Becker's Clinical Quality & Infection Control

6 INFECTION CONTROL & PATIENT SAFETY Boston Children's researchers tap machine learning for better flu surveillance By Mackenzie Bean R esearchers created a surveillance model that uses machine learning to provide highly accurate estimates of local flu activity, according to a study published in Nature Communications. For the study, researchers from the Computational Health Informatics Program at Boston Children's Hospital combined two forecasting methods with machine learn- ing to estimate flu activity. The first model, ARGO, uses data from EHRs, flu-related Google searches and historical flu activity for a spe- cific location. The second model analyzes information on spatial-temporal flu activity for nearby locations. Researchers trained the new machine learning model, called ARGOnet, using flu predictions from both models and actual flu data. "The system continuously evaluates the predictive pow- er of each independent method and recalibrates how this information should be used to produce improved flu estimates," senior author Mauricio Santillana, PhD, a CHIP faculty member, said in a press release. Researchers used ARGOnet to analyze flu seasons from September 2014 to May 2017 and found the model made more accurate predictions than ARGO in more than 75 percent of states included in the analysis. n Drug-resistant bacteria may thrive in certain hospital sink drains, study finds By Mackenzie Bean H ospital sinks stationed near toilets in patient rooms may act as reservoirs for Klebsiella pneumoniae car- bapenemase-producing bacteria, according to a study published in the American Journal of Infection Control. Researchers conducted the study in a 26-bed medical inten- sive care unit at a 600-bed hospital in Milwaukee. The rooms have an open-concept design, with a sink by the door and a second sink stationed next to a toilet. Researchers took speci- men samples from every sink drain in the MICU. A majority (87 percent) of sinks next to toilets tested positive for KPC, compared to 21.7 percent of sinks located near the door. Of the five rooms where the sink near the door tested positive, the sink near the toilet also tested positive in four rooms. The hospital unit did not report any carbapenem-resistant Entero- bacteriaceae infections in the past year, meaning KPC may persist long term in sinks. "This study, if validated, could have major implications for infection control," study authors Blake Buchan, PhD, and Silvia Munoz-Price, MD, PhD, said in a press release. "If sinks next to toilets are indeed a reservoir for KPC, additional interventions — such as modified hand hygiene practices and sink disinfec- tion protocols — may be needed to stem the risk of transmis- sion among healthcare providers and patients alike." n Computer model can help control MRSA outbreaks in hospitals By Megan Knowles A computer-based model could help hospitals control outbreaks of methicillin-resistant Staphylococcus aureus and identify MRSA patients who don't show symptoms, according to a study published in eLife. e research team, led by scientists at the Columbia University Mailman School of Public Health in New York City, created a computer model of MRSA outbreaks using more than 2 million admission records from 66 hospitals in Sweden. e records represent a six-year period. e model recreated outbreaks of the most prevalent MRSA strain, UK EMRSA-15, which is present in 16 countries, the U.S. included. Harnessing statistical techniques used in weather forecasting, the model simulates infection transmission within hospitals and infections imported from the community. e model estimated as many as 400 asymptomatic MRSA cases per month in the Swedish hospitals and found up to 61 percent of MRSA infections diagnosed in the hospital setting came from the community. e MRSA simulation also calculates a patient's chances of being in- fected. e researchers tested the value of these chances by simulating a treatment intervention for high-risk patients. e researchers' targeted intervention was better at controlling an outbreak than current practices, which involve either treating patients who have spent the most time in hospital, treating patients with the most contacts in the hospital, or using contact-tracing to treat patients exposed to another patient with the infection. "Compared with traditional intervention strategies that may over- look a considerable number of invisible colonized patients, this new model-inference system can identify a pivotal group for treatment, namely individuals who may otherwise transmit MRSA asymptomat- ically," said study author Sen Pei, PhD, an associate research scientist at Columbia University Mailman School of Public Health. e researchers plan to apply their system to other antimicrobial-resis- tant pathogens and settings with a higher burden of disease.n

• Cover