We conducted a systematic review with meta-analysis of randomized controlled trials that evaluated the effect of diabetes mobile phone applications. A total of 1550 participants from 21 studies were included. For type 1 diabetes, a significant 0.49% reduction in HbA1c was seen (95% CI, 0.04-0.94; I = 84%), with unexplained heterogeneity and a low GRADE of evidence. For type 2 diabetes, using diabetes apps was associated with a mean reduction of 0.57% (95% CI, 0.32-0.82; I = 77%). The results had severe heterogeneity that was explained by the frequency of HCP feedback. In studies with no HCP feedback, low frequency and high frequency HCP feedback, the mean reduction is 0.24% (95% CI, 0.02-0.49; I = 0%), 0.33% (95% CI, 0.07-0.59; I = 47%) and 1.12% (95% CI, 0.91-1.32; I = 0%), respectively, with a high GRADE of evidence. There is evidence that diabetes apps improve glycaemic control in type 1 diabetes patients. A reduction of 0.57% in HbA1c was found in type 2 diabetes patients. However, HCP functionality is important to achieve clinical effectiveness. Future studies are needed to explore the cost-effectiveness of diabetes apps and the optimal intensity of HCP feedback.

Background Risk-based breast cancer screening is a cost-effective intervention for controlling breast cancer in China, but the successful implementation of such intervention requires an accurate breast cancer prediction model for Chinese women. Objective This study aimed to evaluate and compare the performance of four machine learning algorithms on predicting breast cancer among Chinese women using 10 breast cancer risk factors. Methods A dataset consisting of 7127 breast cancer cases and 7127 matched healthy controls was used for model training and testing. We used repeated 5-fold cross-validation and calculated AUC, sensitivity, specificity, and accuracy as the measures of the model performance. Results The three novel machine-learning algorithms (XGBoost, Random Forest and Deep Neural Network) all achieved significantly higher area under the receiver operating characteristic curves (AUCs), sensitivity, and accuracy than logistic regression. Among the three novel machine learning algorithms, XGBoost (AUC 0.742) outperformed deep neural network (AUC 0.728) and random forest (AUC 0.728). Main residence, number of live births, menopause status, age, and age at first birth were considered as top-ranked variables in the three novel machine learning algorithms. Conclusions The novel machine learning algorithms, especially XGBoost, can be used to develop breast cancer prediction models to help identify women at high risk for breast cancer in developing countries.

There is strong evidence to suggest that obesity-related proteins play a key role in pathways that are related to breast cancer. In this study, we aimed to establish a robust obesity-related protein score (ORPS) that could be used to assess breast cancer risk. Based on evidence from high-quality systematic reviews and population studies, we selected nine such proteins that are stable in vitro, and measured their circulating concentrations by ELISA in a case–control study conducted in Chengdu, Sichuan, China, with 279 breast cancer cases and 260 healthy controls. Two obesity-related protein scores (ORPS) were calculated using a three-step method, with linear-weighted summation, and the one with a larger area under the curve was chosen for further evaluation. As a result, ORPS (PS5pre or PS4post) was positively correlated with breast cancer risk (premenopausal: OR≤63 VS >63 3.696, 95% CI 2.025–6.747; postmenopausal: OR≤38 VS >38 7.100, 95% CI 3.134–16.084), and represented a better risk predictor among obese women compared to non-obese in pre- and postmenopausal women. Among different molecular subtypes, ORPS was positively correlated with Luminal breast cancer, with additionally positive association with triple-negative breast cancer in premenopausal women. The ORPS might be a potential marker of breast cancer risk among Chinese women.

Background: Drug use safety in children is a global public health problem. The potentially inappropriate prescription screening tools are expected to reduce adverse drug reactions and promote rational drug use.Objectives: To systematically evaluate children’s potentially inappropriate prescription screening tools and validation studies on these tools.Methods: We systematically searched six databases PubMed, Embase, Cochrane Library, CNKI, VIP and Wanfang Data. Two reviewers independently selected articles by the eligible criteria and extracted data. Then we evaluated the coverage of diseases or drugs in these tools and the consistency of items between tools.Results: Five children’s potentially inappropriate prescription screening tools were identified, most tools were formed by Delphi expert consensus and focused on respiratory system drugs, anti-infective drugs, and gastrointestinal drugs. The coincidence rates of items between the POPI and the POPI Int, the POPI and the POPI United Kingdom, the POPI United Kingdom and the POPI int, and the POPI United Kingdom and the PIPc were 82.0, 55.1, 51.0 and 2.2% respectively, and the KIDs List did not overlap other four tools. Only the POPI tool developed by French experts was comprehensively validated by studies and most tools have not been validated.Conclusion: The development of screening tools for potentially inappropriate prescribing in children is a neglected field and most tools lack studies to validate clinical applicability. More researchers need to form their national potentially inappropriate prescription screening tools for children based on the best available clinical evidence and the actual clinical situation in their countries.