about how artificial intelligence makes decisions fairness: ensuring artificial intelligence treats everyone equally and does not discriminate against certain groups accountability: making sure someone is responsible for artificial intelligence&#x2019;s actions reducing risks: ensuring artificial Intelligence systems are safe to use and do not cause harm governance and oversight: having strong systems in place to make sure artificial intelligence is used responsibly and ethically. This study examined ethical concerns of artificial intelligence in screening by reviewing research involving the general public, clinicians and patients. Initially focusing on diabetic retinopathy and breast cancer, it expanded to other conditions due to limited evidence. The study highlighted several ethical concerns raised in the literature, such as accountability for artificial intelligence mistakes, bias, data privacy, transparency and artificial intelligence&#x2019;s impact on doctors&#x2019; professional roles. In addition, people in the studies included in the literature expressed worries about related issues, particularly keeping humans in control of decisions, who is responsible when errors occur and whether artificial intelligence systems can be trusted to act fairly. Ethical challenges related to the implementation of artificial intelligence in clinical screening were also highlighted. These included healthcare inequality (with resource-limited hospitals potentially not benefiting equally), risks to patient safety from delays or errors in artificial intelligence-generated reports, the need for trust through rigorous testing and the importance of clear governance guidelines to ensure that artificial intelligence remains an assistive tool rather than replacing human judgement. This study provides useful information by identifying recurring ethical concerns that can inform the development of governance frameworks, guide safe implementation of artificial intelligence in screening and highlight priorities for future research and policy. Despite providing useful information, this study has some limitations due to incomplete research available. Future studies could focus on specific diseases and ethical issues, reassessing ethical considerations as new evidence becomes available.

The aim of this French physician initiated, multicentre, prospective, randomised trial with blinded criteria for re-intervention was to evaluate the efficacy of paclitaxel coated balloons (DCBs) in the treatment of arteriovenous fistula stenosis. One hundred and fifty patients with an arteriovenous fistula stenosis were included, and 145 were randomised 1:1 in the ABISS trial (Angioplastie au Ballon Impr&#xe9;gn&#xe9; de paclitaxel vs. angioplastie Standard pour le traitement des St&#xe9;noses sur fistule art&#xe9;rioveineuse; ClinicalTrials.gov identifier: NCT02753998) between a DCB (Lutonix, BD, Franklin Lakes, NJ, USA) and a placebo balloon. The main outcome was the rate of primary patency loss at 6 months. The primary analysis was performed in the modified intention to treat population. Missing outcome data were handled using a conservative imputation approach. The per-protocol analysis excluded patients who were included by mistake, not treated according to the randomisation arm, lost to follow up, or who missed a visit. In the modified intention to treat analysis, there was no significant difference between the groups regarding primary patency loss at 6 months (31% [22 of 70] in the DCB group vs. 42% [30 of 71] in the placebo group; p = .09; hazard ratio = 0.62 [95% confidence interval {CI} 0.36 - 1.07]). Primary patency loss was lower in the DCB than in the placebo group at 3 months (4.2% vs. 22.5%; p = .002; hazard ratio = 0.18 [95% CI 0.05 - 0.60]), but not at 12 months (58.6% vs. 57.7%; p = .57; hazard ratio = 0.88 [95% CI 0.57 - 1.35]). In the per-protocol analysis, the primary patency loss was significantly lower in the DCB group than in the placebo group (26% [16 of 61] in the DCB group vs. 50% [28 of 56] in the placebo group; p = .004). In the ABISS trial, the use of a DCB was not superior to a placebo balloon at 6 months in the modified intention to treat analysis.

The COVID-19 pandemic marked a significant turning point in public engagement with science, with conservatives exhibiting a more pronounced decline than liberals. Just as the pandemic intensified polarization, we reasoned that post-crisis reflection could offer an opportunity to repair these divides by shaping future public engagement with scientists. Drawing on boundary work theory and classic research on one-sided versus two-sided messaging, we contrasted two widely observed public health communication styles to assess the legitimacy of pandemic-related information. Across four experiment-based studies conducted in 2025 (N&#x202f;=&#x202f;1441), participants read about a scientist reflecting on lessons from COVID-19, with messaging framed as either authoritative (defending institutions, dismissing dissent, minimizing error, and selectively attributing misinformation) or accountability-embracing (acknowledging mistakes, engaging critics, and recognizing cross-ideological misinformation). Participants evaluated the scientist on: (a) immediate, attitude-based indicators of engagement, including generalized trust in scientists, prestige, and conspiracy beliefs, and (b) deliberate, actionable indicators, including funding support and willingness to provide the scientist with a public platform. Across studies, accountability-embracing approaches elicited greater engagement on most metrics, with especially clear benefits among political conservatives in several conditions. In polarized contexts following uncertainty-laden crises, reflexivity and accountability across ideological lines may offer a more promising path to restoring public engagement with science.

This study aimed to develop and validate a brief 5-item version of the Hypoglycemia Problem-Solving Scale (HPSS-5) and a single-item screening tool (HPSS-1) for the rapid assessment of hypoglycemia problem-solving ability in adults with type 1 diabetes. In this post-hoc analysis of the PR-IAH study, data from 233 adults with type 1 diabetes were analyzed to examine reliability and validity. Internal consistency, construct validity, and discriminative performance were evaluated using confirmatory factor analysis (CFA) with a robust estimator and receiver operating characteristic (ROC) analysis. Participants with high and low hypoglycemia problem-solving ability accounted for 17.2% and 27.5% of the sample, respectively. CFA of the five selected items demonstrated excellent model fit (&#x3c7;2(5)&#x2009;=&#x2009;3.74, p&#x2009;=&#x2009;0.588; CFI&#x2009;=&#x2009;1.000; TLI&#x2009;=&#x2009;1.001; RMSEA&#x2009;=&#x2009;0.000; SRMR&#x2009;=&#x2009;0.021). The HPSS-5 showed good internal consistency (&#x3c9;&#x2009;=&#x2009;0.80). ROC analysis indicated high discriminative accuracy, with areas under the curve (AUCs) of 0.969 (95% CI 0.949-0.989) and 0.986 (95% CI 0.976-0.997) for identifying high and low problem-solving abilities, respectively. The single-item screener (HPSS-1; "When my attempt to prevent hypoglycemia fails, I will analyze and identify my mistake") also showed good performance, with AUCs of 0.885 (95% CI 0.836-0.935) and 0.872 (95% CI 0.828-0.915), respectively. The HPSS-5 and HPSS-1 are psychometrically robust, time-efficient instruments for evaluating hypoglycemia problem-solving ability, suitable for use in both clinical practice and research. University hospital Medical Information Network (UMIN) Center: UMIN000039475), Approval date 13 February 2020. The online version contains supplementary material available at 10.1007/s13340-026-00911-8.

Patient advocates increasingly serve as voting members of Scientific Review Committees (SRCs) charged with evaluating the scientific merit and feasibility of clinical trial protocols. In principle, this gives advocates a seat at the table when the scientific architecture of a trial is most malleable. In practice, a recurring pattern undermines that promise. When advocates raise concerns about visit frequency, procedural invasiveness, or restrictive eligibility criteria, those concerns are often deflected on the grounds that they are ethical rather than scientific, and therefore properly the province of the Institutional Review Board. This Comment names that pattern jurisdictional dismissal and argues that it rests on a category error. Patient burden is a primary determinant of accrual, retention, data completeness, and generalizability. These are scientific concerns by any reasonable definition. When SRCs decline to engage with them, they let an unsuccessful trial proceed to ethics review with a fixed design that the IRB cannot meaningfully alter. I describe how jurisdictional dismissal operates, distinguish the scientific feasibility question that belongs to the SRC from the ethical acceptability question that belongs to the IRB, and propose four concrete changes to SRC practice that recognize feasibility as a scientific criterion in its own right. When patients and former research participants join clinical trial review committees, they bring firsthand experience of what it can be like to live through the demands of a trial. They can often tell whether weekly clinic visits are realistic for an older patient who no longer drives, whether three additional biopsies might deter people from enrolling, or which eligibility rules could quietly exclude the people most likely to benefit. No single advocate speaks for every patient, and lived experience is not the same as representative data. It is a vantage point that a committee made up only of clinicians and scientists can easily miss.But here is what often happens when a patient advocate raises one of these concerns at a Scientific Review Committee meeting: someone gently redirects the conversation. &#x201c;That sounds like an ethics question. The Institutional Review Board will look at that.&#x201d; The advocate&#x2019;s concern gets handed off, and by the time it reaches the ethics review, the trial design is already locked in.This article argues that this handoff is a mistake. Patient burden is not only an ethics question. It is also a scientific question, because trials that ask too much of patients fail to enroll enough people, lose participants partway through, and produce results that do not apply to the wider patient community. When committees treat burden as someone else&#x2019;s problem, they miss the chance to fix the trial when it is still possible to fix.I offer a name for this pattern: jurisdictional dismissal. I describe how it shows up in practice, why it weakens science, and what review committees can do about it. The fix is straightforward: treat patient burden as a scientific feasibility question, recognize advocates as scientific reviewers rather than ethics consultants, and weigh their concerns seriously, alongside concerns about statistical power or biological plausibility. The patient voice belongs in the room where the trial is designed, not only in the room where it is approved.Clinical trial number Not applicable.

The first aim of this study was to conduct a systematic and meta-analysis review considering both single-group interventions and controlled design to determine the effectiveness of interventions designed to reduce overall perfectionism, its two higher-order factors (Perfectionistic Strivings and Perfectionistic Concerns), and its dimensions measured separately. The second aim was to determine the relevance of moderators in terms of effectiveness. A systematic search process was performed in the Web of Science, Scopus, PsycINFO, and Psicodoc databases, using a common search strategy. A total of 33 studies were found. The interventions demonstrated clinical effectiveness with small to moderate effect sizes, except for the baseline-to-post-intervention measures of Perfectionistic Concerns, Socially prescribed perfectionism, and Personal standards. Scale type was found to be a significant moderator of Perfectionistic Concerns, Socially prescribed perfectionism, Perfectionistic Strivings, and Self-oriented perfectionism. The country was found to be significant for Perfectionistic Strivings and Self-oriented perfectionism. The treatment goal was only revealed to be significant for Concerns over mistakes.

More than two millennia ago, Plato's Allegory of the Cave illustrated how humans may mistake shadows for reality. Over the last decade, radiology has experienced a similar phenomenon with Artificial Intelligence (AI). Early narratives, often promoted by opinion leaders and commercial stakeholders, portrayed AI as an imminent replacement for radiologists, fostering unrealistic expectations, regulatory minimalism, and a climate of urgency. With increasing evidence and clinical experience, these shadows have gradually dissipated. AI has not supplanted radiologists but has demonstrated value when applied to well-defined tasks such as triage, workflow optimization, image quality enhancement, and quantitative analysis. At the same time, limitations have become evident, including hallucinations, lack of transparency, security risks, and the restricted clinical relevance of most marketed solutions. This decade of exaggerated promises has carried tangible costs, from damaged professional perception to adoption skepticism. As radiology enters the era of agentic AI, critical realism, transparency, and radiologist-led development are essential to avoid repeating past illusions.

The error-related negativity (ERN) is a measurable brain response to mistakes that is thought to reflect a modifiable cognitive/emotional bias contributing to the development of anxiety disorders. We previously demonstrated that a psychosocial intervention to reduce error sensitivity can reduce the ERN among (nonclinical) adults and children who have relatively large ERNs. We have also demonstrated that a brain response (balance N1) evoked by a disturbance to standing balance shares the ERN's relationship to anxiety. We hypothesized that if ERN and N1 reflect the same underlying brain mechanisms, then an intervention to reduce the ERN should similarly reduce the balance N1. In this pre-registered randomized controlled trial, 54 children with anxiety disorders (age 9-12&#xa0;years) were randomized into either a brief (45-min) single-session computerized psychosocial intervention to reduce error sensitivity, or a similarly formatted control condition. Primary outcome measures were changes in the ERN (measured in a Go/NoGo task) and balance N1 (measured in a lean-and-release balance task). The ERN was reduced after the psychosocial intervention, while the balance N1 remained unchanged. A brief computerized psychosocial intervention to reduce error sensitivity can reduce the ERN among clinically anxious children, but the limited effect may warrant a larger dosage. Discrepant outcomes between the ERN and balance N1 suggest the intervention targets mechanisms not shared between these brain responses. We speculate the intervention may have helped children manage overreactions to trivial mistakes while preserving the inherent significance of a loss of balance. CLINICAL TRIAL REGISTRATION: Computerized intervention targeting the error-related negativity and balance N1 in anxious children. https://clinicaltrials.gov/study/NCT05503017.

Wearable recorders are used in research and clinical practice to collect and measure children's vocalizations and the language environment in which they occur. Recordings generate vast amounts of audio, making manual analysis impractical and requiring automated processing. Two automated algorithms have emerged: the proprietary LENA (Language ENvironment Analysis) and the open-source ACLEW (Analyzing Child Language Experiences around the World) systems; yet, systematic performance comparisons remain scarce. Here, we validate and compare the performance of these two algorithms across key measures: audio segmentation into speaker categories, conversational turn count (CTC), adult word count (AWC), and child vocalization count (CVC). This analysis is based on 25 h of manually annotated audio recordings from 50 age-matched U.S. children with diverse neurodevelopmental profiles: children with Down syndrome, Fragile X syndrome, and Angelman syndrome, children at elevated likelihood of autism, and low-risk controls. We hypothesized that the algorithms might be less accurate for children with neurodevelopmental conditions, since these children often show different patterns of volubility and vocal maturity compared to the typically developing children used to train the algorithms. Thus, we assessed the performance of algorithms across diagnostic groups, a crucial validation step for both cross-population research and the evaluation of language interventions. Results reveal that while algorithms achieve similar performance across groups, they show different patterns: LENA makes fewer segmentation mistakes but misses many segments (identification error rate = 81.3%, percent correct = 45.3%), while ACLEW shows the opposite pattern (identification error rate = 129.4%, percent correct = 69.4%). Both LENA and ACLEW achieve reasonable levels of accuracy in their automatic counts (Pearson's r ranging from 0.78 to 0.92) and maintain stable performance across diagnostic groups. We conclude with recommendations for the validation and potential use of these algorithms in research and clinical practice. SUMMARY: Our comparison of the LENA and ACLEW algorithms in analyzing children's language environment and vocal production across five neurodevelopmental profiles reveals similar performance but distinct error patterns. LENA makes fewer errors but misses speech (81.3% error rate, 45.3% correct); ACLEW makes more errors but captures more speech (129.4% error rate, 69.4% correct). Both algorithms maintain consistent performance across diagnostic groups, supporting their reliability for research with these 2-year-old populations with diverse neurodevelopmental profiles. Variations in algorithm performance are primarily driven by the total speaking time of surrounding speakers (other children and adults), rather than the diagnostic group.

Broca's area is a region of the brain involved in the processing of verbal information, including memory encoding and retrieval. This study examined differences in the neural correlates of encoding and retrieval in Broca's area using word lists of varying lengths. Differences in the encoding and retrieval processes were assessed using word lists of different lengths: short (2-29), medium (30-59), and long (60-225) Latvian language nouns. In total, 23 participants completed the memory task. Each participant performed memory tasks with two short lists, two medium lists, and one long list, with varying list lengths within each diapason. We considered the activity of the F3 and F7 to represent our region of interest. We compared time-frequency (TF) data from encoding and retrieval across list lengths. The results revealed significant differences in TF plots for encoding and for correct and incorrect retrieval of information up to a list length of 50 words. Further increases in list length indicated greater similarity in brain functional patterns. For correct stimulus recognition or rejection of a distractor, the observed differences in TF depended on list length, and these differences were more pronounced under correct stimulus recognition. Encoding and correct answers during retrieval also showed statistically significant differences from incorrect answers (misses or false alerts) at different list lengths. Spectral power changes exhibited a nonlinear shape during both encoding and retrieval. The possible mechanisms differed during encoding and correct recognition versus mistakes.

Clinical skill anxiety (CSA) is a common psychological challenge among health science students during their clinical years, potentially impacting clinical competence and patient safety. While previous Ethiopian studies have highlighted high levels of stress and anxiety among students, evidence on context-specific CSA and its determinants among clinical-year students remains limited. This institution-based cross-sectional study was conducted between 20 August and 25 October 2025. A total of 426 clinical-year health science students across medicine, nursing, midwifery, and anaesthesia programs at Debre Tabor Comprehensive Specialized Hospital (DTCSH) were invited to participate in a census. The study assessed clinical skill anxiety (CSA) and associated characteristics using an adapted Clinical Skill Anxiety Scale (CSAS). This scale was informed by established instruments (STAI, BAI, and PSS), modified through a literature review and expert consultation, and demonstrated acceptable internal consistency in this study population (Cronbach's &#x3b1;&#x2009;=&#x2009;0.86). Predictors of CSA were identified using multivariable logistic regression (p&#x2009;<&#x2009;0.05; 95% CI). The overall prevalence of CSA among clinical-year health science students was 64.2% (95% CI: 59.5%-69.0%). Significant determinants included fear of making mistakes (AOR&#x2009;=&#x2009;4.2; 95% CI: 2.80-6.30), limited instructor feedback (AOR&#x2009;=&#x2009;3.6; 95% CI: 2.05-6.32), female sex (AOR&#x2009;=&#x2009;2.2; 95% CI: 1.62-2.98), lack of prior clinical-skills exposure (AOR&#x2009;=&#x2009;2.1; 95% CI: 1.35-3.26), and high academic stress (AOR&#x2009;=&#x2009;1.8; 95% CI: 1.25-2.60). Nearly two-thirds (64.2%) of clinical-year students experienced CSA, which was associated with psychological, institutional, and instructional factors. Based on these cross-sectional findings, we suggest that strengthening mentorship, feedback systems, simulated skill practice, and stress-reduction strategies may be beneficial. Future research could explore interventions such as maintaining lower student-to-instructor ratios, bolstering structured feedback systems, increasing simulation-based learning opportunities, and integrating resilience training into health science courses. However, causal inferences cannot be drawn from this cross-sectional design; therefore, intervention studies are needed to evaluate the effectiveness of these approaches. No patients, caregivers, or public members participated in this study; the research focused exclusively on health science students in Ethiopia.