Search

Background: Accurate estimates of gestational age (GA) at birth are important for preterm birth surveillance but can be challenging to obtain in low income countries. Our objective was to develop machine learning models to accurately estimate GA shortly after birth using clinical and metabolomic data. Methods: We derived three GA estimation models using ELASTIC NET multivariable linear regression using metabolomic markers from heel-prick blood samples and clinical data from a retrospective cohort of newborns from Ontario, Canada. We conducted internal model validation in an independent cohort of Ontario newborns, and external validation in heel prick and cord blood sample data collected from newborns from prospective birth cohorts in Lusaka, Zambia and Matlab, Bangladesh. Model performance was measured by comparing model-derived estimates of GA to reference estimates from early pregnancy ultrasound. Results: Samples were collected from 311 newborns from Zambia and 1176 from Bangladesh. The best-performing model accurately estimated GA within about 6 days of ultrasound estimates in both cohorts when applied to heel prick data (MAE 0.79 weeks (95% CI 0.69, 0.90) for Zambia; 0.81 weeks (0.75, 0.86) for Bangladesh), and within about 7 days when applied to cord blood data (1.02 weeks (0.90, 1.15) for Zambia; 0.95 weeks (0.90, 0.99) for Bangladesh). Conclusions: Algorithms developed in Canada provided accurate estimates of GA when applied to external cohorts from Zambia and Bangladesh. Model performance was superior in heel prick data as compared to cord blood data. [ABSTRACT FROM AUTHOR]

Objective: We investigated the association between a child's birth order and emergency room (ER) visits and hospital admissions following 2-,4-,6- and 12-month pediatric vaccinations. Methods: We included all children born in Ontario between April 1st, 2006 and March 31st, 2009 who received a qualifying vaccination. We identified vaccinations, ER visits and admissions using health administrative data housed at the Institute for Clinical Evaluative Sciences. We used the self-controlled case series design to compare the relative incidence (RI) of events among 1st-born and later-born children using relative incidence ratios (RIR). Results: For the 2-month vaccination, the RIR for 1st-borns versus later-born children was 1.37 (95% CI: 1.19–1.57), which translates to 112 additional events/100,000 vaccinated. For the 4-month vaccination, the RIR for 1st-borns vs. later-borns was 1.70 (95% CI: 1.45–1.99), representing 157 additional events/100,000 vaccinated. At 6 months, the RIR for 1st vs. later-borns was 1.27 (95% CI: 1.09–1.48), or 77 excess events/100,000 vaccinated. At the 12-month vaccination, the RIR was 1.11 (95% CI: 1.02–1.21), or 249 excess events/100,000 vaccinated. Conclusions: Birth order is associated with increased incidence of ER visits and hospitalizations following vaccination in infancy. 1st-born children had significantly higher relative incidence of events compared to later-born children. [ABSTRACT FROM AUTHOR]

Background: Live vaccines have distinct safety profiles, potentially causing systemic reactions one to 2 weeks after administration. In the province of Ontario, Canada, live MMR vaccine is currently recommended at age 12 months and 18 months. Methods: Using the self-controlled case series design we examined 271,495 12 month vaccinations and 184,312 18 month vaccinations to examine the relative incidence of the composite endpoint of emergency room visits or hospital admissions in consecutive one day intervals following vaccination. These were compared to a control period 20 to 28 days later. In a post-hoc analysis we examined the reasons for emergency room visits and the average acuity score at presentation for children during the at-risk period following the 12 month vaccine. Results: Four to 12 days post 12 month vaccination, children had a 1.33 (1.29-1.38) increased relative incidence of the combined endpoint compared to the control period, or at least one event during the risk interval for every 168 children vaccinated. Ten to 12 days post 18 month vaccination, the relative incidence was 1.25 (95%, 1.17-1.33) which represented at least one excess event for every 730 children vaccinated. The primary reason for increased events was statistically significant elevations in emergency room visits following all vaccinations. There were non-significant increases in hospital admissions. There were an additional 20 febrile seizures for every 100,000 vaccinated at 12 months. Conclusions: There are significantly elevated risks of primarily emergency room visits approximately one to two weeks following 12 and 18 month vaccination. Future studies should examine whether these events could be predicted or prevented. [ABSTRACT FROM AUTHOR]

Deep neural networks have been widely adopted in numerous domains due to their high performance and accessibility to developers and application-specific end-users. Fundamental to image-based applications is the development of Convolutional Neural Networks (CNNs), which possess the ability to automatically extract features from data. However, comprehending these complex models and their learned representations, which typically comprise millions of parameters and numerous layers, remains a challenge for both developers and end-users. This challenge arises due to the absence of interpretable and transparent tools to make sense of black-box models. There exists a growing body of Explainable Artificial Intelligence (XAI) literature, including a collection of methods denoted Class Activation Maps (CAMs), that seek to demystify what representations the model learns from the data, how it informs a given prediction, and why it, at times, performs poorly in certain tasks. We propose a novel XAI visualization method denoted CAManim that seeks to simultaneously broaden and focus end-user understanding of CNN predictions by animating the CAM-based network activation maps through all layers, effectively depicting from end-to-end how a model progressively arrives at the final layer activation. Herein, we demonstrate that CAManim works with any CAM-based method and various CNN architectures. Beyond qualitative model assessments, we additionally propose a novel quantitative assessment that expands upon the Remove and Debias (ROAD) metric, pairing the qualitative end-to-end network visual explanations assessment with our novel quantitative "yellow brick ROAD" assessment (ybROAD). This builds upon prior research to address the increasing demand for interpretable, robust, and transparent model assessment methodology, ultimately improving an end-user's trust in a given model's predictions. Examples and source code can be found at: https://omni-ml.github.io/pytorch-grad-cam-anim/. [ABSTRACT FROM AUTHOR]

Objective: To develop and internally validate a deep-learning algorithm from fetal ultrasound images for the diagnosis of cystic hygromas in the first trimester. Methods: All first trimester ultrasound scans with a diagnosis of a cystic hygroma between 11 and 14 weeks gestation at our tertiary care centre in Ontario, Canada were studied. Ultrasound scans with normal nuchal translucency were used as controls. The dataset was partitioned with 75% of images used for model training and 25% used for model validation. Images were analyzed using a DenseNet model and the accuracy of the trained model to correctly identify cases of cystic hygroma was assessed by calculating sensitivity, specificity, and the area under the receiver-operating characteristic (ROC) curve. Gradient class activation heat maps (Grad-CAM) were generated to assess model interpretability. Results: The dataset included 289 sagittal fetal ultrasound images;129 cystic hygroma cases and 160 normal NT controls. Overall model accuracy was 93% (95% CI: 88–98%), sensitivity 92% (95% CI: 79–100%), specificity 94% (95% CI: 91–96%), and the area under the ROC curve 0.94 (95% CI: 0.89–1.0). Grad-CAM heat maps demonstrated that the model predictions were driven primarily by the fetal posterior cervical area. Conclusions: Our findings demonstrate that deep-learning algorithms can achieve high accuracy in diagnostic interpretation of cystic hygroma in the first trimester, validated against expert clinical assessment. [ABSTRACT FROM AUTHOR]

Background: Infants are at high risk for influenza illness, but are ineligible for vaccination before 6 months. Transfer of maternal antibodies to the fetus has been demonstrated for 2009 A/H1N1 pandemic vaccines; however, clinical effectiveness is unknown. Our objective was to evaluate the association between 2009 A/H1N1 pandemic vaccination during pregnancy and rates of infant influenza and pneumonia. Methods: We linked a population-based birth cohort to administrative databases to measure rates of influenza and pneumonia diagnosed during ambulatory physician visits, hospitalizations and emergency department visits during one year of follow-up. We estimated incidence rate ratios and 95% confidence intervals (95% CI) using Poisson regression, comparing infants born to A/H1N1-vaccinated women (vaccine-exposed infants) with unexposed infants, adjusted for confounding using high-dimensional propensity scores. Results: Among 117,335 infants in the study, 36,033 (31%) were born to A/H1N1-vaccinated women. Crude rates of influenza during the pandemic (per 100,000 infant-days) for vaccine-exposed and unexposed infants were similar (2.19, 95% CI: 1.27–3.76 and 3.60, 95% CI: 2.51–5.14, respectively), as were crude rates of influenza and pneumonia combined. We did not observe any significant differences in rates of study outcomes between study groups during the second wave of the 2009 A/H1N1 pandemic, nor during any post-pandemic time period. Conclusion: We observed no difference in rates of study outcomes among infants born to A/H1N1-vaccinated mothers relative to unexposed infants born during the second A/H1N1 pandemic wave; however, due to late availability of the pandemic vaccine, the available follow-up time during the pandemic time period was very limited. [ABSTRACT FROM AUTHOR]

Background: Recurrent acute exacerbations of chronic obstructive pulmonary disease (AECOPD) are common, debilitating, costly and often difficult to prevent. Methods: We reviewed records of patients who had COPD and immunoglobulin (Ig) treatment as adjunctive preventative treatment for AECOPD, and documented all AECOPD episodes for one year before and after initiation of Ig treatment. We graded AECOPD episodes as moderate for prescription of antibiotics and/or corticosteroids or for visit to the Emergency Department, and as severe for hospital admission. We conducted a retrospective within-subject self-controlled risk interval analysis to compare the outcome of annual AECOPD rate before and after treatment. Results: We identified 22 cases of certain COPD, of which three had early discontinuation of Ig treatment due to rash and local swelling to subcutaneous Ig, and five had incomplete records leaving 14 cases for analyses. The median baseline IgG level was 5.9 g/L (interquartile range 4.1–7.4). Eight had CT radiographic bronchiectasis. Overall, the incidence of AECOPD was consistently and significantly reduced in frequency from mean 4.7 (± 3.1) per patient-year before, to 0.6 (± 1.0) after the Ig treatment (p = 0.0001). There were twelve episodes of severe AECOPD (in seven cases) in the year prior, and one in the year after Ig treatment initiation (p = 0.016). Conclusions: Ig treatment appears to decrease the frequency of moderate and severe recurrent AECOPD. A prospective, controlled evaluation of adjunctive Ig treatment to standard therapy of recurrent AECOPD is warranted. [ABSTRACT FROM AUTHOR]