Risk Management and Patient Safety in the Artificial Intelligence Era: A Systematic Review

Healthcare, Journal Year: 2024, Volume and Issue: 12(5), P. 549 - 549

Published: Feb. 27, 2024

Background: Healthcare systems represent complex organizations within which multiple factors (physical environment, human factor, technological devices, quality of care) interconnect to form a dense network whose imbalance is potentially able compromise patient safety. In this scenario, the need for hospitals expand reactive and proactive clinical risk management programs easily understood, artificial intelligence fits well in context. This systematic review aims investigate state art regarding impact AI on processes. To simplify analysis outcomes motivate future standardized comparisons with any subsequent studies, findings present will be grouped according possibility applying prevention different incident type groups as defined by ICPS. Materials Methods: On 3 November 2023, literature Preferred Reporting Items Systematic Reviews Meta-Analyses (PRISMA) guidelines was carried out using SCOPUS Medline (via PubMed) databases. A total 297 articles were identified. After selection process, 36 included review. Results Discussion: The studies allowed identification three main “incident type” domains: healthcare-associated infection, medication. Another relevant application concerns topic reporting. Conclusions: highlighted that can applied transversely various contexts enhance safety facilitate errors. It appears promising tool improve management, although its use requires supervision cannot completely replace skills. outcome enable comparison reviews, it deemed useful refer pre-existing taxonomy adverse events. However, results study usefulness not only practice, but also improving an essential tool, For reason, areas processes should include additional class relating tools. purpose, considered convenient ICPS classification.

Language: Английский

---

Advancing Patient Safety: The Future of Artificial Intelligence in Mitigating Healthcare-Associated Infections: A Systematic Review

Healthcare, Journal Year: 2024, Volume and Issue: 12(19), P. 1996 - 1996

Published: Oct. 6, 2024

Healthcare-associated infections are that patients acquire during hospitalization or while receiving healthcare in other facilities. They represent the most frequent negative outcome healthcare, can be entirely prevented, and pose a burden terms of financial human costs. With development new AI ML algorithms, hospitals could develop automated surveillance prevention models for HAIs, leading to improved patient safety. The aim this review is systematically retrieve, collect, summarize all available information on application impact HAI and/or prevention.

Language: Английский

---

Artificial intelligence in wound care: diagnosis, assessment and treatment of hard-to-heal wounds: a narrative review

Journal of Wound Care, Journal Year: 2024, Volume and Issue: 33(4), P. 229 - 242

Published: April 2, 2024

Objective: The effective assessment of wounds, both acute and hard-to-heal, is an important component in the delivery by wound care practitioners efficacious for patients. Improved diagnosis, optimising treatment regimens, enhanced prevention wounds aid providing patients with a better quality life (QoL). There significant potential use artificial intelligence (AI) health-related areas such as care. However, AI-based systems remain to be developed point where they can used clinically deliver high-quality We have carried out narrative review development AI hard-to-heal wounds. retrieved 145 articles from several online databases other resources, 81 them were included this review. Our shows that application offers benefits assessment/diagnosis, monitoring As well offering improved QoL, may also enable healthcare resources.

Language: Английский

---

Remote digital surgical wound monitoring and surveillance using smartphones

Journal of Wound Care, Journal Year: 2025, Volume and Issue: 34(Sup4b), P. S1 - S25

Published: March 20, 2025

Language: Английский

---

Combining simulation models and machine learning in healthcare management: strategies and applications

Progress in Biomedical Engineering, Journal Year: 2024, Volume and Issue: 6(2), P. 022001 - 022001

Published: Jan. 24, 2024

Abstract Simulation models and artificial intelligence (AI) are largely used to address healthcare biomedical engineering problems. Both approaches showed promising results in the analysis optimization of processes. Therefore, combination simulation AI could provide a strategy further boost quality health services. In this work, systematic review studies applying hybrid approach management challenges was carried out. Scopus, Web Science, PubMed databases were screened by independent reviewers. The main strategies combine as well major application scenarios identified discussed. Moreover, tools algorithms implement proposed described. Results that machine learning appears be most employed with models, which mainly rely on agent-based discrete-event systems. scarcity heterogeneity included suggested standardized framework learning-simulation is yet defined. Future efforts should aim use these design novel intelligent in-silico processes effective translation clinics.

Language: Английский

---

Improving Surgical Site Infection Prediction Using Machine Learning: Addressing Challenges of Highly Imbalanced Data

Diagnostics, Journal Year: 2025, Volume and Issue: 15(4), P. 501 - 501

Published: Feb. 19, 2025

Background: Surgical site infections (SSIs) lead to higher hospital readmission rates and healthcare costs, representing a significant global burden. Machine learning (ML) has demonstrated potential in predicting SSIs; however, the challenge of addressing imbalanced class ratios remains. Objectives: The aim this study is evaluate enhance predictive capabilities machine models for SSIs by assessing effects feature selection, resampling techniques, hyperparameter optimization. Methods: Using routine SSI surveillance data from multiple hospitals Saudi Arabia, we analyzed dataset 64,793 surgical patients, whom 1632 developed SSI. Seven algorithms were created tested: Decision Tree (DT), Gaussian Naive Bayes (GNB), Support Vector (SVM), Logistic Regression (LR), Random Forest (RF), Stochastic Gradient Boosting (SGB), K-Nearest Neighbors (KNN). We also improved several strategies, such as undersampling oversampling. Grid search five-fold cross-validation was employed comprehensive optimization, conjunction with balanced sampling techniques. Features selected using filter method based on their relationships target variable. Results: Our findings revealed that RF achieves highest performance, an MCC 0.72. synthetic minority oversampling technique (SMOTE) best-performing technique, consistently enhancing performance most models, except LR GNB. struggles imbalance due its linear assumptions bias toward majority class, while GNB's reliance independence distribution make it unreliable under-represented classes. For computational efficiency, Instance Hardness Threshold (IHT) offers viable alternative though may compromise some extent. Conclusions: This underscores ML effective tools risk, warranting further clinical exploration improve patient outcomes. By employing advanced techniques robust validation methods, these demonstrate promising accuracy reliability events, even face imbalances. In addition, ensures more reliable evaluation model's particularly presence dataset, where other metrics fail provide accurate evaluation.

Language: Английский

---

Artificial intelligence in hospital infection prevention: an integrative review

Frontiers in Public Health, Journal Year: 2025, Volume and Issue: 13

Published: April 2, 2025

Hospital-acquired infections (HAIs) represent a persistent challenge in healthcare, contributing to substantial morbidity, mortality, and economic burden. Artificial intelligence (AI) offers promising potential for improving HAIs prevention through advanced predictive capabilities. To evaluate the effectiveness, usability, challenges of AI models preventing, detecting, managing HAIs. This integrative review synthesized findings from 42 studies, guided by SPIDER framework inclusion criteria. We assessed quality included studies applying TRIPOD checklist individual AMSTAR 2 tool reviews. demonstrated high accuracy detection, surveillance, multiple HAIs, with surgical site urinary tract frequently achieving area-under-the-curve (AUC) scores exceeding 0.80, indicating strong reliability. Comparative data suggest that while both machine learning deep approaches perform well, some may offer slight advantages complex environments. Advanced algorithms, including neural networks, decision trees, random forests, significantly improved detection rates when integrated EHRs, enabling real-time surveillance timely interventions. In resource-constrained settings, non-real-time utilizing historical EHR showed considerable scalability, facilitating broader implementation infection control. AI-supported systems outperformed traditional methods accurately identifying enhancing compliance hand hygiene protocols. Furthermore, Explainable (XAI) frameworks interpretability tools such as Shapley additive explanations (SHAP) values increased clinician trust facilitated actionable insights. also played pivotal role antimicrobial stewardship predicting emergence multidrug-resistant organisms guiding optimal antibiotic usage, thereby reducing reliance on second-line treatments. However, need comprehensive training, integration costs, ensuring compatibility existing workflows were identified barriers widespread adoption. The HAI management represents potentially transformative shift capabilities supporting effective control measures. Successful necessitates standardized validation protocols, transparent reporting, development user-friendly interfaces ensure seamless adoption healthcare professionals. Variability sources model validations across underscores necessity multicenter collaborations external consistent performance diverse Innovations viable solutions scaling applications low- middle-income countries (LMICs), addressing higher prevalence these regions. Intelligence stands fight against hospital-acquired infections, offering prevention, management. fully realize its potential, sector must prioritize rigorous standards, incorporation build confidence. By adopting scalable fostering interdisciplinary collaborations, can overcome barriers, integrating seamlessly into policies ultimately patient safety care quality. Further research is needed cost-effectiveness, real-world applications, strategies (e.g., training explainable AI) improve broaden clinical

Language: Английский

---

Classification and regression model to manage the hospitalization for laparoscopic cholecystectomy

Scientific Reports, Journal Year: 2023, Volume and Issue: 13(1)

Published: Sept. 7, 2023

Gallstone disease (GD) is one of the most common morbidities in world. Laparoscopic Cholecystectomy (LC) currently gold standard, performed about 96% cases. The affected groups are elderly, who generally have higher pre- and post-operative morbidity mortality rates longer Length Stay (LOS). For this reason, several indicators been defined to improve quality efficiency contain costs. In study, data from patients underwent LC at "San Giovanni di Dio e Ruggi d'Aragona" University Hospital Salerno years 2010-2020 were processed using a Multiple Linear Regression (MLR) model Classification algorithms order identify variables that influence LOS. results 2352 analyzed showed pre-operative LOS Age independent particular, MLR had R2 value equal 0.537 best classification algorithm, Decision Tree, an accuracy greater than 83%. conclusion, both produced significant could provide important support management healthcare process.

Language: Английский

---

Flow Analysis of Mastectomy Patients Using Length of Stay: A Single-Center Study

BioMedInformatics, Journal Year: 2024, Volume and Issue: 4(3), P. 1725 - 1744

Published: July 19, 2024

Background: Malignant breast cancer is the most common affecting women worldwide. The COVID-19 pandemic appears to have slowed diagnostic process, leading an enhanced use of invasive approaches such as mastectomy. increased a surgical procedure pushes towards objective analysis patient flow with measurable quality indicators length stay (LOS) in order optimize it. Methods: In this work, different regression and classification models were implemented analyze total LOS function set independent variables (age, gender, pre-op LOS, discharge ward, year discharge, type procedure, presence hypertension, diabetes, cardiovascular disease, respiratory secondary tumors, surgery complications) extracted from records patients undergoing mastectomy at ‘San Giovanni di Dio e Ruggi d’Aragona’ University Hospital Salerno (Italy) years 2011–2021. addition, impact was assessed by statistically comparing data discharged 2018–2019 those 2020–2021. Results: results obtained generally show good performance characterizing particular case studies. Among models, best predicting described above polynomial regression, R2 value 0.689. algorithms that operated on divided into 3 arbitrary classes also proved be tools, reaching 79% accuracy voting classifier. variables, both showed ward complications during had greatest LOS. final focus assess statically significant increase complications. Conclusion: Through study, it possible validate characterize patients. proves excellent indicator performance, through its advanced methods, machine learning algorithms, understand which demographic organizational collected thus build simple predictors support healthcare management.

Language: Английский

---

Artificial intelligence and prescription of antibiotic therapy: present and future

Expert Review of Anti-infective Therapy, Journal Year: 2024, Volume and Issue: 22(10), P. 819 - 833

Published: Aug. 18, 2024

In the past few years, use of artificial intelligence in healthcare has grown exponentially. Prescription antibiotics is not exempt from its rapid diffusion, and various machine learning (ML) techniques, logistic regression to deep neural networks large language models, have been explored literature support decisions regarding antibiotic prescription.

Language: Английский

---