Artificial Intelligence in NICU and PICU: A Need for Ecological Validity, Accountability, and Human Factors

Neonatal nurses' experiences with generative AI in clinical decision-making: a qualitative exploration in high-risk nicus

BACKGROUND

Neonatal nurses in high-risk Neonatal Intensive Care Units (NICUs) navigate complex, time-sensitive clinical decisions where accuracy and judgment are critical. Generative artificial intelligence (AI) has emerged as a supportive tool, yet its integration raises concerns about its impact on nurses' decision-making, professional autonomy, and organizational workflows.

AIM

This study explored how neonatal nurses experience and integrate generative AI in clinical decision-making, examining its influence on nursing practice, organizational dynamics, and cultural adaptation in Saudi Arabian NICUs.

METHODS

An interpretive phenomenological approach, guided by Complexity Science, Normalization Process Theory, and Tanner's Clinical Judgment Model, was employed. A purposive sample of 33 neonatal nurses participated in semi-structured interviews and focus groups. Thematic analysis was used to code and interpret data, supported by an inter-rater reliability of 0.88. Simple frequency counts were included to illustrate the prevalence of themes but were not used as quantitative measures. Trustworthiness was ensured through reflexive journaling, peer debriefing, and member checking.

RESULTS

Five themes emerged: (1) Clinical Decision-Making, where 93.9% of nurses reported that AI-enhanced judgment but required human validation; (2) Professional Practice Transformation, with 84.8% noting evolving role boundaries and workflow changes; (3) Organizational Factors, as 97.0% emphasized the necessity of infrastructure, training, and policy integration; (4) Cultural Influences, with 87.9% highlighting AI's alignment with family-centered care; and (5) Implementation Challenges, where 90.9% identified technical barriers and adaptation strategies.

CONCLUSIONS

Generative AI can support neonatal nurses in clinical decision-making, but its effectiveness depends on structured training, reliable infrastructure, and culturally sensitive implementation. These findings provide evidence-based insights for policymakers and healthcare leaders to ensure AI integration enhances nursing expertise while maintaining safe, patient-centered care.

Adaptation of a Synergy Model-based Patient Acuity Tool for the Electronic Health Record: Proof of Concept

Nurse staffing decisions are often made without input from high-quality, reliable patient acuity measures, especially in medical-surgical settings. Staffing decisions not aligned with patient care needs can contribute to inadequate patient-to-nurse ratios and nurse burnout, potentially resulting in preventable patient harm and death. We conducted a proof-of-concept study to explore the feasibility of adapting an evidence-based patient acuity tool for use in the EHR. A retrospective cohort of pediatric medical-surgical inpatients was used to map electronic patient data variables. We developed an algorithm to calculate the score for one domain of the tool and validated it by comparing it with a score based on a manual chart review. Through multiple rounds of testing and refinement of the variables and algorithm, we achieved 100% concordance between scores generated by the algorithm and the manual chart review. Our proof-of-concept study demonstrates the feasibility and challenges of adapting an evidence-based patient acuity score for automation in the EHR. Further collaboration with data scientists is warranted to operationalize the tool in the EHR and achieve an automated acuity score that can improve staffing decisions, support nursing practice, and enhance team collaboration.

Creating Perinatal Nursing Care Plans Using ChatGPT: A Pathway to Improve Nursing Care Plans and Reduce Documentation Burden

BACKGROUND

Extensive time spent on documentation in electronic health records (EHRs) impedes patient care and contributes to nurse burnout. Artificial intelligence-based clinical decision support tools within the EHR, such as ChatGPT, can provide care plan recommendations to the perinatal nurse. The lack of explicit methodologies for effectively integrating ChatGPT led to our initiative to build and demonstrate our ChatGPT-4 prompt to support nurse care planning.

METHODS

We employed our process model, previously tested with 22 diverse medical-surgical patient scenarios, to generate a tailored prompt for ChatGPT-4 to produce care plan suggestions for an exemplar patient presenting with preterm labor and gestational diabetes. A comparative analysis was conducted by evaluating the output against a "nurse-generated care plan" developed by our team of nurses on content alignment, accuracy of standardized nursing terminology, and prioritization of care.

RESULTS

ChatGPT-4 delivered suggestions for nursing diagnoses, interventions, and outcomes comparable to the "nurse-generated care plan." It accurately identified major care areas, avoided irrelevant or unnecessary recommendations, and identified top priority care. Of the 24 labels generated by ChatGPT-4, 16 correctly utilized standardized nursing terminology.

CONCLUSION

This demonstration of the use of our ChatGPT-4 prompt illustrates the potential of leveraging a large language model to assist perinatal nurses in creating care plans. The next steps are improving the accuracy of ChatGPT-4-generated standardized nursing terminology and integrating our prompt into EHRs. This work supports our broader goal of enhancing patient outcomes while mitigating the burden of documentation that contributes to nurse burnout.

Early Impressions and Adoption of the AtriAmp for Managing Arrhythmias Following Congenital Heart Surgery

AtriAmp is a new medical device that displays a continuous real-time atrial electrogram on telemetry using temporary atrial pacing leads. Our objective was to evaluate early adoption of this device into patient care within our pediatric intensive care unit (PICU). This is a qualitative study using inductive analysis of semi-structured interviews to identify dominant themes. The study was conducted in a single-center, tertiary, academic 21-bed mixed PICU. The subjects were PICU multidisciplinary team members (Pediatric Cardiac Intensivists, PICU Nurse Practitioners, PICU nurses and Pediatric Cardiologists) who were early adopters of the AtriAmp (n = 14). Three prominent themes emerged: (1) Accelerated time from arrhythmia event to diagnosis and treatment; (2) Increased confidence in the accuracy of providers' arrhythmia diagnosis; and (3) Improvement in the ability to educate providers about post-operative arrhythmias. Providers also noted some learning curves, but none compromised medical care or clinical workflow. Insights from early adopters of AtriAmp signal the need for simplicity and fidelity in new PICU technologies. Our research suggests that such technologies can be pivotal to the support and growth of multi-disciplinary teams, even among those who do not participate in early implementation. Further research is needed to understand when and why novel technology adoption becomes widespread in high-stakes settings.

Clinical Decision Support for Improved Neonatal Care: The Development of a Machine Learning Model for the Prediction of Late-onset Sepsis and Necrotizing Enterocolitis

OBJECTIVE

To develop an artificial intelligence-based software system for predicting late-onset sepsis (LOS) and necrotizing enterocolitis (NEC) in infants admitted to the neonatal intensive care unit (NICU).

STUDY DESIGN

Single-center, retrospective cohort study, conducted in the NICU of the Antwerp University Hospital. Continuous monitoring data of 865 preterm infants born at <32 weeks gestational age, admitted to the NICU in the first week of life, were used to train an XGBoost machine learning (ML) algorithm for LOS and NEC prediction in a cross-validated setup. Afterward, the model's performance was assessed on an independent test set of 148 patients (internal validation).

RESULTS

The ML model delivered hourly risk predictions with an overall sensitivity of 69% (142/206) for all LOS/NEC episodes and 81% (67/83) for severe LOS/NEC episodes. The model showed a median time gain of ≤10 hours (IQR, 3.1-21.0 hours), compared with historical clinical diagnosis. On the complete retrospective dataset, the ML model made 721 069 predictions, of which 9805 (1.3%) depicted a LOS/NEC probability of ≥0.15, resulting in a total alarm rate of <1 patient alarm-day per week. The model reached a similar performance on the internal validation set.

CONCLUSIONS

Artificial intelligence technology can assist clinicians in the early detection of LOS and NEC in the NICU, which potentially can result in clinical and socioeconomic benefits. Additional studies are required to quantify further the effect of combining artificial and human intelligence on patient outcomes in the NICU.