Conectar
With ambient listening systems increasingly adopted in healthcare, analyzing clinician-patient conversations has become essential. The Omaha System is a standardized terminology for documenting patient care, classifying health problems into four domains across 42 problems and 377 signs/symptoms. Manually identifying and mapping these problems is time-consuming and labor-intensive. This study aims to automate health problem identification from clinician-patient conversations using large language models (LLMs) with retrieval-augmented generation (RAG).
Using the Omaha System framework, we analyzed 5118 utterances from 22 clinician-patient encounters in home healthcare. RAG-enhanced LLMs detected health problems and mapped them to Omaha System terminology. We evaluated different model configurations, including embedding models, context window sizes, parameter settings (top k, top p), and prompting strategies (zero-shot, few-shot, and chain-of-thought). Three LLMs—Llama 3.1-8B-Instruct, GPT-4o-mini, and GPT-o3-mini—were compared using precision, recall, and F1-score against expert annotations.
The optimal configuration used a 1-utterance context window, top k = 15, top p = 0.6, and few-shot learning with chain-of-thought prompting. GPT-4o-mini achieved the highest F1-score (0.90) for both problem and sign/symptom identification, followed by GPT-o3-mini (0.83/0.82), while Llama 3.1-8B-Instruct performed worst (0.73/0.72).
Using the Omaha System, LLMs with RAG effectively automate health problem identification in clinical conversations. This approach can enhance documentation completeness, reduce documentation burden, and potentially improve patient outcomes through more comprehensive problem identification, translating into tangible improvements in clinical efficiency and care delivery.
Automating health problem identification from clinical conversations can improve documentation accuracy, reduce burden, and ensure alignment with standardized frameworks like the Omaha System, enhancing care quality and continuity in home healthcare.
Falls among older adults are a major public health concern, often leading to serious outcomes such as fractures, head trauma, and increased mortality. Virtual reality (VR) interventions have emerged as a promising strategy for fall prevention by improving balance, reducing fear of falling, and enhancing confidence. However, the impact of VR interventions on specific outcomes such as fear of falling, balance, and postural control in older adults remains insufficiently synthesized.
Systematic review and meta-analysis.
A comprehensive systematic search of six databases was conducted from inception to January 20, 2025. Randomized controlled trials (RCTs) evaluating VR interventions targeting fear of falling, balance, and postural control in older adults were included. Methodological quality was assessed using the Cochrane risk-of-bias tool (RoB-2). Pooled standardized mean differences (SMDs) with 95% confidence intervals (CIs) were calculated using random-effects models for each outcome.
Seventeen RCTs involving 988 older adults, published between 2016 and 2025, met the inclusion criteria. VR interventions demonstrated significant effects in reducing fear of falling (SMD = −0.40; 95% CI: −0.72 to −0.08; I 2 = 45.10%; p = 0.02), improving balance (SMD = 0.45; 95% CI: 0.07–0.83; I 2 = 73.54%; p = 0.02), and enhancing postural control (SMD = 0.50; 95% CI: 0.13–0.86; I2 = 46.89%; p = 0.01).
This meta-analysis highlights the effectiveness of VR interventions in reducing fear of falling and improving balance and postural control among older adults.
VR represents a valuable tool in fall prevention strategies, addressing key outcomes essential for maintaining independence and mobility in this population.
Nurses and healthcare support staff have a higher suicide risk than the public. This elevated risk calls for increased efforts to support mental health. Additionally, nursing leaders' education on employee-specific suicide prevention is lacking.
An evidence-based project was implemented using the PICO question: Among nurse leaders at an academic healthcare system in California, does the provision of an educational program using role-playing practice and the creation of a suicide prevention toolkit versus no standard education or training improve self-efficacy and knowledge on how to take action with a team member who is suspected of being suicidal or voicing suicidal ideation?
Education sessions were planned based on the literature, with surveys collected preintervention, immediately posteducation, and 1-month postintervention to assess suicide prevention self-efficacy and knowledge. Knowledge was measured using a researcher-constructed questionnaire validated by six suicide prevention experts. The General Self-Efficacy Scale (range: 10–40) was used.
Sixty participants attended one of 11 scheduled remote-learning sessions. Mean self-efficacy significantly improved (pre: 31.3 [n = 46, min: 18, max: 40]; immediate post: 33.49 [n = 37, min: 24, max: 40]; 1-month post: 33.77 [n = 31, min: 28, max: 40]) (X 2 = 8.0184, df = 2, p = 0.01815). The proportion of incorrect knowledge questions was significantly lower postintervention (mean pre: 24.5%, immediate post: 11.5%, 1-month post: 10.7%, X 2 = 23.195, df = 2, p = 0.000001). All participants (100%, n = 55) recommended the program. Leaders reported feeling better prepared to support suicidal employees.
Project results demonstrate the need to provide suicide prevention training for leaders. The authors recommend requiring training/return demonstration competency as a component of new leaders' onboarding. This program can easily be modified for nurses from prelicensure through senior leadership.
Suicide rates in healthcare members are higher than those of the general population. Suicide prevention programs can help nursing leaders feel better prepared to support and connect at-risk healthcare workers with resources.
Pain is a frequent post-stroke health concern, and several non-pharmacological interventions are commonly employed to manage it. However, few reviews have examined the effectiveness of such interventions, making it difficult to draw conclusions about their usefulness. Furthermore, subgroup analysis based on post-stroke pain level or intervention characteristics is rarely performed. This study aimed to investigate the effectiveness of non-pharmacological interventions and evaluate the significant factors associated with post-stroke pain through subgroup analysis.
Systematic review and meta-analysis.
Relevant studies were obtained from seven databases, from their commencement up to March 2024, as well as from the gray literature. The PICOS approach was used to evaluate the eligibility criteria of the studies. The RoB-2 tool was used to determine the risk of bias in each randomized trial. Pooled estimations of standardized mean difference and heterogeneity (quantified with I 2) were obtained using a random-effects model. The stability of the pooled result was then assessed using the leave-one-out approach. STATA 17.0 was used to run the meta-analysis.
Non-pharmacological interventions were effective in reducing pain immediately after intervention (pooled SMDs: −0.79; 95% confidence interval [CI]: −1.06 to −0.53; p < 0.001). The approach involving acupuncture, aquatic therapy, or laser therapy and rehabilitation training was effective for post-stroke hemiplegic shoulder pain. A pooled analysis of non-pharmacological interventions showed that both less than 4 weeks and more than 4 weeks of interventions were effective in alleviating pain in stroke patients.
Non-pharmacological approaches appear to be beneficial for reducing post-stroke pain. The outcomes based on the modalities merit further research.
Further studies are needed to determine the effects of different modalities on pain intensity following a stroke. Furthermore, to avoid overestimation of intervention efficacy, future randomized trials should consider blinding approaches to the interventions delivered.
Identifying health problems in audio-recorded patient–nurse communication is important to improve outcomes in home healthcare patients who have complex conditions with increased risks of hospital utilization. Training machine learning classifiers for identifying problems requires resource-intensive human annotation.
To generate synthetic patient–nurse communication and to automatically annotate for common health problems encountered in home healthcare settings using GPT-4. We also examined whether augmenting real-world patient–nurse communication with synthetic data can improve the performance of machine learning to identify health problems.
Secondary data analysis of patient–nurse verbal communication data in home healthcare settings.
The data were collected from one of the largest home healthcare organizations in the United States. We used 23 audio recordings of patient–nurse communications from 15 patients. The audio recordings were transcribed verbatim and manually annotated for health problems (e.g., circulation, skin, pain) indicated in the Omaha System Classification scheme. Synthetic data of patient–nurse communication were generated using the in-context learning prompting method, enhanced by chain-of-thought prompting to improve the automatic annotation performance. Machine learning classifiers were applied to three training datasets: real-world communication, synthetic communication, and real-world communication augmented by synthetic communication.
Average F1 scores improved from 0.62 to 0.63 after training data were augmented with synthetic communication. The largest increase was observed using the XGBoost classifier where F1 scores improved from 0.61 to 0.64 (about 5% improvement). When trained solely on either real-world communication or synthetic communication, the classifiers showed comparable F1 scores of 0.62–0.61, respectively.
Integrating synthetic data improves machine learning classifiers' ability to identify health problems in home healthcare, with performance comparable to training on real-world data alone, highlighting the potential of synthetic data in healthcare analytics.
This study demonstrates the clinical relevance of leveraging synthetic patient–nurse communication data to enhance machine learning classifier performances to identify health problems in home healthcare settings, which will contribute to more accurate and efficient problem identification and detection of home healthcare patients with complex health conditions.
Nurses increasingly use mindfulness as an effective mental health intervention to reduce psychological distress. The effectiveness of mindfulness-based interventions remains inconclusive, which may lead to implementation of interventions in an inefficient or ineffective manner. This study aimed to examine the effects of mindfulness-based interventions on reducing stress, anxiety, and depression among nurses.
Systematic review and meta-analysis.
Randomized controlled trials (RCTs) were searched using six databases published through May 20, 2023, which evaluated the effects of mindfulness-based interventions on reducing psychological distress among nurses. To assess the quality of methodology included in the RCTs, version 2 of the Cochrane risk-of-bias instrument for RCTs with five domains was used. Standardized mean difference (SMD) with 95% confidence interval (CI) were calculated using the random–effects model in the meta-analyses. Publication bias was assessed using Egger's regression test. Further, the robustness effect size of the pooled analysis was assessed using leave-one-out sensitivity analysis.
A total of 16 RCTs were included in the final analysis. Overall, the modalities appeared to alleviate stress (pooled SMD: −0.50 [95% CI: −0.82 to −0.18]; p < 0.001) and depression (pooled SMD: −0.42 [95% CI: −0.78 to −0.06]; p = 0.02) among nurses.
Mindfulness-based interventions appear to alleviate stress and depression in nurses. Future research evaluating mindfulness-based interventions among working nurses with more rigorous methodological and larger sample size.
Support for nurses' mental health must be included while implementing personal and professional development plans.
FreshRSS 