Autonomous Artificial Intelligence Increases Access and Health Equity in Underserved Populations with Diabetes.

Diabetic eye disease (DED) is a leading cause of blindness in the world. Early detection and treatment of DED have been shown to be both sight-saving and cost-effective. As such, annual testing for DED is recommended for adults with diabetes and is a Healthcare Effectiveness Data and Information Set (HEDIS) measure. However, adherence to this guideline has historically been low, and access to this sight-saving intervention has particularly been limited for specific populations, such as Black or African American patients. In 2018, the US Food and Drug Agency (FDA) De Novo cleared autonomous artificial intelligence (AI) for diagnosing DED in a primary care setting. In 2020, Johns Hopkins Medicine (JHM), an integrated healthcare system with over 30 primary care sites, began deploying autonomous AI for DED testing in some of its primary care clinics. In this retrospective study, we aimed to determine whether autonomous AI implementation was associated with increased adherence to annual DED testing, and whether this was different for specific populations. JHM primary care sites were categorized as "non-AI" sites (sites with no autonomous AI deployment over the study period and where patients are referred to eyecare for DED testing) or "AI-switched" sites (sites that did not have autonomous AI testing in 2019 but did by 2021). We conducted a difference-in-difference analysis using a logistic regression model to compare change in adherence rates from 2019 to 2021 between non-AI and AI-switched sites. Our study included all adult patients with diabetes managed within our health system (17,674 patients for the 2019 cohort and 17,590 patients for the 2021 cohort) and has three major findings. First, after controlling for a wide range of potential confounders, our regression analysis demonstrated that the odds ratio of adherence at AI-switched sites was 36% higher than that of non-AI sites, suggesting that there was a higher increase in DED testing between 2019 and 2021 at AI-switched sites than at non-AI sites. Second, our data suggested autonomous AI improved access for historically disadvantaged populations. The adherence rate for Black/African Americans increased by 11.9% within AI-switched sites whereas it decreased by 1.2% within non-AI sites over the same time frame. Third, the data suggest that autonomous AI improved health equity by closing care gaps. For example, in 2019, a large adherence rate gap existed between Asian Americans and Black/African Americans (61.1% vs. 45.5%). This 15.6% gap shrank to 3.5% by 2021. In summary, our real-world deployment results in a large integrated healthcare system suggest that autonomous AI improves adherence to a HEDIS measure, patient access, and health equity for patients with diabetes - particularly in historically disadvantaged patient groups. While our findings are encouraging, they will need to be replicated and validated in a prospective manner across more diverse settings.

[Application and comparison of three occupational health risk assessment methods in an automobile manufacturing industry].

Objective: Three occupational health risk assessment methods were used to assess the occupational health risk of noise exposed posts in an automobile manufacturing enterprise. According to the results, the selection of risk assessment methods and risk management of such occupational noise enterprises were provided. Methods: Form April to November 2021, The occupational health field survey was carried out in an automobile manufacturing industry in Tianjin. The occupational health MES risk assessment method, occupational health risk index risk assessment method and Australian occupational hazard risk assessment method were used to evaluate the occupational health risk of noise-exposed posts in this enterprise, and the evaluation results of different methods were analyzed and compared. Results: The average value of L(Aeq, 8 h) in the four workshops of automobile manufacturing industry was 82.95 dB (A) , and the noise detection exceeding rate was 22.41% (26/116) . The LAeq, 8h and exceeding rate noise of welding workshop were higher than those of other workshops (χ(2)=23.56, 32.94, P<0.01) . The three occupational health risk assessment methods have the same risk assessment results for the four major workshops. The assembly and painting workshops are level 4 risk (possible risk) , and the stamping and welding workshops are level 3 risk (significant risk) . Conclusion: Occupational noise has certain potential hazards to workers in automobile manufacturing enterprises. Therefore, in the future work, corresponding organizational management measures should be taken to improve the working environment and reduce the actual exposure level of workers in order to protect the health of occupational workers.

Social Determinants of Health and Impact on Screening, Prevalence, and Management of Diabetic Retinopathy in Adults: A Narrative Review.

Diabetic retinal disease (DRD) is the leading cause of blindness among working-aged individuals with diabetes. In the United States, underserved and minority populations are disproportionately affected by diabetic retinopathy and other diabetes-related health outcomes. In this narrative review, we describe racial disparities in the prevalence and screening of diabetic retinopathy, as well as the wide-range of disparities associated with social determinants of health (SDOH), which include socioeconomic status, geography, health-care access, and education.

Quality assessment of clinical practice guidelines for the management of wheezing disorders in children.

BACKGROUND: The quality of available clinical practice guidelines (CPGs) for childhood wheezing disorders have not been systematically evaluated.METHODS: CPGs were systematically evaluated by four independent reviewers using Appraisal of Guidelines Research and Evaluation (AGREE) II instrument and the Reporting Items for Practice Guidelines in HealTHcare (RIGHT) checklist. We calculated the overall agreement among reviewers with the intraclass correlation coefficient (ICC).RESULTS: A total of 35 CPGs published between January 2000 and December 2020 were evaluated. The overall agreement among reviewers was good (ICC 0.85, 95% CI 0.83-0.87). The average CPGs score was 42% (range: 25-79). The mean scores of four domains were low: 37% for Stakeholder Involvement (range: 10-85), 28% for Rigour of Development (range: 42-81), 35% for Applicability (range: 11-73) and 24% for Editorial Independence (range: 0-83). The mean reporting rate of the RIGHT checklist was 31%. The Basic Information domain had the highest reporting rate (65%); the Review and Quality Assurance domain had the lowest rate (3%).CONCLUSIONS: The quality of the CPGs was poor. Greater efforts are needed to improve quality in domains to provide high-quality guidelines that can be used as reliable tools for clinical decision-making.

[Health hazards and hearing loss risk assessment of workers exposed to noise in an automobile manufacturing enterprise].

Objective: To investigate the current situation of occupational exposure to noise among noise workers in an automobile manufacturing enterprise in Tianjin, understand the impact of noise on workers＇ nervous system and hearing, and assess the risk of hearing loss among noise workers. Methods: In May 2021, 3516 workers in an automobile manufacturing enterprise were investigated by using a self-made questionnaire＂Noise Workers Questionnaire＂ and cluster sampling method. The occupational noise hygiene survey and occupational hazards detection were carried out in their workplaces. They were divided into noise exposure group and non-noise exposure group according to whether they were exposed to noise or not. The general characteristics, hearing and nervous system symptoms of the two groups of workers were compared, and the risk of hearing loss was assessed. Results: There were 758 workers in the noise exposure group, aged (26±5) years old, with a working age of 3.0 (2.0, 6.0) years exposed to noise. 2758 workers in the non-noise exposure group, aged (25±6) years old, with a working age of 2.0 (1.0, 4.0) years. There were statistically significant differences in the distribution of workers＇education level, working age and memory loss between the two groups (χ(2)=37.98, 38.70, 5.20, P<0.05). The workers in the noise exposure group showed a decreasing trend of insomnia, dreaminess, sweating and fatigue with the increase of working age (χ(2trend)=6.16, 7.99, P<0.05). The risk classification of binaural high-frequency hearing loss for workers in all noise positions until the age of 50 and 60 was negligible, the risk of occupational noise deafness was low for workers in stamping and welding noise positions until the age of 60. Conclusion: The occupational noise exposed to automobile manufacturing workers may cause certain harm to their nervous and auditory systems. Noise protection measures should be taken to reduce the risk of hearing loss and occupational noise deafness.

[Current status of thromboembolism risk assessment in patients hospitalized with non-valvular atrial fibrillation in tertiary hospitals in China].

Objective: To analyze the current status, trend and predictors of thromboembolism risk assessment in patients hospitalized with non-valvular atrial fibrillation (NVAF) in tertiary hospitals in China. Methods: The study was based on data from the Improving Care for Cardiovascular disease in China (CCC)-Atrial Fibrillation (AF) project. About 10% of the tertiary hospitals in each geographic-economic stratum were recruited. Participating hospitals reported the first 10 to 20 patients with a discharge diagnosis of atrial fibrillation monthly. From February 2015 to December 2019, a total of 49 104 NVAF patients from 151 tertiary hospitals in 30 provinces, municipalities and autonomous regions were enrolled. Clinical data of the patients was collected. The proportion of NVAF patients receiving thromboembolism risk assessment, variations in the proportion between different hospitals, the time trend of the application of thromboembolism risk assessment, and the predictors of the application of thromboembolism risk assessment were analyzed. Results: The age of the NVAF patients was (68.7±12.1) years, 27 709 patients (56.4%) were male. Only 17 251 patients (35.1%) received thromboembolism risk assessment. The proportion varied substantially between hospitals with the lowest value of 0 and the highest value of 100%. Among the hospitals, which enrolled more than 30 patients, no patients received thromboembolism risk assessment in 18.4% (26/141) of the hospitals, more than 50% of the patients received thromboembolism risk assessment in 21.3% (30/141) of the hospitals, and all the patients received thromboembolism risk assessment in only 1 hospital. The proportion of NVAF patients receiving thromboembolism risk assessment was 16.2% (220/1 362) in the first quarter of 2015, and significantly increased to 67.1% (1 054/1 572) in the last quarter of 2019 (P<0.001). Patients' characteristics were associated with the application of thromboembolism risk assessment. The odds of receiving thromboembolism risk assessment was lower in male patients compared to female patients(OR=0.94,95%CI 0.89-0.99), lower in patients with acute coronary syndrome or other cardiovascular diseases compared to those with AF as the primary admission reason (OR=0.59, 95%CI 0.55-0.63, OR=0.52, 95%CI 0.45-0.61, respectively), and lower in patients with paroxysmal, persistent and long-standing/permanent AF compared to those with first detected AF (OR=0.62, 95%CI 0.57-0.67, OR=0.72, 95%CI 0.66-0.79, OR=0.57, 95%CI 0.52-0.64, respectively). The odds was higher in patients with a history of hypertension, heart failure, stroke/TIA, and previous anticoagulant therapy compared to those without the above conditions (OR=1.17, 95%CI 1.11-1.23, OR=1.18, 95%CI 1.07-1.30, OR=1.17, 95%CI 1.08-1.27, OR=1.28, 95%CI 1.19-1.37, respectively) (P all<0.05). Conclusion: Thromboembolism risk assessment was underused in patients hospitalized with NVAF in tertiary hospitals in China, and there were substantial variations between hospitals in the application of thromboembolism risk assessment. The application of thromboembolism risk assessment in tertiary hospitals has been improved in recent years, but there is still plenty of room for future improvement. Patients' characteristics could affect the application of thromboembolism risk assessment in China.

The SEE Study: Safety, Efficacy, and Equity of Implementing Autonomous Artificial Intelligence for Diagnosing Diabetic Retinopathy in Youth.

OBJECTIVE: Diabetic retinopathy (DR) is a leading cause of vision loss worldwide. Screening for DR is recommended in children and adolescents, but adherence is poor. Recently, autonomous artificial intelligence (AI) systems have been developed for early detection of DR and have been included in the American Diabetes Association's guidelines for screening in adults. We sought to determine the diagnostic efficacy of autonomous AI for the diabetic eye exam in youth with diabetes. RESEARCH DESIGN AND METHODS: In this prospective study, point-of-care diabetic eye exam was implemented using a nonmydriatic fundus camera with an autonomous AI system for detection of DR in a multidisciplinary pediatric diabetes center. Sensitivity, specificity, and diagnosability of AI was compared with consensus grading by retinal specialists, who were masked to AI output. Adherence to screening guidelines was measured before and after AI implementation. RESULTS: Three hundred ten youth with diabetes aged 5-21 years were included, of whom 4.2% had DR. Diagnosability of AI was 97.5% (302 of 310). The sensitivity and specificity of AI to detect more-than-mild DR was 85.7% (95% CI 42.1-99.6%) and 79.3% (74.3-83.8%), respectively, compared with the reference standard as defined by retina specialists. Adherence improved from 49% to 95% after AI implementation. CONCLUSIONS: Use of a nonmydriatic fundus camera with autonomous AI was safe and effective for the diabetic eye exam in youth in our study. Adherence to screening guidelines improved with AI implementation. As the prevalence of diabetes increases in youth and adherence to screening guidelines remains suboptimal, effective strategies for diabetic eye exams in this population are needed.

[Comparison of Three CNN Models Applied in Bone Age Assessment of Pelvic Radiographs of Adolescents].

ABSTRACT: Objective To compare the performance of three deep-learning models （VGG19, Inception-V3 and Inception-ResNet-V2） in automatic bone age assessment based on pelvic X-ray radiographs. Methods A total of 962 pelvic X ray radiographs taken from adolescents （481 males, 481 females） aged from 11.0 to 21.0 years in five provinces and cities of China were collected, preprocessed and used as objects of study. Eighty percent of these X ray radiographs were divided into training set and validation set with random sampling method and used for model fitting and hyper-parameters adjustment. Twenty percent were used as test sets, to evaluate the ability of model generalization. The performances of the three models were assessed by comparing the root mean square error （RMSE）, mean absolute error （MAE） and Bland-Altman plots between the model estimates and the chronological ages. Results The mean RMSE and MAE between bone age estimates of the VGG19 model and the chronological ages were 1.29 and 1.02 years, respectively. The mean RMSE and MAE between bone age estimates of the Inception-V3 model and the chronological ages were 1.17 and 0.82 years, respectively. The mean RMSE and MAE between bone age estimates of the Inception-ResNet-V2 model and the chronological ages were 1.11 and 0.84 years, respectively. The Bland-Altman plots showed that the mean value of differences between bone age estimates of Inception-ResNet-V2 model and the chronological ages was the lowest. Conclusion In the automatic bone age assessment of adolescent pelvis, the Inception-ResNet-V2 model performs the best while the Inception-V3 model achieves a similar accuracy as VGG19 model.

[Risk assessment and early warning of imported COVID-19 in Guangdong province].

Objective: To assess the imported risk of COVID-19 in Guangdong province and its cities, and conduct early warning. Methods: Data of reported COVID-19 cases and Baidu Migration Index of 21 cities in Guangdong province and other provinces of China as of February 25, 2020 were collected. The imported risk index of each city in Guangdong province were calculated, and then correlation analysis was performed between reported cases and the imported risk index to identify lag time. Finally, we classified the early warming levels of epidemic by imported risk index. Results: A total of 1 347 confirmed cases were reported in Guangdong province, and 90.0% of the cases were clustered in the Pearl River Delta region. The average daily imported risk index of Guangdong was 44.03. Among the imported risk sources of each city, the highest risk of almost all cities came from Hubei province, except for Zhanjiang from Hainan province. In addition, the neighboring provinces of Guangdong province also had a greater impact. The correlation between the imported risk index with a lag of 4 days and the daily reported cases was the strongest (correlation coefficient: 0.73). The early warning base on cumulative 4-day risk of each city showed that Dongguan, Shenzhen, Zhongshan, Guangzhou, Foshan and Huizhou have high imported risks in the next 4 days, with imported risk indexes of 38.85, 21.59, 11.67, 11.25, 6.19 and 5.92, and the highest risk still comes from Hubei province. Conclusions: Cities with a large number of migrants in Guangdong province have a higher risk of import. Hubei province and neighboring provinces in Guangdong province are the main source of the imported risk. Each city must strengthen the health management of migrants in high-risk provinces and reduce the imported risk of Guangdong province.

Technical Note: The development of a multi-physics simulation tool to estimate the background dose by systemic targeted alpha therapy.

PURPOSE: To predict biological effects of targeted alpha therapy (TAT) in preclinical studies, dosimetry calculations based on the micro-level distributions of emitters are essential. Due to the saturation of the tumor antigenic sites and bonding breaks by decay, some of Alpha-immuno-conjugate and decay daughters may inevitably be transported by convection and diffusion along with blood or lymphatic circulation. This results in highly nonuniform and unsteady distributions of irradiation sources. Since the micro-level distribution of emitters cannot be measured and obtained in patients with current technology, a modeling toolset to give more insight of the internal dose could be an alternative. METHODS: A multi-physics model based on a Monte Carlo microdosimetry technique and computational fluid dynamics (CFD) modeling was developed and applied to multiple internal irradiation sources. The CFD model tracks the path of the radionuclides and the dose model is capable of evaluating the time-dependent absorbed dose to the target. RESULTS: The conceptual model is capable of handling complex nonuniform irradiation sources in vasculature. The results from the simulations indicate that the assumption of homogeneous and motionless distribution of the administered activity used in the conventional dose calculation tends to significantly underestimate or overestimate the absorbed dose to the vascular system in various scenarios. CONCLUSION: Modeling the in vivo transport of radionuclides has the potential to improve the accuracy of TAT dose estimates. It could be the first step to develop a simulation tool set for assessing absorbed dose to tumor or normal tissues and predict the corresponding biological responses in the future.

[Risk assessment of exported risk of COVID-19 from Hubei Province].

Objective: To evaluate the exported risk of COVID-19 from Hubei Province and the imported risk in various provinces across China. Methods: Data of reported COVID-19 cases and Baidu Migration Indexin all provinces of the country as of February 14, 2020 were collected. The correlation analysis between cumulative number of reported cases and the migration index from Hubei was performed, and the imported risks from Hubei to different provinces across China were further evaluated. Results: A total of 49 970 confirmed cases were reported nationwide, of which 37 884 were in Hubei Province. The average daily migration index from Hubei to other provinces was 312.09, Wuhan and other cities in Hubei were 117.95 and 194.16, respectively. The cumulative COVID-19 cases of provinces was positively correlated with the migration index derived from Hubei Province, also in Wuhan and other cities in Hubei, with correlation coefficients of 0.84, 0.84, and 0.81. In linear model, population migration from Hubei Province, Wuhan and other cities in Hubei account for 71.2%, 70.1%, and 66.3% of the variation, respectively. The period of high exported risk from Hubei occurred before January 27, of which the risks before January 23 mainly came from Wuhan, and then mainly from other cities in Hubei. Hunan Province, Henan Province and Guangdong Province ranked the top three in terms of cumulative imported risk (the cumulative risk indices were 58.61, 54.75 and 49.62 respectively). Conclusion: The epidemic in each province was mainly caused by the importation of Hubei Province. Taking measures such as restricting the migration of population in Hubei Province and strengthening quarantine measures for immigrants from Hubei Province may greatly reduce the risk of continued spread of the epidemic.

Estimating the economic impact of pneumococcal conjugate, Haemophilus influenzae type b and rotavirus vaccines in India: National and state-level analyses.

BACKGROUND: To support vaccine decision-making we estimated from the societal perspective the potential health impact and costs averted through immunization with three vaccines - Haemophilus influenzae type b (Hib), pneumococcal conjugate vaccine (PCV) and rotavirus vaccine (RVV). METHODS: Based on variability in disease burden, strength of health system and economic status, we selected four states in India: Bihar, New Delhi, Maharashtra and Tamil Nadu. We used secondary data sources to estimate the number of under-5 deaths averted from Hib, pneumococcus and rotavirus in each state and back-calculated the total cases averted. We synthesized available data to estimate the disease burden, treatment cost, caretaker productivity loss and vaccine coverage in each state. A Delphi Survey and roundtable among Indian experts was conducted to reach consensus on model inputs. RESULTS: By scaling up coverage of Hib, PCV and RVV, India could save over US$1 billion (uncertainty range: US$0.9-US$2.4 billion) in economic benefits and avert more than 90,000 needless child deaths each year. An estimated US$1 billion (US$0.9-US$2 billion) or 88% of the total amount of cost savings would be attributable to lost productivity due to premature pneumococcal death. Another US$112.8 million (US$105-297 million), or 10% of the total cost would be accounted by costs related to loss of productivity due to disability as a result of these diseases. Treatment costs of Hib, pneumococcal disease and rotavirus gastroenteritis, would account for US$8.4 million (US$4-12 million) or <1% of the total costs of these diseases. Finally, caretaker productivity loss from seeking care would represent US$1.5 million (US$ 1-4.9 million). Cost savings varied by vaccine, coverage scenarios and states. CONCLUSIONS: Hib, PCV and RVV vaccine introduction in India can result in immediate benefits to the government and households in terms of savings.

Methodological approaches to situational analysis in global mental health: a scoping review.

Global inequity in access to and availability of essential mental health services is well recognized. The mental health treatment gap is approximately 50% in all countries, with up to 90% of people in the lowest-income countries lacking access to required mental health services. Increased investment in global mental health (GMH) has increased innovation in mental health service delivery in LMICs. Situational analyses in areas where mental health services and systems are poorly developed and resourced are essential when planning for research and implementation, however, little guidance is available to inform methodological approaches to conducting these types of studies. This scoping review provides an analysis of methodological approaches to situational analysis in GMH, including an assessment of the extent to which situational analyses include equity in study designs. It is intended as a resource that identifies current gaps and areas for future development in GMH. Formative research, including situational analysis, is an essential first step in conducting robust implementation research, an essential area of study in GMH that will help to promote improved availability of, access to and reach of mental health services for people living with mental illness in low- and middle-income countries (LMICs). While strong leadership in this field exists, there remain significant opportunities for enhanced research representing different LMICs and regions.

[The definition of heat-wave based on mortality risk assessment in different regions of China].

Objective: To identify the definition of heat wave based on mortality risk assessment in different regions of China. Methods: Daily mortality (from China Information System for Disease Control and Prevention) and meteorological data (from National Meteorological Information Center in China) from 66 counties with a population of over 200 000 were collected from 2006-2011. With the consideration of climate type and administrative division, China was classified as seven regions. Firstly, distributed lag non-linear model (DLNM) was used to estimate community-specific effects of temperature on non-accidental mortality. Secondly, a multivariate meta-analysis was applied to pool the estimates of community-specific effects to explore the region-specific temperature threshold and the duration for definition of heat wave. Results: We defined regional heat wave of Northeast, North, Northwest, East, Central and Southwest China as being two or more consecutive days with daily mean temperature higher than or equal to the P(64), P(71), P(85), P(67), P(75) and P(77) of warm season (May to October) temperature, respectively, while the thresholds of temperature were 21.6, 23.7, 24.3, 25.7, 28.0 and 25.3 ℃. The heat wave in South China was defined as five or more consecutive days with daily mean temperature higher than or equal to the P(93) (30.4 ℃) of warm season (May to October) temperature. Conclusion: The region-specific definition of heat wave developed in our study may provide local government with the guidance of establishment and implementation of early heat-health response systems to address the negative health outcomes due to heat wave.

Assessment of Deep Generative Models for High-Resolution Synthetic Retinal Image Generation of Age-Related Macular Degeneration.

Importance: Deep learning (DL) used for discriminative tasks in ophthalmology, such as diagnosing diabetic retinopathy or age-related macular degeneration (AMD), requires large image data sets graded by human experts to train deep convolutional neural networks (DCNNs). In contrast, generative DL techniques could synthesize large new data sets of artificial retina images with different stages of AMD. Such images could enhance existing data sets of common and rare ophthalmic diseases without concern for personally identifying information to assist medical education of students, residents, and retinal specialists, as well as for training new DL diagnostic models for which extensive data sets from large clinical trials of expertly graded images may not exist. Objective: To develop DL techniques for synthesizing high-resolution realistic fundus images serving as proxy data sets for use by retinal specialists and DL machines. Design, Setting, and Participants: Generative adversarial networks were trained on 133 821 color fundus images from 4613 study participants from the Age-Related Eye Disease Study (AREDS), generating synthetic fundus images with and without AMD. We compared retinal specialists' ability to diagnose AMD on both real and synthetic images, asking them to assess image gradability and testing their ability to discern real from synthetic images. The performance of AMD diagnostic DCNNs (referable vs not referable AMD) trained on either all-real vs all-synthetic data sets was compared. Main Outcomes and Measures: Accuracy of 2 retinal specialists (T.Y.A.L. and K.D.P.) for diagnosing and distinguishing AMD on real vs synthetic images and diagnostic performance (area under the curve) of DL algorithms trained on synthetic vs real images. Results: The diagnostic accuracy of 2 retinal specialists on real vs synthetic images was similar. The accuracy of diagnosis as referable vs nonreferable AMD compared with certified human graders for retinal specialist 1 was 84.54% (error margin, 4.06%) on real images vs 84.12% (error margin, 4.16%) on synthetic images and for retinal specialist 2 was 89.47% (error margin, 3.45%) on real images vs 89.19% (error margin, 3.54%) on synthetic images. Retinal specialists could not distinguish real from synthetic images, with an accuracy of 59.50% (error margin, 3.93%) for retinal specialist 1 and 53.67% (error margin, 3.99%) for retinal specialist 2. The DCNNs trained on real data showed an area under the curve of 0.9706 (error margin, 0.0029), and those trained on synthetic data showed an area under the curve of 0.9235 (error margin, 0.0045). Conclusions and Relevance: Deep learning-synthesized images appeared to be realistic to retinal specialists, and DCNNs achieved diagnostic performance on synthetic data close to that for real images, suggesting that DL generative techniques hold promise for training humans and machines.

[Applied study of Singapore occupational semi-quantitative risk assessment technique in occupational health risk assessment in a chair furniture manufacturing enterprise].

Objective: To assess the occupational health risk level of a small-scale furniture manufacturer, and to explore the applicability of the Singapore-developed semi-quantitative occupational risk assessment model for chemical exposure (Singapore model) in small-scale furniture manufacturers, and to provide a basis for the continuous occupational health management of manufacturers. Methods: A small-scale furniture manufacturer was selected as the study subject; an on-site occupational hygiene investigation was performed on the above manufacturer during April to June in 2017, and a risk assessment was carried out using the Singapore model. Results: The assessment results of the Singapore model indicated that risk levels of occupational exposure to harmful chemicals for the key positions in the workplace were inconsistent between the actual exposure level method and the exposure index method except for the following: high risk for formaldehyde exposure (risk level: 3.5 and 4.1, respectively) during woodworking process, high risk for dimethyl benzene exposure (risk level: 3.5 and 3.5, respectively) during burnishing process, medium risk for methyl benzene and dimethyl benzene exposure (risk level: 3.0 and 3.4, respectively) as well as for dimethyl benzene exposure (risk level: 3.0 and 3.3, respectively) during primer coating process, medium risk for methyl benzene exposure (risk level: 3.0 and 3.4, respectively) during gel painting process, and medium risk for cyclohexanone exposure (risk level: 2.8 and 2.8, respectively) during oil polishing process. The exposure index method yielded a higher risk level than the actual exposure level method, especially in the risk level of benzene exposure, which was rated as "high" and "very high" by the former but "low" by the latter. Conclusion: The Singapore model is suitable for risk assessment of occupational exposure to harmful chemicals in small-scale furniture manufacturers, which can provide a basis for further prevention and control measures taken by manufacturers.

[Automated Assessment for Bone Age of Left Wrist Joint in Uyghur Teenagers by Deep Learning].

OBJECTIVES: To realize the automated bone age assessment by applying deep learning to digital radiography （DR） image recognition of left wrist joint in Uyghur teenagers, and explore its practical application value in forensic medicine bone age assessment. METHODS: The X-ray films of left wrist joint after pretreatment, which were taken from 245 male and 227 female Uyghur nationality teenagers in Uygur Autonomous Region aged from 13.0 to 19.0 years old, were chosen as subjects. And AlexNet was as a regression model of image recognition. From the total samples above, 60％ of male and female DR images of left wrist joint were selected as net train set, and 10％ of samples were selected as validation set. As test set, the rest 30％ were used to obtain the image recognition accuracy with an error range in ±1.0 and ±0.7 age respectively, compared to the real age. RESULTS: The modelling results of deep learning algorithm showed that when the error range was in ±1.0 and ±0.7 age respectively, the accuracy of the net train set was 81.4% and 75.6% in male, and 80.5% and 74.8% in female, respectively. When the error range was in ±1.0 and ±0.7 age respectively, the accuracy of the test set was 79.5% and 71.2% in male, and 79.4% and 66.2% in female, respectively. CONCLUSIONS: The combination of bone age research on teenagers' left wrist joint and deep learning, which has high accuracy and good feasibility, can be the research basis of bone age automatic assessment system for the rest joints of body.

[Advantages and Application Prospects of Deep Learning in Image Recognition and Bone Age Assessment].

Deep learning and neural network models have been new research directions and hot issues in the fields of machine learning and artificial intelligence in recent years. Deep learning has made a breakthrough in the applications of image and speech recognitions, and also has been extensively used in the fields of face recognition and information retrieval because of its special superiority. Bone X-ray images express different variations in black-white-gray gradations, which have image features of black and white contrasts and level differences. Based on these advantages of deep learning in image recognition, we combine it with the research of bone age assessment to provide basic datum for constructing a forensic automatic system of bone age assessment. This paper reviews the basic concept and network architectures of deep learning, and describes its recent research progress on image recognition in different research fields at home and abroad, and explores its advantages and application prospects in bone age assessment.

Association of body composition with bone mineral density in northern Chinese men by different criteria for obesity.

UNLABELLED: With impressive economic development, obesity has emerged as a critical public health issue in China. Recently it was reported that obesity has taken an adverse effect on osteoporosis. Because there is different body mass index (BMI) for obesity globally, studies based on BMI levels on association of obesity with osteoporosis were quite few. Therefore, we discussed the relationship of body composition with skeletal BMD according to WHO BMI and BMI on Working Group on Obesity in China (WGOC). METHODS: A total of 502 adult men aged 20-89 were enrolled as healthy subjects for osteoporosis study at Qianfoshan Hospital, Shandong University between September 2008 and August 2010. According to WHO BMI, all subjects were divided into three groups: normal weight (18.5 ≤ BMI < 25 kg/m(2), n = 202), overweight (25 ≤ BMI < 30 kg/m(2), n = 242), and obesity (BMI ≥ 30 kg/m(2), n = 58). According to WGOC BMI, normal weight (18.5 ≤ BMI < 24 kg/m(2), n = 137), overweight (24 ≤ BMI < 28 kg/m(2), n = 225), and obesity (BMI ≥ 28 kg/m(2), n = 140). Total body and regional BMD, lean mass (LM), lean body mass index (LBMI), fat mass (FM), percent body fat (%BF) and fat mass index (FMI) were measured by dual-energy X-ray absorptiometry. Age-partial Pearson correlation analyses between body composition-related parameters and BMD. Multiple regression analyses were performed to explore the associations of BMD with LM, LBMI, FM, %BF and FMI. RESULTS: Fat mass (FM), %BF, FMI, LM and LBMI were positively correlated with BMD at almost sites (P < 0.001) in all subjects. However, the relationship was not different among groups. LM, LBMI, FM and FMI were positively correlated with BMD (P < 0.01) in normal weight. LM and LBMI appeared significantly positive with BMD in overweight and obesity according to WHO and WGOC criteria. %BF and FMI were negative significance with BMD at total body and some regional BMD according to WHO criteria in overweight (P < 0.05). In two obese groups, %BF appeared negatively significant with BMD (P < 0.05) according to WGOC criteria, and %BF and FMI appeared negatively significant with BMD (P < 0.05) according to WHO criteria. In regression of independent variables as FM and LM, LM showed statistically positively significant relations with BMD at almost sites (P < 0.05) in all groups. FM appeared positively significant with BMD in normal groups and overweight group according to WGOC criteria. In regression of independent variables as %BF and FMI, %BF and FMI appeared statistically negatively significant relations with BMD in overweight and obesity, but %BF and FMI were inconsistent in same site. CONCLUSIONS: Lean mass (LM) and LBMI could help to determinant of BMD, and %BF and FMI were adverse to BMD in overweight and obesity. Comparing with two criteria, we found the differences in fat-related parameters and BMD according to WHO criteria were more obvious than that according to WGOC criteria. We also found that %BF and FMI were useful to research the relationship between osteoporosis and obesity at the same time.