Automated cooling tower detection through deep learning for Legionnaires' disease outbreak investigations: a model development and validation study.

BACKGROUND: Cooling towers containing Legionella spp are a high-risk source of Legionnaires' disease outbreaks. Manually locating cooling towers from aerial imagery during outbreak investigations requires expertise, is labour intensive, and can be prone to errors. We aimed to train a deep learning computer vision model to automatically detect cooling towers that are aerially visible. METHODS: Between Jan 1 and 31, 2021, we extracted satellite view images of Philadelphia (PN, USA) and New York state (NY, USA) from Google Maps and annotated cooling towers to create training datasets. We augmented training data with synthetic data and model-assisted labelling of additional cities. Using 2051 images containing 7292 cooling towers, we trained a two-stage model using YOLOv5, a model that detects objects in images, and EfficientNet-b5, a model that classifies images. We assessed the primary outcomes of sensitivity and positive predictive value (PPV) of the model against manual labelling on test datasets of 548 images, including from two cities not seen in training (Boston [MA, USA] and Athens [GA, USA]). We compared the search speed of the model with that of manual searching by four epidemiologists. FINDINGS: The model identified visible cooling towers with 95·1% sensitivity (95% CI 94·0-96·1) and a PPV of 90·1% (95% CI 90·0-90·2) in New York City and Philadelphia. In Boston, sensitivity was 91·6% (89·2-93·7) and PPV was 80·8% (80·5-81·2). In Athens, sensitivity was 86·9% (75·8-94·2) and PPV was 85·5% (84·2-86·7). For an area of New York City encompassing 45 blocks (0·26 square miles), the model searched more than 600 times faster (7·6 s; 351 potential cooling towers identified) than did human investigators (mean 83·75 min [SD 29·5]; mean 310·8 cooling towers [42·2]). INTERPRETATION: The model could be used to accelerate investigation and source control during outbreaks of Legionnaires' disease through the identification of cooling towers from aerial imagery, potentially preventing additional disease spread. The model has already been used by public health teams for outbreak investigations and to initialise cooling tower registries, which are considered best practice for preventing and responding to outbreaks of Legionnaires' disease. FUNDING: None.

Suspected Legionella Transmission from a Single Donor to Two Lung Transplant Recipients - Pennsylvania, May 2022.

In July 2022, the Pennsylvania Department of Health received two reports of laboratory-confirmed Legionnaires disease in patients who had recently received lung transplants from the same donor at a single Pennsylvania hospital. The donor's cause of death was freshwater drowning in a river, raising suspicion of potential donor-derived transmission, because Legionella bacteria naturally live in fresh water. Further investigation of patients receiving other organs from the same donor did not identify additional legionellosis cases. Health care-associated infection caused by water exposure at the hospital was also evaluated as a potential source of infection and was found to be unlikely. Hospital water quality parameter measurements collected during May-June 2022 were within expected ranges and no water disruptions were noted, although no testing for Legionella was performed during this period. Notifiable disease data did not identify any other Legionnaires disease cases with exposure to this hospital within the 6 months before or after the two cases. Although laboratory testing did not confirm the source of recipient infections, available data suggest that the most likely source was the donor lungs. This cluster highlights the need for increased clinical awareness of possible infection with Legionella in recipients of lungs from donors who drowned in fresh water before organ recovery.

Suspected Legionella Transmission from a Single Donor to Two Lung Transplant Recipients - Pennsylvania, May 2022.

In July 2022, the Pennsylvania Department of Health received two reports of laboratory-confirmed Legionnaires disease in patients who had recently received lung transplants from the same donor at a single Pennsylvania hospital. The donor's cause of death was freshwater drowning in a river, raising suspicion of potential donor-derived transmission, because Legionella bacteria naturally live in fresh water. Further investigation of patients receiving other organs from the same donor did not identify additional legionellosis cases. Health care-associated infection caused by water exposure at the hospital was also evaluated as a potential source of infection and was found to be unlikely. Hospital water quality parameter measurements collected during May-June 2022 were within expected ranges and no water disruptions were noted, although no testing for Legionella was performed during this period. Notifiable disease data did not identify any other Legionnaires disease cases with exposure to this hospital within the 6 months before or after the two cases. Although laboratory testing did not confirm the source of recipient infections, available data suggest that the most likely source was the donor lungs. This cluster highlights the need for increased clinical awareness of possible infection with Legionella in recipients of lungs from donors who drowned in fresh water before organ recovery.

Estimating Waterborne Infectious Disease Burden by Exposure Route, United States, 2014.

More than 7.15 million cases of domestically acquired infectious waterborne illnesses occurred in the United States in 2014, causing 120,000 hospitalizations and 6,600 deaths. We estimated disease incidence for 17 pathogens according to recreational, drinking, and nonrecreational nondrinking (NRND) water exposure routes by using previously published estimates. In 2014, a total of 5.61 million (95% credible interval [CrI] 2.97-9.00 million) illnesses were linked to recreational water, 1.13 million (95% CrI 255,000-3.54 million) to drinking water, and 407,000 (95% CrI 72,800-1.29 million) to NRND water. Recreational water exposure was responsible for 36%, drinking water for 40%, and NRND water for 24% of hospitalizations from waterborne illnesses. Most direct costs were associated with pathogens found in biofilms. Estimating disease burden by water exposure route helps direct prevention activities. For each exposure route, water management programs are needed to control biofilm-associated pathogen growth; public health programs are needed to prevent biofilm-associated diseases.

Outbreaks Associated with Treated Recreational Water - United States, 2015-2019.

Outbreaks associated with treated recreational water can be caused by pathogens or chemicals in aquatic venues such as pools, hot tubs, water playgrounds, or other artificially constructed structures that are intended for recreational or therapeutic purposes. For the pseriod 2015-2019, public health officials from 36 states and the District of Columbia (DC) voluntarily reported 208 outbreaks associated with treated recreational water. Almost all (199; 96%) of the outbreaks were associated with public (nonbackyard) pools, hot tubs, or water playgrounds. These outbreaks resulted in at least 3,646 cases of illness, 286 hospitalizations, and 13 deaths. Among the 155 (75%) outbreaks with a confirmed infectious etiology, 76 (49%) were caused by Cryptosporidium (which causes cryptosporidiosis, a gastrointestinal illness) and 65 (42%) by Legionella (which causes Legionnaires' disease, a severe pneumonia, and Pontiac fever, a milder illness with flu-like symptoms). Cryptosporidium accounted for 2,492 (84%) of 2,953 cases resulting from the 155 outbreaks with a confirmed etiology. All 13 deaths occurred in persons affected by a Legionnaires' disease outbreak. Among the 208 outbreaks, 71 (34%) were associated with a hotel (i.e., hotel, motel, lodge, or inn) or a resort, and 107 (51%) started during June-August. Implementing recommendations in CDC's Model Aquatic Health Code (MAHC) (1) can help prevent outbreaks associated with treated recreational water in public aquatic venues.