Current perspectives on the use of artificial intelligence in critical patient safety

Intensive Care Units (ICUs) have undergone enhancements in patient safety, and artificial intelligence (AI) emerges as a disruptive technology offering novel opportunities. While the published evidence is limited and presents methodological issues, certain areas show promise, such as decision support systems, detection of adverse events, and prescription error identification. The application of AI in safety may pursue predictive or diagnostic objectives. Implementing AI-based systems necessitates procedures to ensure secure assistance, addressing challenges including trust in such systems, biases, data quality, scalability, and ethical and confidentiality considerations. The development and application of AI demand thorough testing, encompassing retrospective data assessments, real-time validation with prospective cohorts, and efficacy demonstration in clinical trials. Algorithmic transparency and explainability are essential, with active involvement of clinical professionals being crucial in the implementation process.

Reclassification by applying the Framingham equation 30 years to subjects with intermediate cardiovascular risk. MARK study

BACKGROUND AND OBJECTIVE

The estimation of cardiovascular risk (CVR) with scores at 30 years old has a special interest in reclassifying in a suitable way <60 year subjects with intermediate CVR. This study analyzes what percentage of patients with intermediate CVR included in the MARK study is reclassified by applying the 30-year Framingham score (FS30). It also analyzes the degree of agreement between the two equations to classify high risk subjects.

PATIENTS AND METHODS

Cross-sectional study of 966 subjects included in the MARK study. The CVR was calculated with the two versions of the FS30 (based on lipids and body mass index) for "hard" cardiovascular events in subjects with intermediate CVR.

RESULTS

The 59% and 61% of the subjects with intermediate CVR would be classified as if they had high CVR to undergo a hard event if we used the FS30 in both versions. 70% of men and 35% of women would be classified as high CVR (p<0.01). The agreement percentage, measured with the Kappa index, between the equations FS30L and FS30BMI to classify the high-risk subjects was 67.9% (in men 67.4% and in women 68.7%).

CONCLUSIONS

In subjects with intermediate CVR the FS30 reclassifies more than the half as high RCV, 2 out of 3 men and 1 out of 3 women.

Validation of the CUETO scoring model for predicting recurrence and progression in T1G3 urothelial carcinoma of the bladder

INTRODUCTION AND OBJECTIVES

Various studies tried to validate Club Urológico Español de Tratamiento Oncológico (CUETO) tables, yet, none of this papers focused on the high and very high risk bladder cancers. The aim of the study was to externally validate the CUETO model for predicting disease recurrence and progression in group of T1G3 tumors treated with BCG immunotherapy.

PATIENTS OR MATERIALS AND METHODS

Data from 414 patients with primary T1G3 bladder cancer were analysed. To evaluate the model discrimination, Cox proportional hazard regression models were created and concordance indexes were calculated.

RESULTS

The median follow-up was 68 months. The recurrence was observed in 212 (51.2%) and 64 patients (15.5%) experienced the recurrence more than once during the study follow-up. Progression of the cancer was observed in 106 patients (25.6%). Radical cystectomy was performed in 115 patients (27.8%) and there were 64 (15.5%) cancer specific deaths. For recurrence and progression probability, the concordance index of the CUETO models was 0.633 and 0.697 respectively. CUETO tables underestimated significantly the risk of recurrence and marginally the risk of progression in the first year of observation. For 5 years of observation, the trend for the recurrence was much less clear. On the contrary, there was slight overestimation in the risk of progression. The study is limited by retrospective nature.

CONCLUSIONS

It was shown that the CUETO risk tables exhibit a fair discrimination for both disease recurrence and progression in T1G3 patients treated with BCG. CUETO scoring model underestimates the risk of tumor recurrence, but predicts well risk of progression.

[Patient safety in antibiotics administration: Risk assessment]

OBJECTIVE

To determine the level of risk in the preparation and administration of antibiotics frequently used in the Intensive Care Unit using a risk matrix.

MATERIAL AND METHOD

A study was conducted using situation analysis and literature review of databases, protocols and good practice guidelines on intravenous therapy, drugs, and their administration routes. The most used antibiotics in the ICU registered in the ENVIN-HELICS program from 1 April to 30 June 2015 were selected. In this period, 257 patients received antimicrobial treatment and 26 antibiotics were evaluated. Variables studied: A risk assessment of each antibiotic using the scale Risk Assessment Tool, of the National Patient Safety Agency, as well as pH, osmolarity, type of catheter recommended for administration, and compatibility and incompatibility with other antibiotics studied.

RESULTS

Almost two-thirds (65.3%) of antibiotics had more than 3 risk factors (represented by a yellow stripe), with the remaining 34.7% of antibiotics having between 0 and 2 risk factors (represented by a green stripe). There were no antibiotics with 6 or more risk factors (represented by a red stripe). Most drugs needed reconstitution, additional dilution, and the use of part of the vial to administer the prescribed dose.

CONCLUSION

More than half of the antibiotics studied had a moderate risk level; thus measures should be adopted in order to reduce it. The risk matrix is a useful tool for the assessment and detection of weaknesses associated with the preparation and administration of intravenous antibiotics.