Health Insurance CEO Pulls Back Curtain on Prior Auth Reform Newsweek
source
More measles outbreaks put US total within single digits of modern-day record – CIDRAP
Natalya Maisheva/iStock
In its weekly update today, the US Centers for Disease Control and Prevention (CDC) reported 40 more measles cases today, boosting the number of infections this year to 1,267, which is just 8 shy of passing the total in 2019, which was the highest since the disease was eliminated in the country in 2000.
Though the large outbreak in West Texas has slowed substantially, the number of smaller outbreaks and travel-related cases continues to grow. The CDC this week reported 4 more outbreaks, raising the national total to 27. So far this year, 88% of confirmed cases have been linked to outbreaks. For comparison, the United States had 16 outbreaks for all of 2024.
The US measles surge is occurring amid even bigger rises in Canada and Mexico, and all three countries have had large outbreaks fueled by virus spread in Mennonite communities, though health officials warn that outbreaks can affect anywhere pockets of undervaccinated people live.
In an update on measles in the Americas yesterday, the Pan American Health Organization (PAHO) said Canada has confirmed 3,170 cases, 1 of them fatal, and Mexico has recorded 2,597 cases, which includes 9 deaths. Canada has the most cases since the country eliminated the disease in 1998; meanwhile, most of Mexico’s cases are centered in Chihuahua state.
Measles cases in the Americas this year are up 29-fold compared to the same period in 2024, PAHO said. For the region, cases began rising in the third week of the year, peaking in late April will infections concentrated in vaccine-hesitant communities in multiple regions of the Americas. The proportion of cases was highest in children in young adults, but the incidence rate was highest in younger children. Thirty percent of patients were unvaccinated, while vaccination status wasn’t known for 65%.
The Wyoming Department of Health yesterday announced the state’s first case since 2010, which involves an unvaccinated child from Natrona County, which is home to Casper.
So far, the source of the child’s infection isn’t known, and the WDH said public exposure may have occurred during the child’s brief visits to a Casper emergency department on June 24 and June 25.
Alexia Harrist, MD. PhD, state health officer warned that measles is one of the most contagious diseases, but is preventable. “The MMR [measles, mumps, and rubella] vaccine is safe and highly effective, providing long-lasting protection. Two doses of MMR vaccine are about 97% effective in preventing measles, and we recommend that all Wyoming residents ensure they and their children are up to date on MMR vaccinations.”
In other developments, a handful of states reported more cases. Utah, which recently confirmed its first cases that were followed by more related detections, reported two more infections, raising its total to nine.
In Michigan, the Kent County Health Department, located in Grand Rapids, announced a measles case in a young child whose family had recently traveled internationally. It added that the new case marks the county’s second case of the year and the 17th to be confirmed in Michigan.
Meanwhile, Florida has reported its third case of the year, involving a young adult from Leon County, home to Tallahassee, who was exposed during travel outside the country in June, according to a local media report that cited the Florida Department of Health.
The Kansas Department of Health and Environment today reported 3 more cases, all linked to an outbreak in the southwest of the state that was tied to the large West Texas outbreak. The state now has 83 cases, 80 of them linked to the main outbreak that spans 11 counties.
This week’s meeting of ACIP is likely to mark its end—for now—as a vaccine advisory body.
ACIP’s chair signaled that new work groups will take up 2 new topics: cumulative vaccine doses in children and impacts from established vaccines, including hepatitis B in infants.
Also, the group recommends Merck’s RSV preventive for babies and weighs in on the upcoming season’s flu vaccines.
It is not yet known which strain of H5N1 was involved in the latest Cambodian case.
The announcement comes a week after New Mexico officials reported measles in a wastewater sample collected in the area, hinting of an undiagnosed infection.
The patients had contact with sick poultry, a feature of most H5N1 infections in Cambodia.
Experts warn that the federal vaccine advisory group’s decision to revisit childhood immunizations and its controversial vote on thimerosal will undermine confidence in vaccines.
In a pre-recorded video delivered to Gavi leaders, Kennedy accuses the group of ignoring vaccine safety concerns and said the US government will withhold financial support.
Listed in critical condition, the woman had recently handled and cooked sick poultry.
The initial patient in Michigan’s outbreak was exposed to a sick out-of-state traveler.
Help make CIDRAP’s vital work possible
CIDRAP – Center for Infectious Disease Research & Policy
Research and Innovation Office, University of Minnesota, Minneapolis, MN
Email us
© 2025 Regents of the University of Minnesota. All rights Reserved.
The University of Minnesota is an equal opportunity educator and employer
Research and Innovation Office | Contact U of M | Privacy Policy
Newsletter subscribe
Leave a Comment
Startup builds tool to streamline health care credentialing – Spokane Journal of Business
Spokane-based health tech startup Credential Network Inc. has secured a $250,000 grant from the Health Sciences & Services Authority of Spokane County to expand its team and establish its headquarters in the region.
Founded late last year, the company is developing software that aims to streamline the credentialing process for health care professionals, a cumbersome yet essential process in ensuring that medical staff meet the standards set by regulatory bodies, insurance companies, and health care institutions.
Dylan Avatar, founder of Credential Network, says that with the HSSA grant, Credential Network plans to grow its Spokane-based workforce and deepen its ties to the region’s innovation ecosystem.
“Our goal is to bring the fast pass to credentialing,” Avatar says. “We don’t want to bypass credentialing. We want to streamline it and in doing so, give people higher trust in the providers who are going through it because we’re utilizing intelligent technologies to help build these profiles.”
The company is currently beta testing its software tool and has seven beta clients based in the Pacific Northwest. Avatar anticipates to launch the software sometime in the third quarter of this year.
The medical credentialing process faces several challenges that make it difficult and lengthy for health care professionals to verify their qualifications and competence, Avatar contends. Such challenges often include redundant paperwork, manual data entry, and long approval times.
Credential Network’s platform aims to centralize and automate this process, offering a digital interface for credentialing specialists and a secure digital wallet for providers to manage their own credentials, he says.
The goal, he adds, is to speed up onboarding without compromising on safety or compliance, a process he and his team believe is possible through intelligent design and secure technology.
Credential Network is led by Avatar, alongside co-founders’ chief business officer Jeff Falconer, and chief technology officer, Thomas Boyles. Their team includes two full-time staff members and contractors. Two additional team members are also in the process of relocating to Spokane, Avatar notes. The company plans to have at least seven Spokane-based employees within the next six months, pending additional fundraising. The company is also supported by a group of health care and credentialing advisors who help guide and ensure that Credential Network’s tools are aligned with their mission while also meeting the needs of health care professionals.
The company operates on a hybrid model, utilizing local coworking spaces like Fuel, the downtown coworking space launched by Cowles Ventures, the venture capital arm of Cowles Co., and working remotely from home.
While Spokane is laying the foundation for the company’s eventual headquarters, Credential Network has its eyes set on national reach, Avatar says.
“We’re listening to our clients, and we’re listening to the industry and we’re developing new features and functionality by the day,” he says.
The Credential Network founders bring a mix of technical and industry-specific expertise. Avatar previously worked for and was part of the team that founded California-based digital credentialing platform Merit, which offers digital credentialing services for large-scale government programs. Falconer has over two decades of experience in medical software and government sales. Boyles brings decades of experience leading teams in medical technology.
The idea for Credential Network began to take shape during an unexpected chapter in Avatar’s life. In December 2023, Avatar was seriously injured in a wingsuit base jumping accident that left him hospitalized for six weeks. While his physical recovery was demanding, it also offered him the opportunity to learn more about the inner workings of the health care system.
“I couldn’t really lift my arms, but I could ask questions,” Avatar says. “So I spent a lot of time understanding the different perspectives of the health care providers.”
Drawing on his experience in digital credentialing and verified identity, he began to have conversations with the people aiding his recovery about the health care credentialing process. What he heard was consistent: it’s a critical, highly regulated system, but one that is still manual and often a source of major delays.
“To quote one of our first things I heard was ‘it’s a constant headache,’” Avatar says. “And that’s been mirrored within the dozens of provider conversations and administrator conversations we’ve had.”
Avatar’s accident happened in Arizona, and he was initially treated there, but eventually returned home to Spokane for multiple follow-up surgeries and a long course of physical therapy. During his physical therapy journey, Avatar developed a close relationship with Brian Cronin, owner of U-District Physical Therapy & Institute of Sports Performance and an adjunct professor at Whitworth University. Cronin would go on to become a trusted advisor to Credential Network, providing initial feedback on their platform to ensure the company was building something valuable and viable for health care businesses like U-District Physical Therapy.
After his recovery, Avatar returned to his remote job at Merit, but his near-death experience had changed his perspective and his priorities. After some time, he left amicably and took some time to think about what he wanted to do next, and decided he wanted to work in the medical space in some form.
“I just started putting a lot of time and focus into things that I think matter,” Avatar says. “I did go back to my last company, and I helped them for a little, but I knew my true calling was to come and help the medical space. I just felt indebted.”
Avatar launched Credential Network remotely from his home in Spokane, with his dining room table as the centerpiece where he and his partners would lay the foundation for their new startup. He then dove into the startup, innovation, and tech scene in Spokane and found a community that cared about innovation, coupled with a healthy amount of mentorship, he says.
As time went on and he considered moving from a remote model to a hybrid model, or possibly establishing a brick-and-mortar headquarters, Spokane seemed like the perfect combination of creative, innovative leaders and players set within a strong health care community.
For Avatar, Credential Network was born out of a newfound passion and purpose to root his life into the things that matter, he says.
“It’s all being inspired heavily from my accident,” Avatar says. “Everything I’ve been doing, ever since the accident, has a tremendous amount of intentionality behind it. I’m so stoked to be alive, and every day I just carry that energy forward.”
You must have JavaScript enabled to enjoy a limited number of articles over the next 30 days.
Subscribe today to our free E-Newsletters!
All content copyright © 2025 by the Journal of Business and Northwest Business Press Inc. All rights reserved.
Design, CMS, Hosting & Web Development :: ePublishing
Leave a Comment
A study on factors influencing digital sports participation among Chinese secondary school students based on explainable machine learning – nature.com
Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.
Advertisement
Scientific Reports volume 15, Article number: 15657 (2025)
1574
1
Metrics details
This study utilized data from 4,925 Hong Kong students in the 2018 Programme for International Student Assessment (PISA) to investigate factors influencing secondary school students’ use of digital devices for sports participation and their threshold effects. Univariate analysis revealed significant differences in 16 variables including gender. Multilevel logistic regression identified five significant influencing factors (p < 0.05): academic performance, weekly physical education class days, household ICT resources, school ICT resources, and ICT social perception, which were incorporated as features in machine learning models. Through grid search with 5-fold cross-validation, we constructed and optimized four basic machine learning models (GNB, GBDT, KNN, and LR), then developed an ensemble stacking model using base models with AUC values exceeding 0.70. The best-performing stacking model was selected for interpretability analysis. These findings were supported by swarm plots consistent with the multilevel logistic regression results, confirming the reliability of the conclusions. Feature dependence plots further demonstrated threshold effects for specific variables. The study concludes that the entrenched cultural preference for academics over physical activity, weekly physical education class frequency (with two days as the threshold), household and school ICT resources (using high requirements as the threshold), and differences in ICT social perception significantly contribute to digital sports participation disparities among secondary students. Therefore, academically high-achieving students require particular attention in digital sports promotion. Schools should strive to increase the number of weekly physical education classes to at least two days or more, and actively configure equipment and environment to promote the digitalization of physical education. Furthermore, relevant departments can explore the social attributes of digital sports to enhance the possibility of secondary school students’ participation in digital sports.
Digital technology has brought unprecedented changes to human society, and Digital Sports has become a focal point of international research, increasingly valued and applied in the sports and health industries1. The General Administration of Sport of China defines Digital Sports as physical activities that transcend temporal and spatial boundaries, connecting sports scenarios and smart devices through information technology to help participants better achieve the goals of physical fitness and mental well-being2. However, every indirect consequence of technological and economic transformation threatens class differentiation and drives changes in the current class structure3. With the deep integration of digital technology and society, a digital divide has emerged between different social groups4. The “divide” reflects a state in which two groups are “split” and differentiated5. Currently, the digital divide has begun to manifest in new forms among micro sports organizations and athletes6, leading to inequalities in Digital Sports. Therefore, the phenomenon of the Digital Sports divide should be given significant attention, particularly for secondary school students, who are at a critical stage of physical, psychological, and social-behavioral development.
Several critical issues warrant explanation: Among secondary students divided by the digital sports “divide,” in which aspects do significant differences manifest? What factors substantially contribute to the formation of these disparities? Furthermore, it must be noted that positively (negatively) correlated influencing factors do not inherently denote positive (negative) impacts, as their effects depend on variable thresholds. At what specific thresholds do these factors precipitate divisions in students’ digital sports participation? Given that these divisions adversely affect educational outcomes and returns while hindering the popularization of digital sports, this study aims to elucidate the mechanisms underlying the digital sports divide by addressing these questions. Ultimately, our findings seek to provide insights for bridging the digital divide, eliminating digital “privilege” in sports, and advancing digital equity initiatives.
Ecological Systems Theory (EST), proposed by American psychologist Urie Bronfenbrenner, provides a theoretical framework for selecting independent variables to study multilevel influences on secondary school students’ participation in digital sports7. The theory emphasizes that individual development arises from progressive and reciprocal interactions between individuals and their environments. In this context, the individual is positioned at the center of an ecosystem. The environmental system comprises five nested levels: microsystems, mesosystems, exosystems, macrosystems, and chronosystems8. Among these, the home and school environments fall under microsystems, which are most directly relevant to students9. Consequently, this study focuses variable selection on three core levels: personal attributes, home environment, and school environment. Additionally, this paper adopts the definition of digital sports as “physical activity mediated by digital technology,” which inherently involves two essential dimensions: digital technology use and physical activity participation. The selection of independent variables must account for potential factors affecting both dimensions. Building on this discussion, the present study incorporates factors related to students’ use of information technology or participation in physical activities at the individual, family, and school levels. It further integrates findings from prior research to explore and identify the factors and mechanisms influencing students’ engagement in digital sports.
Students’ use of ICT is influenced by multiple factors across individual, family, and school levels. At the individual level, gender, stage of school, and digital literacy significantly impact ICT utilization. Research indicates a notable gap in ICT use between male and female students10. Digital skills, as a critical determinant of device usage, shape students’ ability to engage fully in academic, cultural, and social activities11. Additionally, students’ attitudes toward ICT vary significantly12, These differences have a critical impact on educational outcomes and returns13.
At the family level, factors such as parental education, family wealth, and parental support have been extensively studied. Adolescents from lower socioeconomic backgrounds are more likely to use ICT for recreational purposes, whereas those from upper-middle-class families tend to utilize ICT for personal development and have greater access to digital infrastructure14,15. Parental education and family wealth are key determinants of the digital divide16. parental education, are strongly correlated with adolescents’ technology use habits17, Students with higher levels of parental education tend to exhibit greater confidence in using digital devices and the internet to complete tasks18. Furthermore, families with higher cultural capital are more likely to use the internet for activities such as information gathering, education, and training, leveraging its potential to support learning and work19. Parental support also plays a crucial role in facilitating effective online learning experiences20.
At the school level, infrastructure configuration and teachers’ digital literacy are critical influencing factors. Studies reveal that limitations in facilities and technology hinder the progress of digital transformation in education, while the successful implementation of digital teaching and the development of students’ digital competencies heavily depend on teachers’ digital pedagogical skills21. Moreover, teachers’ educational backgrounds influence their digital literacy22, with those lacking adequate qualifications often struggling to meet the demands of digital education due to insufficient technological proficiency.
At the individual level, factors such as gender, obesity level, physical health, body image perception, and academic performance significantly influence students’ participation in physical activities. Research indicates a gender disparity in physical activity behaviors, with boys exhibiting higher motivation for physical exercise compared to girls23. The relationship between obesity and physical activity participation is complex. While some studies suggest that adolescents with normal weight engage in higher levels of physical activity than their obese peers24,25, others find no significant impact of obesity on physical activity participation rates among school-aged children26. Self-assessed physical health and positive body image perception are positively associated with increased physical activity participation among secondary school students27. The relationship between academic performance and physical activity remains inconclusive. Castro’s research found a positive correlation between physical activity participation and higher academic performance28, whereas Daley et al. reported no significant association29.
At the family level, socioeconomic status and parental attitudes toward physical activity play crucial roles. Studies show that 15-year-old adolescents from higher socioeconomic backgrounds are significantly more likely to participate in organized individual sports than those from lower socioeconomic families30. Additionally, parents serve as role models for their children, and their involvement in physical activities increases the likelihood of their children’s participation31.
At the school level, factors such as school type and the frequency of physical education classes are influential. Research indicates that students from private schools are more likely to participate in physical activities compared to their counterparts in public schools32. Moreover, the availability and quality of physical education programs are critical determinants of students’ physical activity engagement. Adolescents who participate in physical education classes exhibit significantly higher levels of physical activity than those who do not33.
Based on the above discussion, it is evident that students’ use of information technology or participation in physical activities is influenced by multi-level factors. However, digital sports behavior is a complex and integrated behavior. Therefore, this study utilizes data from the 2018 Program for International Student Assessment (PISA) in Hong Kong, China, to explore and identify the influencing factors and mechanisms of students’ participation in digital sports from individual, family, and school perspectives. The findings aim to provide insights for preventing and bridging the digital sports divide.
The dataset was obtained from the Programme for International Student Assessment (PISA), initiated by the Organisation for Economic Co-operation and Development (OECD). Since 2000, the OECD has used questionnaires to assess various background factors at the student, school, and educational system levels. The data used in this study comes from the 2018 PISA survey conducted in Hong Kong, China, which included 6,037 participants. The 2018 PISA cycle is the only version to date that investigates students’ use of digital devices to learn sport. Not only that, but in 2018 digital sport is considered to have moved into the 2.0 era, a key period in the development of digital sport, which is having a significant effect on sports participation. The use of this dataset therefore holds significant research importance and reference value34. Regions participating in PISA 2018 from China included mainland China (B-S-J-Z, representing Beijing, Shanghai, Jiangsu, and Zhejiang), Taipei, Hong Kong, and Macao. However, the B-S-J-Z region, Macao, and Taipei exhibited significant missing data in the ICT questionnaire, well-being questionnaire, and parent questionnaire, whereas the Hong Kong sample provided more complete responses, making it suitable for this study. For the data cleaning portion, 549 samples with missing outcomeing variables (n = 549) were first removed, and 26 samples with a percentage of missing independent variables greater than 20% (n = 563) were removed. Missing values for continuous variables were filled in using the median, while categorical variables were filled in using the plurality. Ultimately, 4925 valid samples were included in the analysis. During the sampling process, a fixed random seed (175) was set to ensure the reproducibility of the results.
The dependent variable in this study is whether students use digital technology during sports participation. Two items from the PISA ICT familiarity questionnaire are directly related to this variable: IC151Q07HA (“Time spent using digital devices outside of classroom lessons in a typical school week: Sports”) and IC150Q07HA (“Time spent using digital devices during classroom lessons in a typical school week: Sports”). These two items were combined to construct a binary variable, “Digital Sports Participation,” where 0 indicates no use of digital technology for sports participation (0 min per week both inside and outside the classroom) and 1 indicates the use of digital technology for sports participation (1–60 min or more per week).
Based on the above analysis, 26 independent variables were selected from the individual, family, and school levels. These include gender, stage of secondary school, academic performance, BMI, self-assessed physical health, body image, interest in ICT, perceived ICT competence, ICT autonomy, ICT Social Perception, mother’s education, father’s highest education level, ISEI of mother, ISEI of father, parents’ emotional support index, parents’ voluntary participation in sports, Home cultural possessions, home educational resources, family wealth, ICT available at home, ICT available at school, whether the school regularly discusses digital education with staff, the level of digital teaching skills training provided to teachers by the school, school type, and the Weekly Number of Physical Education Days. The coding and corresponding questionnaire items for these variables are detailed in Table 1.
All data analyses were conducted using Stata and Python software. Variables with P > 0.05 were first excluded through univariate significance analysis. Subsequently, multilevel logistic regression35 was employed to select variables for inclusion in the machine learning models. This method allows for the analysis of independent and interactive effects of Level 1 (student and family variables) and Level 2 (school variables) predictors on secondary school students’ Digital Sports behavior. The study systematically implemented a four-stage hierarchical modeling procedure consistent with previous research36: the Null Model assessed cluster dependency to justify the necessity of multilevel analysis; Model 1 specifically examined the independent effects of school-level predictors while controlling for individual/family factors; Model 2 integrated cross-level predictors, quantifying the direct effects and explanatory contributions of variables at both levels through a multivariate framework. Non-significant predictors (P < 0.05) were progressively eliminated to derive the final Model 3, which included only significant variables. Variables selected through multilevel logistic regression were used to construct optimal machine learning models, including Gaussian Naive Bayes (GNB), Gradient Boosting Decision Tree (GBDT), K-Nearest Neighbors (KNN), and Logistic Regression (LR). Hyperparameter optimization was performed using GridSearchCV with 5-fold cross-validation, and model performance was validated on an independent test set. Additionally, an ensemble model was constructed for multi-model comparison, with the predictive performance comprehensively evaluated using the area under the ROC curve (AUC), supplemented by metrics such as accuracy, precision, recall, and F1 score37. All statistical analyses were conducted using two-tailed tests, with the significance level set at (:alpha:)=0.05.
Given that machine learning models are inherently “black-box” in nature, it is challenging to intuitively interpret the importance and direction of influencing factors on secondary school students’ digital sports participation. To address this, the SHAP (SHapley Additive exPlanations) value algorithm, proposed by Lundberg and Lee38, was employed. Based on the concept of Shapley values from cooperative game theory, SHAP quantifies the contribution of each variable to the prediction outcome, thereby providing an additive explanation for black-box machine learning models. In this study, students’ digital sports participation is viewed as the result of the combined effects of multiple variables, and SHAP values are calculated using an additive feature attribution method to measure the contribution and magnitude of each independent variable.
In summary, the sample flow diagram is presented in Fig. 1.
Research flowchart.
Detailed results are presented in Table 2. The participation of secondary school students in Digital Sports has resulted in two distinct groups: participants (n = 1007) and non-participants (n = 3918). Statistically significant differences (P < 0.05) were observed between the two groups in the following variables: gender, stage of secondary school, academic performance, self-assessed physical health, interest in ICT, perceived ICT competence, ICT Social Perception, mother’s education, father’s education, parents’ voluntary participation in sports, family wealth, ICT available at home, ICT available at school, Digital Teaching Training Level for Teachers, index proportion of all teachers with ISCED5A Master’s qualifications, and Weekly Number of Physical Education Days. No statistically significant differences (P > 0.05) were found for the following variables: BMI, body image, ICT autonomy, ISEI of mother, ISEI of father, parents’ emotional support, Home cultural possessions, home educational resources, whether School Digital Teaching Discussions with Staff, and school type.
16 variables with P < 0.05 from univariate analysis were included in the multilevel logistic regression model, with Table 3 presenting the results. The null model revealed significant between-school random variation in secondary school students’ likelihood of Digital Sports participation [σ_u2 = 0.26 (0.05), P < 0.001]. The ICC value indicated that 7.4% of the variance in Digital Sports behavior probability was attributable to school-level factors, justifying the use of multilevel modeling. Model 1 demonstrated significant positive correlations of weekly physical education class days (P < 0.001) and school ICT resources (P < 0.001) with Digital Sports engagement, accounting for 24.98% of between-school variance. However, neither digital training for teachers (P = 0.275) nor the proportion of teachers with master’s degrees (P = 0.172) showed significant associations. Model 2 identified significant student- and family-level predictors after adjusting for school-level variables (weekly physical education class days and school ICT resources). Even after controlling for student and family variables, weekly physical education class days (AOR 1.30, 95% CI 1.16–1.46) and school ICT resources (AOR 1.06, 95% CI 1.02–1.10) remained significantly positively correlated with Digital Sports participation. The final Model 3 highlighted household ICT resources (AOR 1.08, 95% CI 1.04–1.13) and students’ ICT social perception (AOR 1.18, 95% CI 1.07–1.30) as significant positive correlates, whereas academic performance exhibited a significant negative correlation (AOR 0.995, 95% CI 0.994–0.996).
The five variables with P < 0.05 in the multilevel logistic regression were selected as input variables for model construction. After data standardization, the dataset was divided into training and test sets at a ratio of 85% (n = 4186) to 15% (n = 739). The grid search method exhaustively evaluates all possible combinations within specified parameter ranges to identify the globally optimal parameter configuration, thereby significantly enhancing the model’s generalization ability and computational efficiency. To optimize model performance, GridSearchCV combined with 5-fold cross-validation was employed for hyperparameter tuning of K-nearest neighbors (KNN), Gaussian naive Bayes (GNB), gradient boosting decision tree (GBDT), and logistic regression (LR). The optimized key parameters for each model are presented in Table 4, while non-critical parameters were set to their default values.
As shown in Table 5, the AUC values of GNB, GBDT, KNN, and LR models on the test set were 0.700, 0.641, 0.707, and 0.717, respectively. In this study, the AUC values of GNB, KNN, and LR models all exceeded 0.7, indicating their good discriminative performance in predicting secondary school students’ digital sports participation. Figure 2 presents the receiver operating characteristic (ROC) curves of the four prediction models, where the x-axis represents the false positive rate (FPR) and the y-axis denotes the true positive rate (TPR). Points closer to the top-left corner indicate higher model accuracy. The area under the ROC curve (AUC) reflects the predictive performance of the models, with larger AUC values indicating higher prediction accuracy. Among the primary models, the LR model performed best; however, as logistic regression is a linear model, it may fail to capture the nonlinear relationships between variables and the outcome. Therefore, this study constructed an ensemble stacking model using the LR model as the meta-model and GNB and KNN models as base models to enhance the interpretability of the conclusions. As shown in Table 5, the stacking model achieved an AUC of 0.723, outperforming the LR model and demonstrating better performance in terms of the F1 score.
Working feature curve of the test set using a machine learning model.
After completing the model development, explainability analysis was conducted on the best-performing Stacking model using the SHAP (SHapley Additive exPlanations) method. This analysis included global explanation of variable impacts and individual variable dependency explanation.
Figure 3a ranks the relative importance of the five influencing factors from high to low based on their mean absolute SHAP values: academic performance, weekly physical education class days, household ICT resources, school ICT resources, and ICT social perception. Among these, the SHAP mean values of academic performance and weekly physical education class days are significantly higher than those of other factors, indicating their importance as key determinants.
Figure 3b presents the SHAP beeswarm plot, where each point represents a sample distributed along the y-axis, and the horizontal position reflects its impact on the prediction outcome. The color gradient (red for high values, blue for low values) reveals that lower academic performance, more weekly physical education class days, richer household ICT resources, more sufficient school ICT resources, and higher ICT social perception are all significantly associated with an increased probability of secondary school students’ digital sports participation.
Importance ranking of SHAP features with swarm map.
Figure 4 presents the feature dependence plots, illustrating the threshold effects and nonlinear relationships of five variables on secondary school students’ Digital Sports participation. When the number of weekly physical education class days (0 to 5 days) is 0 or 1, the SHAP values are near the zero line on the y-axis, indicating minimal or even negative effects on students’ Digital Sports participation. However, when the number of days reaches 2 or more, the model is more likely to predict participation, and the strength of this positive effect increases with the number of days. The impacts of school ICT resources and household ICT resources (0 to 10 points) on Digital Sports participation are similar: as these resources increase, the slope changes, suggesting that the effect strengthens with higher ICT resource availability. For values below 8 points, SHAP values remain below the zero line, indicating a model prediction of non-participation; only when values exceed 8 points does the model predict participation, highlighting the high ICT resource requirements for Digital Sports behavior. The lowess curve for ICT social perception reveals a largely linear positive relationship: when social perception exceeds the student population average, SHAP values rise above the zero line, and the model predicts participation. The slope changes in academic performance indicate that when scores range from 250 to 500, the model tends to predict participation, with a stronger negative effect as scores increase; beyond 500 points, the model tends to predict non-participation, and the negative effect gradually weakens with higher scores.
SHAP dependency plots.
This study identified the influencing factors of secondary school students’ participation in Digital Sports through univariate analysis and multilevel logistic regression. Models were constructed using KNN, GBDT, GNB, and LR algorithms combined with hyperparameter optimization, and an ensemble stacking model with stronger classification performance was built. The stacking model was further analyzed using the SHAP method. The results of the data analysis revealed that secondary school students’ Digital Sports participation formed two distinct groups: participants and non-participants, with significant differences in multiple variables between the two groups. The influencing factors screened by multilevel logistic regression were ranked in order of importance using the stacking model as follows: academic performance, number of physical education classes per week, family ICT resources, school ICT resources, and ICT social perception. Among these, academic performance and the number of physical education classes per week were identified as key factors. The SHAP bar plot results were consistent with the multilevel logistic regression results, indicating strong robustness of the findings. The feature dependence plots demonstrated the threshold effects and nonlinear relationships of the five influencing factors, effectively illustrating the threshold targets that should be achieved to promote secondary school students’ participation in Digital Sports.
Academic performance has a significant negative correlation with secondary school students’ participation in Digital Sports. Previous studies have shown that digital technology can enhance learning outcomes across various subjects39, highlighting the close relationship between academic performance and digital device usage. Combined with Chen’s perspective, there is a strong link between academic pressure and physical exercise, as high academic pressure leads to increased sedentary behavior among high school students40. Therefore, the negative impact may be similar to factors influencing physical activity participation. The prioritization of academics over physical exercise may lead high-achieving students to focus on academic learning when using digital devices, thereby reducing their time spent on Digital Sports activities. The number of weekly physical education class days has a significant positive correlation with students’ participation in Digital Sports. Existing studies41、42 suggest that schools have great potential to increase participation rates in both in-school and out-of-school physical activities and serve as an important tool for implementing physical activity interventions among children and adolescents. Increasing the frequency of physical education classes may be an effective strategy to enhance physical activity levels among adolescents. Therefore, promoting physical education in secondary schools may influence students’ physical activity behaviors, thereby affecting their participation in Digital Sports. The dependence plot indicates that two weekly physical education classes are the critical threshold for positively influencing students’ Digital Sports participation, with the positive effect strengthening as the number of days increases.Both household ICT resources and school ICT resources have a positive impact on secondary school students’ participation in Digital Sports. The dependence plot reveals that only when the availability of ICT resources reaches a higher level can a positive effect be observed. This imposes certain requirements on the ICT infrastructure of households and schools. In addition to enhancing household and school ICT resources, lowering the barriers to access Digital Sports equipment and software is also an effective way to promote its widespread adoption. ICT social perception has a significantly positive influence on students’ participation in Digital Sports. The underlying mechanism may be that social cognition of information and communication technologies strengthens peer motivation and social support in the internet environment, thereby increasing students’ sustained motivation to participate in sports activities. These factors partially explain the disparities in secondary students’ use of digital devices for sports participation. Such differences lead to an imbalanced distribution of the benefits from rapidly developing digital sports technologies among student populations, thereby widening the digital sports divide and potentially creating inequalities in educational outcomes and returns. Consequently, these disparities warrant serious attention.
In conclusion, to enhance secondary students’ digital sports participation and bridge the digital sports divide, particular attention should be given to high-achieving students in digital sports promotion initiatives. Schools must guarantee a minimum of two physical education sessions per week while increasing the number of PE classes as much as possible, concurrently advancing the digital transformation of physical education through timely allocation of necessary digital resources. Development departments should establish intuitive, user-friendly, and efficient digital sports service platforms to facilitate equitable sharing and distribution of quality sports resources, ensuring universal student access to digital sports facilities. Additionally, they should implement interactive online sports models that enable user-to-user engagement, thereby expanding the social dimensions of digital sports.
This study, based on Urie Bronfenbrenner’s Ecological Systems Theory (EST) and the available data from the PISA 2018 dataset, primarily examined the influence of individual objective conditions and microsystem-level factors (e.g., family and school) on sports participation. However, due to the limitations of the dataset, this study has certain constraints. First, psychological and behavioral factors (e.g., self-efficacy, habits) have been demonstrated to significantly impact sports participation43, but they were not included in the analytical framework of this study. Second, although EST emphasizes the interaction of multi-level environmental factors, this study’s measurement of higher-level environmental factors (e.g., local policies) and the chronosystem is limited, particularly due to the dataset’s constraints on measuring the chronosystem. Furthermore, as the concept of Digital Sports continues to evolve and diversify, the measurement of Digital Sports in this study may not fully capture its latest developments. Future research needs to adopt more comprehensive measurements to capture this dynamic change. Future studies could further explore the influence of these complex factors by integrating additional data sources or adopting mixed-methods approaches.
The dataset used in this study is derived from publicly accessible data from the Programme for International Student Assessment (PISA). The PISA data, which is an integral part of this research, can be freely obtained from the official PISA website (https://www.oecd.org/pisa/data/).
Saheb, T. An empirical investigation of the adoption of mobile health applications: integrating big data and social media services. Health Technol. 10(5), 1063–1077 (2020).
Article Google Scholar
General Administration of Sport of China. Sports Development Plan for the Fourteenth Five-Year Plan [EB/OL]. -10-25) [2024-12-02]. (2021). https://www.sport.gov.cn/zfs/n4977/c23655706/part/23656158.pdf
Weber, M. & Roth, G. Economy and Society: an Outline of Interpretive Sociology 2nd edn (University of California Press, 1978).
Bonfadelli, H. The internet and knowledge gaps: A theoretical and empirical investigation. Eur. J. Communication. 17(1), 65–84 (2002).
Article Google Scholar
Van Dijk, J. & Hacker, K. The digital divide as a complex and dynamic phenomenon. Inform. Soc. 19(4), 315–326 (2003).
Article Google Scholar
Baerg, A. Big Data: Big Data Sport, and the digital divide. J. Sport Soc. Issues 41(1), 3–20 (2016).
Bronfenbrenner, U. The Ecology of Human Development: Experiments by Nature and Design (Harvard University Press, 1979).
Bronfenbrenner, U. Developmental ecology through space and time: A future perspective. In Examining Lives in Context: Perspectives on the Ecology of Human Development (eds Moen, P., Elder, G. H. Jr. & Lüscher, K.) 619–647 (American Psychological Association, 1995).
Chafouleas, S. M. & Iovino, E. A. Engaging a whole child, school, and community lens in positive education to advance equity in schools. Front. Psychol. 12, 758788 (2021).
Article PubMed PubMed Central Google Scholar
Purushothaman, A., Holmfeld, L. D. & Kuruvilla, M. Second order divide in internet usage in developing countries: an overlooked issue? In Gender and Diversity: Concepts, Methodologies, Tools, and Applications 38–51 (IGI Global, 2019).
OECD & Students Computers and Learning: Making the Connection 3 (OECD Publishing, 2015).
Gurung, B. & Rutledge, D. Digital learners and the overlapping of their personal and educational digital engagement. Comput. Educ. 77, 91–100 (2014).
Article Google Scholar
Hargittai, E. & Hinnant, A. Digital inequality: differences in young adults’ use of the internet. Commun. Res. 5, 602–621 (2008).
Article Google Scholar
Micheli, M. What is new in the digital divide? Understanding internet use by teenagers from different social backgrounds. In Communication and Information Technologies Annual 55–87 (Emerald Group Publishing Limited, 2015).
Reddy, A. B. & Jose, S. Of access and inclusivity: digital divide in online education. ArXiv e-prints. arXiv:2107.10723 (2021).
Maji, S. K. & Laha, A. State of digital divide and its determinants: evidence from student level survey in West Bengal, India. Int. J. Bus. Environ. 4, 392–417 (2022).
Article Google Scholar
North, S., Snyder, I. & Bulfin, S. Digital tastes: social class and young People’s technology use. Inf. Commun. Soc. 7, 895–911 (2008).
Article Google Scholar
Vekiri, I. Socioeconomic differences in elementary students’ ICT beliefs and out-of-school experiences. Comput. Educ. 4, 941–950 (2010).
Article Google Scholar
Ren, W., Zhu, X. & Yang, J. The SES-based difference of adolescents’ digital skills and usages: an explanation from family cultural capital. Comput. Educ. 177 (2022).
Liu, X., Zhao, L. & Su, Y. S. Impact of parents’ attitudes on learning ineffectiveness: the mediating role of parental self-efficacy. Int. J. Environ. Res. Public. Health. 1, 615 (2022).
Article Google Scholar
Sipilä, K. Educational use of information and communications technology: teachers’ perspective. Technol. Pedagog Educ. 23(2), 225–241 (2014).
Article Google Scholar
Rahmawati, S. et al. The distributions of vocational high school teachers’ advanced digital competence (ADC). J. Tech. Educ. Train. 14(2), 190–201 (2022).
Google Scholar
Portela-Pino, I. et al. Gender differences in motivation and barriers for the practice of physical exercise in adolescence. Int. J. Environ. Res. Public. Health. 17(1), 168 (2020).
Article Google Scholar
Butte, N. F. et al. Physical activity in nonoverweight and overweight Hispanic children and adolescents. Med. Sci. Sports Exerc. 8, 1257–1266 (2007).
Article Google Scholar
Wright, K. N. Influence of body mass index, gender, and Hispanic ethnicity on physical activity in urban children. J. Spec. Pediatr. Nurs. 16(2), 90–104 (2011).
Article PubMed PubMed Central Google Scholar
Uvacsek, M., Wágner, B. & Simkó, G. Sport and physical activity participation by weight groups in school-aged Hungarian children. Appl. Sci. 14(15) (2024).
Sabiston, C. M. et al. Body image, physical activity, and sport: A scoping review. Psychol. Sport Exerc. 42, 48–57 (2019).
Article Google Scholar
Castro-Sánchez, M. et al. Motivational climate in sport is associated with life stress levels, academic performance and physical activity engagement of adolescents. Int. J. Environ. Res. Public. Health. 16(7), 1198 (2019).
Article PubMed PubMed Central Google Scholar
Daley, A. J. & Ryan, J. Academic performance and participation in physical activity by secondary school adolescents. Percept. Mot Skills. 91(2), 531–534 (2000).
Article CAS PubMed Google Scholar
Hobza, V., Maracek, M. & Hamrik, Z. Organized sport activities of 11 to 15-year-old adolescents: trends from 2010–2018 and socioeconomic context. Int. J. Environ. Res. Public. Health. 19(10), 6074 (2022).
Article PubMed PubMed Central Google Scholar
Quinto Romani, A. Parental behaviour and children’s sports participation: evidence from a Danish longitudinal school study. Sport Educ. Soc. 25(3), 332–347 (2020).
Article Google Scholar
García, J. & Suárez, M. J. Organised and non-organised after-school physical activity among children in Spain: the role of school-related variables. Eur. Sport Manag Q. 20(2), 171–188 (2020).
Article Google Scholar
Diamant, A. L., Babey, S. H. & Wolstein, J. Adolescent physical education and physical activity in California. UCLA Cent. Health Policy Res. 3 (2011).
Miah, A. Sport 2.0: Transforming Sports for a Digital World (MIT Press, 2017).
Snijders, T. A. & Bosker, R. Multilevel Analysis: An Introduction to Basic and Advanced Multilevel Modeling (SAGE, 2011).
Leatherdale, S. T., Cameron, R., Brown, K. S. & McDonald, P. W. Senior student smoking at school, student characteristics, and smoking onset among junior students: a multilevel analysis. Prev. Med. 40(6), 853–859 (2005).
Article PubMed Google Scholar
Badillo, S. et al. An introduction to machine learning. Clin. Pharmacol. Ther. 4, 871–885 (2020).
Article Google Scholar
Lundberg, S. M. & Lee, S. I. A unified approach to interpreting model predictions. Adv. Neural Inf. Process. Syst. 30 (2017).
Sormunen, M. et al. Learning outcomes of digital learning interventions in higher education: A scoping review. Comput. Inf. Nurs. 40(3), 154–164 (2022).
Google Scholar
Chen, X. et al. Influence of learning stress on anxiety levels among high school students in China: the chain-mediated role of physical activity and mobile phone addiction. BMJ Open. 14(11), e078488 (2024).
Rezende, L. F. et al. The role of school environment in physical activity among Brazilian adolescents. PLoS One 10(6), e0131342 (2015).
Da Silva, D. J. et al. Is participation in physical education classes related to physical activity and sedentary behavior? A systematic review. J. Phys. Act. Health. 19(11), 786–808 (2022).
Article PubMed Google Scholar
Di Maio, S., Keller, J., Hohl, D. H., Schwarzer, R. & Knoll, N. Habits and self-efficacy moderate the effects of intentions and planning on physical activity. Br. J. Health Psychol. 26(1), 50–66 (2021).
Article PubMed Google Scholar
Download references
Funding for this research came from Project 23CDTXQ002 supported by the Annual Project of the Sichuan University Student Sports Association and Project 2024PE012 supported by the Fundamental Research Funds for the Central Universities of Sichuan University.
Institute of Physical Education, Sichuan University, Chengdu, 610065, China
XiaoTao Cai, Yi Xian, TongYi Liu & Qing Chen
College of Biomedical Engineering, Sichuan University, Chengdu, 610065, China
YuXin Zhou
Institute of Physical Education, Southwest Jiaotong University, Chengdu, 611756, China
HaoNan Cui
PubMed Google Scholar
PubMed Google Scholar
PubMed Google Scholar
PubMed Google Scholar
PubMed Google Scholar
PubMed Google Scholar
X.C was responsible for research design, experimental data analysis, and manuscript writing. Y.X and Y.Z were responsible for data interpretation. Q.C and H.C were responsible for data collection. T.L was responsible for research design and supervision. All authors reviewed the manuscript.
Correspondence to TongYi Liu.
The authors declare no competing interests.
Ethical review and approval was not required for the study on human participants in accordance with the local legislation and institutional requirements. Written informed consent to participate in this study was provided by the participants’ legal guardian/next of kin. This study is based on the public databases of the PISA 2018 assessment (OECD). Data collection for OECD-PISA studies is under the responsibility of the governments from the participating countries.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
Reprints and permissions
Cai, X., Xian, Y., Liu, T. et al. A study on factors influencing digital sports participation among Chinese secondary school students based on explainable machine learning. Sci Rep 15, 15657 (2025). https://doi.org/10.1038/s41598-025-00769-x
Download citation
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41598-025-00769-x
Anyone you share the following link with will be able to read this content:
Sorry, a shareable link is not currently available for this article.
Provided by the Springer Nature SharedIt content-sharing initiative
Advertisement
Scientific Reports (Sci Rep)
ISSN 2045-2322 (online)
© 2025 Springer Nature Limited
Sign up for the Nature Briefing: AI and Robotics newsletter — what matters in AI and robotics research, free to your inbox weekly.
Leave a Comment
Leave a Comment
Leave a Comment
Leave a Comment
PEEHIP says its members won’t be effected by any UAB, UnitedHealthcare changes – Alabama Daily News
As contract negotiations between the University of Alabama at Birmingham Health System and UnitedHealthcare are ongoing, state officials said Wednesday any changes won’t affect Alabama educators or retirees.
The Public Education Employees Health Insurance Plan currently uses UHC for its Medicare-eligible members.
“We are pleased to inform you that UnitedHealthcare has confirmed that PEEHIP members may continue to receive services from UAB and its affiliates at no additional cost to either members or the PEEHIP plan,” a Wednesday statement from the plan said.
“… UAB will continue to provide care regardless of whether members are considered in-network or out-of-network under UnitedHealthcare’s plan. This means there will be no disruption in services provided by UAB and its affiliates.”
UnitedHealthcare has confirmed that PEEHIP members may continue to receive services from UAB and its affiliates at no additional cost to either members or the PEEHIP plan. There will be no disruption in services provided by UAB and its affiliates.https://t.co/3QIncjhLTy
— PEEHIP (@peehip) July 2, 2025
UnitedHealthcare and UAB have until the end of the month to reach a new contract. Patients with United coverage have access to UAB sites this month as talks continue.
If an agreement is not reached by the end of the month, services for United members would be considered out-of-network at dozens of UAB-owned and affiliated sites, including the UAB hospital, UAB’s primary, specialty and urgent care clinics and Baptist Health in Montgomery.
“We are deeply disappointed that UnitedHealthcare has created this uncertainty for our patients,” said UAB Health System CEO Dawn Bulgarella said in a statement last month. “We will continue to diligently negotiate with
United in good faith to reach a reasonable agreement before July 31. Our goal is to remain a
participating provider and continue delivering the highest-quality care to the people of Alabama. We encourage patients and employers to contact United and express the importance of keeping UAB Health System entities in-network.”
United last month said UAB is one of the most expensive health systems in the Southeast, “yet they’re demanding a near 30% price hike for our employer-sponsored commercial plans as well as a significant rate increase for our Medicare Advantage plans.”
“Agreeing to UAB’s demands would significantly increase premiums and out-of-pocket costs for consumers as well as the cost of doing business for employers,” the insurer said.
A full list of UAB facilities that would be considered out-of-network if the contract ends is available here.
The State Employees Insurance Board does not currently use UnitedHealthcare for its coverage.
1228 Magnolia Curve,
Montgomery AL, 36106
[email protected]
Copyright © 2025 Alabama Daily News
Web Development By Infomedia
Leave a Comment
Leave a Comment
One Actionable Way to Offset Your Diabetes Risk—Even If It’s In Your Genes – MindBodyGreen
Account
If you needed one more reason to hit the weights, here it is: new research from the UK Biobank shows that higher muscle strength is strongly linked to a lower risk of type 2 diabetes—even in people genetically predisposed to the disease.
The study analyzed data from over 140,000 British adults who were diabetes-free at the start of the study. Over the course of 7.4 years, researchers tracked their health, focusing on two main things: grip strength (as a proxy for overall muscle strength) and their genetic risk score for type 2 diabetes, based on 138 known gene variants.
What they found is compelling: people with high muscle strength had a 44% lower relative risk of developing type 2 diabetes compared to those with low muscle strength. That’s after adjusting for genetic risk and other factors.
But the most surprising part? Even individuals with high genetic risk had lower absolute risk if they were strong. In other words, building muscle might be able to buffer some of the impact of your DNA.
Muscle tissue plays a key role in how our bodies handle glucose. When you build muscle, you increase your body’s ability to take up and store blood sugar efficiently, improving insulin sensitivity. Strength training also boosts levels of GLUT4 (a glucose transporter) and mitochondrial function in skeletal muscle, both of which help keep blood sugar in check.
Loss of muscle mass and strength, on the other hand, shrinks this metabolic machinery, making it harder for your body to manage glucose and potentially leading to insulin resistance.
That’s one reason why maintaining (or better yet, increasing) muscle strength is so critical as we age.
Interestingly, the researchers found that while higher muscle strength helped across the board, the protective effect was slightly weaker in people with high genetic risk. But even then, the data showed that these high-risk individuals with strong muscles had a lower 8-year absolute risk of developing diabetes than low- or medium-risk individuals with weak muscles.
This suggests that building muscle may be a powerful way to counteract inherited risk and that lifestyle choices can make a meaningful difference, even when your DNA stacks the odds against you.
Your genes may shape your risk for type 2 diabetes, but they don’t get the final say. This study adds to growing evidence that building muscle can play a meaningful role in protecting your metabolic health, even if you’re genetically predisposed. The good news? Muscle strength is something you can actively improve.
Grip strength, the measure used in the study, responds well to regular resistance training and functional movement—think lifting weights, carrying groceries, or doing pushups. Whether you’re managing a known risk or simply aiming to stay strong as you age, prioritizing strength could be one of the most powerful things you do for your long-term health.
*These statements have not been evaluated by the Food and Drug Administration. This product is not intended to diagnose, treat, cure or prevent any disease.