Using Machine Learning to Reduce Burden on Infection Control Staff

Ferreira, Stephani Amanda Lukasewicz; Meneses, Arateus Crysham Franco; Vaz, Tiago Andres; Carvalho, Otávio Luiz da Fontoura; Dalmora, Camila Hubner; Vanni, Daiane Pressoto; Berti, Isabele Ribeiro; Rocca, Fabiane; Ramos, Márcio Pereira; Santos, Rodrigo Pires dos

doi:10.1056/cat.22.0071

Cited by 1 publication

(8 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…[18][19][20] The semiautomated surveillance method studied here has shown a sensitivity of 97% and specificity of 98.2%. 9 In this single-center ML study, we used 1 database and 1 gold standard for performance comparisons. Performance results may change related to the reference golden standard chosen or different data analysis by the algorithm.…”

Section: Discussionmentioning

confidence: 99%

“…Manual surveillance was performed by the hospital infection control team based on bacterial culture results, and patient records were collected for review. Between July and December 2021, a semiautomated surveillance method based on ML algorithms as described by Ferreira et al 9 was implemented in the hospital. In this semiautomated process, the artificial intelligence (AI) tool classifies patients with potential HAIs, and ICPs subsequently validate the classification.…”

Section: Methodsmentioning

confidence: 99%

“…The performance of the ML algorithm has been reported elsewhere. 9,10 For this study, the manual diagnosis was reviewed by independent specialists and the golden standard was reclassified strictly according to CDC Anvisa criteria. The ML model was retrained according to these new standards.…”

Section: Methodsmentioning

confidence: 99%

“…The random forest algorithm randomly uses decision trees to explore potential connections of each variable to an outcome. 9 The features in the model refer to the individual measurable properties or characteristics of the data that are used as inputs to an ML algorithm. The importance of a feature is the average information gained during the forest construction; it is expressed as a percentage.…”

Section: Methodsmentioning

confidence: 99%

“…8 In a previous study, in which we searched for infections globally, the machine learning (ML) model outperformed manual surveillance by 42%, and time spent on record review decreased by 71%. 9 In this study, we compared manual surveillance with the supervised ML semiautomated method, and we explored the types of infection that each type of surveillance identified as well as the features of importance depicted by the ML model.…”

mentioning

confidence: 99%

See 4 more Smart Citations

Hospital-acquired infections surveillance: The machine-learning algorithm mirrors National Healthcare Safety Network definitions

Lukasewicz Ferreira,

Franco Meneses,

Vaz

et al. 2024

Infect. Control Hosp. Epidemiol.

View full text Add to dashboard Cite

Background: Surveillance of hospital-acquired infections (HAIs) is the foundation of infection control. Machine learning (ML) has been demonstrated to be a valuable tool for HAI surveillance. We compared manual surveillance with a supervised, semiautomated, ML method, and we explored the types of infection and features of importance depicted by the model. Methods: From July 2021 to December 2021, a semiautomated surveillance method based on the ML random forest algorithm, was implemented in a Brazilian hospital. Inpatient records were independently manually searched by the local team, and a panel of independent experts reviewed the ML semiautomated results for confirmation of HAI. Results: Among 6,296 patients, manual surveillance classified 183 HAI cases (2.9%), and a semiautomated method found 299 HAI cases (4.7%). The semiautomated method added 77 respiratory infections, which comprised 93.9% of the additional HAIs. The ML model considered 447 features for HAI classification. Among them, 148 features (33.1%) were related to infection signs and symptoms; 101 (22.6%) were related to patient severity status, 51 features (11.4%) were related to bacterial laboratory results; 40 features (8.9%) were related to invasive procedures; 34 (7.6%) were related to antibiotic use; and 31 features (6.9%) were related to patient comorbidities. Among these 447 features, 229 (51.2%) were similar to those proposed by NHSN as criteria for HAI classification. Conclusion: The ML algorithm, which included most NHSN criteria and >200 features, augmented the human capacity for HAI classification. Well-documented algorithm performances may facilitate the incorporation of AI tools in clinical or epidemiological practice and overcome the drawbacks of traditional HAI surveillance.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%

mentioning

confidence: 99%

See 3 more Smart Citations