Shaomei Sun
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Research Interests
Shaomei Sun has a total of 73 co-authors. Here are the top co-authors by number of shared publications:
- Ge Meng
Nutritional Epidemiology Institute and School of Public Health, Tianjin Medical University, Tianjin, China
4 shared publications - Kaijun Niu
School of Public Health Tianjin University of Traditional Chinese Medicine Tianjin China
4 shared publications - Kun Song
Health Management Centre, Tianjin Medical University General Hospital, Tianjin, China
4 shared publications - Ming Zhou
Health Management Centre, Tianjin Medical University General Hospital, Tianjin, China
4 shared publications - Li Liu
Department of Urology, Zhongshan Hospital, Fudan University, Shanghai, China
3 shared publications - Shunming Zhang
School of Public Health, Xi'an Jiaotong University Health Science Center, Xi'an, China; Nutritional Epidemiology, Department of Clinical Sciences Malmö, Lund University, Malmö, Sweden
3 shared publications
Latest Publications
- Open Access
- 29 December 2022
- Published by JMIR Publications Inc. in JMIR Public Health and Surveillance
- Vol. 8 (12) , e35750
An Assessment of the Predictive Performance of Current Machine Learning–Based Breast Cancer Risk Prediction Models: Systematic Review
Background: Several studies have explored the predictive performance of machine learning–based breast cancer risk prediction models and have shown controversial conclusions. Thus, the performance of the current machine learning–based breast cancer risk prediction models and their benefits and weakness need to be evaluated for the future development of feasible and efficient risk prediction models. Objective: The aim of this review was to assess the performance and the clinical feasibility of the currently available machine learning–based breast cancer risk prediction models. Methods: We searched for papers published until June 9, 2021, on machine learning–based breast cancer risk prediction models in PubMed, Embase, and Web of Science. Studies describing the development or validation models for predicting future breast cancer risk were included. The Prediction Model Risk of Bias Assessment Tool (PROBAST) was used to assess the risk of bias and the clinical applicability of the included studies. The pooled area under the curve (AUC) was calculated using the DerSimonian and Laird random-effects model. Results: A total of 8 studies with 10 data sets were included. Neural network was the most common machine learning method for the development of breast cancer risk prediction models. The pooled AUC of the machine learning–based optimal risk prediction model reported in each study was 0.73 (95% CI 0.66-0.80; approximate 95% prediction interval 0.56-0.96), with a high level of heterogeneity between studies (Q=576.07, I2=98.44%; P<.001). The results of head-to-head comparison of the performance difference between the 2 types of models trained by the same data set showed that machine learning models had a slightly higher advantage than traditional risk factor–based models in predicting future breast cancer risk. The pooled AUC of the neural network–based risk prediction model was higher than that of the nonneural network–based optimal risk prediction model (0.71 vs 0.68, respectively). Subgroup analysis showed that the incorporation of imaging features in risk models resulted in a higher pooled AUC than the nonincorporation of imaging features in risk models (0.73 vs 0.61; Pheterogeneity=.001, respectively). The PROBAST analysis indicated that many machine learning models had high risk of bias and poorly reported calibration analysis. Conclusions: Our review shows that the current machine learning–based breast cancer risk prediction models have some technical pitfalls and that their clinical feasibility and reliability are unsatisfactory.
- 29 December 2022
- Published by JMIR Publications Inc. in JMIR Public Health and Surveillance
- Vol. 8 (12) , e35750
Open AccessAn Assessment of the Predictive Performance of Current Machine Learning–Based Breast Cancer Risk Prediction Models: Systematic Review
Background: Several studies have explored the predictive performance of machine learning–based breast cancer risk prediction models and have shown controversial conclusions. Thus, the performance of the current machine learning–based breast cancer risk prediction models and their benefits and weakness need to be evaluated for the future development of feasible and efficient risk prediction models. Objective: The aim of this review was to assess the performance and the clinical feasibility of the currently available machine learning–based breast cancer risk prediction models. Methods: We searched for papers published until June 9, 2021, on machine learning–based breast cancer risk prediction models in PubMed, Embase, and Web of Science. Studies describing the development or validation models for predicting future breast cancer risk were included. The Prediction Model Risk of Bias Assessment Tool (PROBAST) was used to assess the risk of bias and the clinical applicability of the included studies. The pooled area under the curve (AUC) was calculated using the DerSimonian and Laird random-effects model. Results: A total of 8 studies with 10 data sets were included. Neural network was the most common machine learning method for the development of breast cancer risk prediction models. The pooled AUC of the machine learning–based optimal risk prediction model reported in each study was 0.73 (95% CI 0.66-0.80; approximate 95% prediction interval 0.56-0.96), with a high level of heterogeneity between studies (Q=576.07, I2=98.44%; P<.001). The results of head-to-head comparison of the performance difference between the 2 types of models trained by the same data set showed that machine learning models had a slightly higher advantage than traditional risk factor–based models in predicting future breast cancer risk. The pooled AUC of the neural network–based risk prediction model was higher than that of the nonneural network–based optimal risk prediction model (0.71 vs 0.68, respectively). Subgroup analysis showed that the incorporation of imaging features in risk models resulted in a higher pooled AUC than the nonincorporation of imaging features in risk models (0.73 vs 0.61; Pheterogeneity=.001, respectively). The PROBAST analysis indicated that many machine learning models had high risk of bias and poorly reported calibration analysis. Conclusions: Our review shows that the current machine learning–based breast cancer risk prediction models have some technical pitfalls and that their clinical feasibility and reliability are unsatisfactory.
- Open Access
- 28 February 2022
- Published by Cambridge University Press (CUP) in British Journal of Nutrition
- Vol. 130 (2) , 276-283
Organ meat consumption and risk of non-alcoholic fatty liver disease: the Tianjin Chronic Low-grade Systemic Inflammation and Health cohort study
Prospective cohort studies linking organ meat consumption and non-alcoholic fatty liver disease (NAFLD) are limited, especially in Asian populations. This study aimed to prospectively investigate the association between organ meat consumption and risk of NAFLD in a general Chinese adult population. This prospective cohort study included a total of 15 568 adults who were free of liver disease, CVD and cancer at baseline. Dietary information was collected at baseline using a validated FFQ. NAFLD was diagnosed by abdominal ultrasound after excluding other causes related to chronic liver disease. Cox proportional regression models were used to assess the association between organ meat consumption and risk of NAFLD. During a median of 4·2 years of follow-up, we identified 3604 incident NAFLD cases. After adjusting for demographic characteristics, lifestyle factors, vegetable, fruit, soft drink, seafood and red meat consumption, the multivariable hazard ratios (95 % CI) for incident NAFLD across consumption of organ meat were 1·00 (reference) for almost never, 1·04 (0·94, 1·15) for tertile 1, 1·08 (0·99, 1·19) for tertile 2 and 1·11 (1·01, 1·22) for tertile 3, respectively (Pfor trend < 0·05). Such association did not differ substantially in the sensitivity analysis. Our study indicates that organ meat consumption was related to a modestly higher risk of NAFLD among Chinese adults. Further investigations are needed to confirm this finding.
- 28 February 2022
- Published by Cambridge University Press (CUP) in British Journal of Nutrition
- Vol. 130 (2) , 276-283
Open AccessOrgan meat consumption and risk of non-alcoholic fatty liver disease: the Tianjin Chronic Low-grade Systemic Inflammation and Health cohort study
Prospective cohort studies linking organ meat consumption and non-alcoholic fatty liver disease (NAFLD) are limited, especially in Asian populations. This study aimed to prospectively investigate the association between organ meat consumption and risk of NAFLD in a general Chinese adult population. This prospective cohort study included a total of 15 568 adults who were free of liver disease, CVD and cancer at baseline. Dietary information was collected at baseline using a validated FFQ. NAFLD was diagnosed by abdominal ultrasound after excluding other causes related to chronic liver disease. Cox proportional regression models were used to assess the association between organ meat consumption and risk of NAFLD. During a median of 4·2 years of follow-up, we identified 3604 incident NAFLD cases. After adjusting for demographic characteristics, lifestyle factors, vegetable, fruit, soft drink, seafood and red meat consumption, the multivariable hazard ratios (95 % CI) for incident NAFLD across consumption of organ meat were 1·00 (reference) for almost never, 1·04 (0·94, 1·15) for tertile 1, 1·08 (0·99, 1·19) for tertile 2 and 1·11 (1·01, 1·22) for tertile 3, respectively (Pfor trend < 0·05). Such association did not differ substantially in the sensitivity analysis. Our study indicates that organ meat consumption was related to a modestly higher risk of NAFLD among Chinese adults. Further investigations are needed to confirm this finding.
- 1 February 2022
- Published by Royal Society of Chemistry (RSC) in Food & Function
- Vol. 13 (6) , 3431-3440
Ultra-processed food consumption and the risk of subclinical thyroid dysfunction: a prospective cohort study
Background: prospective cohort studies linking ultra-processed foods (UPF) and subclinical thyroid dysfunction (SCTD) are limited, especially in Chinese adults. Objective: we designed a large-scale cohort study to examine whether UPF consumption is a risk factor for SCTD in adults. Methods: this prospective cohort study investigated 8732 participants (mean age: 44.7 ± 10.7 years; 53.3% men). UPF consumption was measured at the baseline using a validated food frequency questionnaire and classified according to the NOVA classification system. SCTD was characterized by abnormal serum thyroid-stimulating hormone levels and normal free thyroxine. Cox proportional hazards regression models were used to examine the association of UPF consumption with the risk of various types of SCTD. Results: during the follow-up period, the incidence rate of subclinical hypothyroidism was 8.3/1000 person-years and that of subclinical hyperthyroidism was 7.0/1000 person-years. After adjusting the potential confounding factors, the multivariable hazard ratios (HRs) (95% confidence intervals) across increasing tertiles of energy-adjusted UPF consumption for subclinical hypothyroidism were 1.00 (reference), 0.82 (0.57, 1.18) and 0.87 (0.60, 1.27) (P for trend = 0.47) and for subclinical hyperthyroidism were 1.00 (reference), 1.09 (0.71, 1.68) and 1.69 (1.12, 2.56) (P for trend = 0.01). Per one standard deviation (58.0 g per 1000 kcal per day) increase in ultra-processed food consumption a 14% increase in risk of subclinical hyperthyroidism was observed (HR = 1.14; 95% CI: 1.00, 1.30; P = 0.04). Further sensitivity analysis showed that the fully adjusted HRs (95% CIs) of subclinical hyperthyroidism across tertiles of the proportion of UPF in weight for subclinical hyperthyroidism were 1.00 (reference), 1.06 (0.69, 1.63) and 1.63 (1.09, 2.46) (P for trend = 0.01). Conclusions: this population-based prospective cohort study has firstly demonstrated that higher UPF consumption was associated with higher risk of subclinical hyperthyroidism in adults. The UMIN clinical trial registry number is UMIN000027174 (https://upload.umin.ac.jp/cgi-open-bin/ctr_e/ctr_view.cgi?recptno=R000031137).
- 1 February 2022
- Published by Royal Society of Chemistry (RSC) in Food & Function
- Vol. 13 (6) , 3431-3440
Ultra-processed food consumption and the risk of subclinical thyroid dysfunction: a prospective cohort study
Background: prospective cohort studies linking ultra-processed foods (UPF) and subclinical thyroid dysfunction (SCTD) are limited, especially in Chinese adults. Objective: we designed a large-scale cohort study to examine whether UPF consumption is a risk factor for SCTD in adults. Methods: this prospective cohort study investigated 8732 participants (mean age: 44.7 ± 10.7 years; 53.3% men). UPF consumption was measured at the baseline using a validated food frequency questionnaire and classified according to the NOVA classification system. SCTD was characterized by abnormal serum thyroid-stimulating hormone levels and normal free thyroxine. Cox proportional hazards regression models were used to examine the association of UPF consumption with the risk of various types of SCTD. Results: during the follow-up period, the incidence rate of subclinical hypothyroidism was 8.3/1000 person-years and that of subclinical hyperthyroidism was 7.0/1000 person-years. After adjusting the potential confounding factors, the multivariable hazard ratios (HRs) (95% confidence intervals) across increasing tertiles of energy-adjusted UPF consumption for subclinical hypothyroidism were 1.00 (reference), 0.82 (0.57, 1.18) and 0.87 (0.60, 1.27) (P for trend = 0.47) and for subclinical hyperthyroidism were 1.00 (reference), 1.09 (0.71, 1.68) and 1.69 (1.12, 2.56) (P for trend = 0.01). Per one standard deviation (58.0 g per 1000 kcal per day) increase in ultra-processed food consumption a 14% increase in risk of subclinical hyperthyroidism was observed (HR = 1.14; 95% CI: 1.00, 1.30; P = 0.04). Further sensitivity analysis showed that the fully adjusted HRs (95% CIs) of subclinical hyperthyroidism across tertiles of the proportion of UPF in weight for subclinical hyperthyroidism were 1.00 (reference), 1.06 (0.69, 1.63) and 1.63 (1.09, 2.46) (P for trend = 0.01). Conclusions: this population-based prospective cohort study has firstly demonstrated that higher UPF consumption was associated with higher risk of subclinical hyperthyroidism in adults. The UMIN clinical trial registry number is UMIN000027174 (https://upload.umin.ac.jp/cgi-open-bin/ctr_e/ctr_view.cgi?recptno=R000031137).
- 1 October 2019
- Published by Elsevier in Journal of the American Medical Directors Association
- Vol. 20 (10) , 1236-1241
Thyroid Function as a Predictor of Handgrip Strength Among Middle-Aged and Older Euthyroid Adults: The TCLSIH Cohort Study
- 1 October 2019
- Published by Elsevier in Journal of the American Medical Directors Association
- Vol. 20 (10) , 1236-1241
Thyroid Function as a Predictor of Handgrip Strength Among Middle-Aged and Older Euthyroid Adults: The TCLSIH Cohort Study
- 1 October 2018
- Published by Elsevier in Journal of Thoracic Oncology
- Vol. 13 (10) , S342
OA09.07 Association Between Outdoor Air Pollution And Lung Cancer in Female Never Smokers
- 1 October 2018
- Published by Elsevier in Journal of Thoracic Oncology
- Vol. 13 (10) , S342
OA09.07 Association Between Outdoor Air Pollution And Lung Cancer in Female Never Smokers
- 28 July 2016
- Published by Springer Nature in Journal of Human Hypertension
- Vol. 31 (2) , 116-120
Serum uric acid is an independent predictor for developing prehypertension: a population-based prospective cohort study
Although the prevalence of prehypertension is rapidly increasing in China, the medical community has paid little attention to its prevention. Prior studies have demonstrated that uric acid directly contributes to vascular remodelling and endothelial dysfunction. However, few prospective studies have assessed the relationship between serum uric acid and prehypertension. We therefore designed a larger-scale cohort study to examine whether uric acid level is a predictive factor for developing prehypertension in adults. Participants were recruited from Tianjin Medical University General Hospital-Health Management Centre. A prospective assessment (n=15 143) was performed. Participants without a history of hypertension or prehypertension were followed up for 2 to 6 years with a median follow-up duration of 2.8 years. Serum uric acid levels and blood pressure were assessed yearly during the follow-up. Adjusted Cox proportional hazards regression models were used to assess relationships between the quintiles of uric acid levels and the incidence of prehypertension. The incidence of prehypertension was 191 per 1000 person-years. In the final multivariate models, the hazard ratios (95% confidence interval) for prehypertension across uric acid quintiles were 1.00 (reference), 0.98 (0.90–1.07), 1.01 (0.93–1.10), 1.09 (1.001–1.20) and 1.17 (1.06–1.29) (P for trend <0.001), respectively. This population-based prospective cohort study has demonstrated that uric acid level is an independent predictor for developing prehypertension.
- 28 July 2016
- Published by Springer Nature in Journal of Human Hypertension
- Vol. 31 (2) , 116-120
Serum uric acid is an independent predictor for developing prehypertension: a population-based prospective cohort study
Although the prevalence of prehypertension is rapidly increasing in China, the medical community has paid little attention to its prevention. Prior studies have demonstrated that uric acid directly contributes to vascular remodelling and endothelial dysfunction. However, few prospective studies have assessed the relationship between serum uric acid and prehypertension. We therefore designed a larger-scale cohort study to examine whether uric acid level is a predictive factor for developing prehypertension in adults. Participants were recruited from Tianjin Medical University General Hospital-Health Management Centre. A prospective assessment (n=15 143) was performed. Participants without a history of hypertension or prehypertension were followed up for 2 to 6 years with a median follow-up duration of 2.8 years. Serum uric acid levels and blood pressure were assessed yearly during the follow-up. Adjusted Cox proportional hazards regression models were used to assess relationships between the quintiles of uric acid levels and the incidence of prehypertension. The incidence of prehypertension was 191 per 1000 person-years. In the final multivariate models, the hazard ratios (95% confidence interval) for prehypertension across uric acid quintiles were 1.00 (reference), 0.98 (0.90–1.07), 1.01 (0.93–1.10), 1.09 (1.001–1.20) and 1.17 (1.06–1.29) (P for trend <0.001), respectively. This population-based prospective cohort study has demonstrated that uric acid level is an independent predictor for developing prehypertension.