Effect of Applying a Real-Time Medical Record Input Assistance System With Voice Artificial Intelligence on Triage Task Performance in the Emergency Department: Prospective Interventional Study

IVF laboratory management through workflow-based RFID tag witnessing and real-time information entry

BACKGROUND

Dual-person inspection in IVF laboratories cannot fully avoid mix-ups or embryo transfer errors, and data transcription or entry is time-consuming and redundant, often leading to delays in completing medical records.

METHODS

This study introduced a workflow-based RFID tag witnessing and real-time information entry platform for addressing these challenges. To assess its potential in reducing mix-ups, we conducted a simulation experiment in semen preparation to analyze its error correction rate. Additionally, we evaluated its impact on work efficiency, specifically in operation and data entry. Furthermore, we compared the cycle costs between paper labels and RFID tags. Finally, we retrospectively analyzed clinical outcomes of 20,424 oocyte retrieval cycles and 15,785 frozen embryo transfer cycles, which were divided into paper label and RFID tag groups.

RESULTS

The study revealed that comparing to paper labels, RFID tag witnessing corrected 100% of tag errors, didn't affect gamete/embryo operations, and notably shorten the time of entering data, but the cycle cost of RFID tags was significantly higher. However, no significant differences were observed in fertilization, embryo quality, blastocyst rates, clinical pregnancy, and live birth rates between two groups.

CONCLUSIONS

RFID tag witnessing doesn't negatively impact gamete/embryo operation, embryo quality and pregnancy outcomes, but it potentially reduces the risk of mix-ups or errors. Despite highly increased cost, integrating RFID tag witnessing with real-time information entry can remarkably decrease the data entry time, substantially improving the work efficiency. This workflow-based management platform also enhances operational safety, ensures medical informational integrity, and boosts embryologist's confidence.

A Systematic Review of the Application of Artificial Intelligence in Nursing Care: Where are We, and What's Next?

Background

Integrating Artificial Intelligence (AI) into healthcare has transformed the landscape of patient care and healthcare delivery. Despite this, there remains a notable gap in the existing literature synthesizing the comprehensive understanding of AI's utilization in nursing care.

Objective

This systematic review aims to synthesize the available evidence to comprehensively understand the application of AI in nursing care.

Methods

Studies published between January 2019 and December 2023, identified through CINAHL Plus with Full Text, Web of Science, PubMed, and Medline, were included in this review. The Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines guided the identification, screening, exclusion, and inclusion of articles. The convergent integrated analysis framework, as proposed by the Joanna Briggs Institute, was employed to synthesize data from the included studies for theme generation.

Results

A total of 337 records were identified from databases. Among them, 35 duplicates were removed, and 302 records underwent eligibility screening. After applying inclusion and exclusion criteria, eleven studies were deemed eligible and included in this review. Through data synthesis of these studies, six themes pertaining to the use of AI in nursing care were identified: 1) Risk Identification, 2) Health Assessment, 3) Patient Classification, 4) Research Development, 5) Improved Care Delivery and Medical Records, and 6) Developing a Nursing Care Plan.

Conclusion

This systematic review contributes valuable insights into the multifaceted applications of AI in nursing care. Through the synthesis of data from the included studies, six distinct themes emerged. These findings not only consolidate the current knowledge base but also underscore the diverse ways in which AI is shaping and improving nursing care practices.

Deep learning-based natural language processing for detecting medical symptoms and histories in emergency patient triage

OBJECTIVE

The manual recording of electronic health records (EHRs) by clinicians in the emergency department (ED) is time-consuming and challenging. In light of recent advancements in large language models (LLMs) such as GPT and BERT, this study aimed to design and validate LLMs for automatic clinical diagnoses. The models were designed to identify 12 medical symptoms and 2 patient histories from simulated clinician-patient conversations within 6 primary symptom scenarios in emergency triage rooms.

MATERIALS AND METHOD

We developed classification models by fine-tuning BERT, a transformer-based pre-trained model. We subsequently analyzed these models using eXplainable artificial intelligence (XAI) and the Shapley additive explanation (SHAP) method. A Turing test was conducted to ascertain the reliability of the XAI results by comparing them to the outcomes of tasks performed and explained by medical workers. An emergency medicine specialist assessed the results of both XAI and the medical workers.

RESULTS

We fine-tuned four pre-trained LLMs and compared their classification performance. The KLUE-RoBERTa-based model demonstrated the highest performance (F1-score: 0.965, AUROC: 0.893) on human-transcribed script data. The XAI results using SHAP showed an average Jaccard similarity of 0.722 when compared with explanations of medical workers for 15 samples. The Turing test results revealed a small 6% gap, with XAI and medical workers receiving the mean scores of 3.327 and 3.52, respectively.

CONCLUSION

This paper highlights the potential of LLMs for automatic EHR recording in Korean EDs. The KLUE-RoBERTa-based model demonstrated superior classification performance. Furthermore, XAI using SHAP provided reliable explanations for model outputs. The reliability of these explanations was confirmed by a Turing test.