Background Machine learning tools that semi-automate data extraction may create efficiencies in systematic review production. We evaluated a machine learning and text mining tool’s ability to (a) automatically extract data elements from randomized trials, and (b) save time compared with manual extraction and verification. Methods For 75 randomized trials, we manually extracted and verified data for 21 data elements. We uploaded the randomized trials to an online machine learning and text mining tool, and quantified performance by evaluating its ability to identify the reporting of data elements (reported or not reported), and the relevance of the extracted sentences, fragments, and overall solutions. For each randomized trial, we measured the time to complete manual extraction and verification, and to review and amend the data extracted by the tool. We calculated the median (interquartile range [IQR]) time for manual and semi-automated data extraction, and overall time savings. Results The tool identified the reporting (reported or not reported) of data elements with median (IQR) 91% (75% to 99%) accuracy. Among the top five sentences for each data element at least one sentence was relevant in a median (IQR) 88% (83% to 99%) of cases. Among a median (IQR) 90% (86% to 97%) of relevant sentences, pertinent fragments had been highlighted by the tool; exact matches were unreliable (median (IQR) 52% [33% to 73%]). A median 48% of solutions were fully correct, but performance varied greatly across data elements (IQR 21% to 71%). Using ExaCT to assist the first reviewer resulted in a modest time savings compared with manual extraction by a single reviewer (17.9 vs. 21.6 h total extraction time across 75 randomized trials). Conclusions Using ExaCT to assist with data extraction resulted in modest gains in efficiency compared with manual extraction. The tool was reliable for identifying the reporting of most data elements. The tool’s ability to identify at least one relevant sentence and highlight pertinent fragments was generally good, but changes to sentence selection and/or highlighting were often required. Protocol https://doi.org/10.7939/DVN/RQPJKS

Background. Machine learning tools that semi-automate data extraction may create efficiencies in systematic review production. We prospectively evaluated an online machine learning and text mining tool’s ability to (a) automatically extract data elements from randomized trials, and (b) save time compared with manual extraction and verification. Methods. For 75 randomized trials published in 2017, we manually extracted and verified data for 21 unique data elements. We uploaded the randomized trials to ExaCT, an online machine learning and text mining tool, and quantified performance by evaluating the tool’s ability to identify the reporting of data elements (reported or not reported), and the relevance of the extracted sentences, fragments, and overall solutions. For each randomized trial, we measured the time to complete manual extraction and verification, and to review and amend the data extracted by ExaCT (simulating semi-automated data extraction). We summarized the relevance of the extractions for each data element using counts and proportions, and calculated the median and interquartile range (IQR) across data elements. We calculated the median (IQR) time for manual and semiautomated data extraction, and overall time savings. Results. The tool identified the reporting (reported or not reported) of data elements with median (IQR) 91 percent (75% to 99%) accuracy. Performance was perfect for four data elements: eligibility criteria, enrolment end date, control arm, and primary outcome(s). Among the top five sentences for each data element at least one sentence was relevant in a median (IQR) 88 percent (83% to 99%) of cases. Performance was perfect for four data elements: funding number, registration number, enrolment start date, and route of administration. Among a median (IQR) 90 percent (86% to 96%) of relevant sentences, pertinent fragments had been highlighted by the system; exact matches were unreliable (median (IQR) 52 percent [32% to 73%]). A median 48 percent of solutions were fully correct, but performance varied greatly across data elements (IQR 21% to 71%). Using ExaCT to assist the first reviewer resulted in a modest time savings compared with manual extraction by a single reviewer (17.9 vs. 21.6 hours total extraction time across 75 randomized trials). Conclusions. Using ExaCT to assist with data extraction resulted in modest gains in efficiency compared with manual extraction. The tool was reliable for identifying the reporting of most data elements. The tool’s ability to identify at least one relevant sentence and highlight pertinent fragments was generally good, but changes to sentence selection and/or highlighting were often required.

Emerging evidence suggests that the COVID-19 pandemic and associated public health measures, including lockdowns and school closures, have been negatively affecting school-aged children’s psychological wellbeing. To identify supports required to mitigate the negative impacts of the COVID-19 pandemic, we gathered in-depth information on school-aged children’s and parents’ lived experiences of COVID-19 and perceptions of its impact on psychological wellbeing in grade 4–6 students in Canada. In this qualitative study, we conducted telephone-based semi-structured interviews with parents (n = 15) and their children (n = 16) from six schools in small and mid-sized northern prairie communities in Canada. Interviews were analyzed through thematic analysis. Three interrelated themes have emerged. First, the start of COVID-19 brought sudden and stressful changes to children’s lives. Second, disruptions to daily life led to feelings of boredom and lack of purpose. Third, limited opportunities for social interaction led to loneliness and an increase in screen time to seek social connection with peers. Results underscore the need for resilience building and the promotion of positive coping strategies to help school-aged children thrive in the event of future health crises or natural disasters.

Capturing socioeconomic inequalities in relation to chronic disease is challenging since socioeconomic status (SES) encompasses many aspects. We constructed a comprehensive individual-level SES index based on a broad set of social and demographic indicators (gender, education, income adequacy, occupational prestige, employment status) and examined its relationship with smoking intensity, a leading chronic disease risk factor. Analyses were based on baseline data from 17,371 participants of Alberta’s Tomorrow Project (ATP), a prospective cohort of adults aged 35–69 years with no prior personal history of cancer. To construct the SES index, we used principal component analysis (PCA) and to illustrate its utility, and examined the association with smoking intensity and smoking history using multiple regression models, adjusted for age and gender. Two components were retained from PCA, which explained 61% of the variation. The SES index was best aligned with educational attainment and occupational prestige, and to a lesser extent, with income adequacy. In the multiple regression analysis, the SES index was negatively associated with smoking intensity (p < 0.001). Study findings highlight the potential of using individual-level SES indices constructed from a broad set of social and demographic indicators in epidemiological research.

Objective: Increasing evidence links unhealthy food environments with diet quality and overweight/obesity. Recent evidence has demonstrated that relative food environment measures outperform absolute measures. Few studies have examined the interplay between the two measures. We examined the separate and combined effects of the absolute and relative densities of unhealthy food outlets within 1600 metre buffers around elementary schools on children’s diet- and weight-related outcomes. Design: This is a cross-sectional study of 812 children, from 39 schools. The Youth Healthy Eating Index (Y-HEI) and daily vegetables and fruit servings were derived from the Harvard Food Frequency Questionnaire (FFQ) for Children and Youth. Measured heights and weights determined body mass index (BMI) z-scores. Food outlets were ranked as healthy, somewhat healthy, and unhealthy according to provincial pediatric nutrition guidelines. Multilevel mixed-effects regression models were used to assess the effect of absolute (number) and relative (proportion) densities of unhealthy food outlets within 1600 metres around schools on diet quality and weight status. Setting: Two urban centres in the province of Alberta, Canada. Participants: Grade 5 students (10-11 years). Results: For children attending schools with a higher absolute number (36+) of unhealthy food outlets within 1600 metres, every 10% increase in the proportion of unhealthy food outlets was associated with 4.1 lower Y-HEI score and 0.9 fewer daily vegetables and fruit. Conclusions: Children exposed to a higher relative density of unhealthy food outlets around a school had lower diet quality, specifically in areas where the absolute density of unhealthy food outlets was also high.