Objective:To construct a prediction model for the risk of sarcopenia in middle-aged and elderly patients with knee osteoarthritis (KOA) based on machine learning, and to provide a basis for carrying out the prevention of sarcopenia in patients with KOA.Methods:Clinical data of KOA patients from three tertiary hospitals in Guangdong Province were collected between December 2023 and September 2024 using a convenience sampling method. The data were randomly split into training and test sets at an 8:2 ratio, with the occurrence of sarcopenia as the outcome variable. Risk prediction models for sarcopenia were constructed using eight machine learning algorithms: logistic regression, K-nearest neighbors, support vector machine, decision tree, neural network, random forest, gradient boosting machine (GBM), and eXtreme gradient boosting. Model performance was evaluated based on metrics including the area under the receiver operating characteristic curve (AUC), accuracy, precision, sensitivity, specificity, and F1 score. The optimal model was selected, and feature importance was visualized using the Shapley Additive exPlanations (SHAP) method.Results:Data from 640 KOA patients were analyzed, 143 males and 497 females, (67.51± 7.72) years, with 136 cases (21.25%) developing sarcopenia. All eight prediction models showed high AUC values, with the GBM model demonstrating the best performance. Its metrics included an AUC of 0.926 (95% CI 0.874 - 0.965), accuracy of 0.852, precision of 0.611, sensitivity of 0.815, specificity of 0.861, and F1 score of 0.698. SHAP analysis identified body mass index, calf circumference, body fat percentage, WOMAC score, and age as the most important predictive features. Conclusions:The GBM-based risk prediction model for sarcopenia in middle- aged and elderly KOA patients demonstrated optimal performance, enabling healthcare professionals to accurately and promptly identify high-risk groups among these patients and to develop effective, evidence-based intervention strategies.

Objective:To construct a prediction model for the risk of sarcopenia in middle-aged and elderly patients with knee osteoarthritis (KOA) based on machine learning, and to provide a basis for carrying out the prevention of sarcopenia in patients with KOA.Methods:Clinical data of KOA patients from three tertiary hospitals in Guangdong Province were collected between December 2023 and September 2024 using a convenience sampling method. The data were randomly split into training and test sets at an 8:2 ratio, with the occurrence of sarcopenia as the outcome variable. Risk prediction models for sarcopenia were constructed using eight machine learning algorithms: logistic regression, K-nearest neighbors, support vector machine, decision tree, neural network, random forest, gradient boosting machine (GBM), and eXtreme gradient boosting. Model performance was evaluated based on metrics including the area under the receiver operating characteristic curve (AUC), accuracy, precision, sensitivity, specificity, and F1 score. The optimal model was selected, and feature importance was visualized using the Shapley Additive exPlanations (SHAP) method.Results:Data from 640 KOA patients were analyzed, 143 males and 497 females, (67.51± 7.72) years, with 136 cases (21.25%) developing sarcopenia. All eight prediction models showed high AUC values, with the GBM model demonstrating the best performance. Its metrics included an AUC of 0.926 (95% CI 0.874 - 0.965), accuracy of 0.852, precision of 0.611, sensitivity of 0.815, specificity of 0.861, and F1 score of 0.698. SHAP analysis identified body mass index, calf circumference, body fat percentage, WOMAC score, and age as the most important predictive features. Conclusions:The GBM-based risk prediction model for sarcopenia in middle- aged and elderly KOA patients demonstrated optimal performance, enabling healthcare professionals to accurately and promptly identify high-risk groups among these patients and to develop effective, evidence-based intervention strategies.

OBJECTIVE To explore how to shorten the time from approval to start-up of drug clinical trial project. METHODS Twenty-two phase Ⅱ-Ⅲ multi-center drug clinical trial projects start up in The First People's Hospital of Nanning from 2020 to 2021 were selected. The time-consuming of each link before the launch was analyzed, and the time- consuming of project approval, ethical review and contract review between the sponsor and research institution was compared, as well as the influence of using the contract template of each party on the time-consuming of contract review was compared. RESULTS Contract review took the longest time. There was no significant difference in the time-consuming between the sponsor and the research institutions in the three links of project approval, ethical review and contract review. Used the contract template of the research institutions, the time spent by the sponsor and the research institution in the review process, as well as the contract signing time of the project were shorter. CONCLUSION Using the clinical trial management system, conduct differentiated ethical review methods, advance drug delivery and commitment letter submission, use the contract template of research institutions or sign a framework contracts, establish a effective communication methods are all effective ways to reduce the time taken before start-up.

Objective:To explore the existing issues in radiation protection during the treatment of 131I by means of measuring the ambient dose equivalent rate to patients with thyroid cancer and the dose equivalent to the surface of chest of patients during hospitalization. Methods:The ambient dose equivalent rate (peer) was measured by using gamma ray detector for selected 78 patients who received 131I treatment in a hospital 10 min, 1 d, 2 d, 3 d and 5 d after administration with 131I. The measurements were made at distances of 5 cm, 0.5 m and 1 m from the body surface in front, rear, left and right directions. The photoluminescence dosimeter on the chest of the patients was used to measure the effective dose during hospitalization period (6 d). Results:The ambient dose equivalent rate on the surface of chest of patients was up to 4.81 mSv/h 10 min after administration of medicine. The dose equivalent on the surface of chest of patients before discharge ranged 2.6-64.1 μSv/h. The cumulative dose on chest surface during hospitalization was 15.9-58.8 mGy. There was a significant difference in the dose rate at 5 cm from the body surface between 3.7 GBq group and 5.55 GBq group 10 min after medication ( t=-6.11, P<0.05). There was a significant difference in the dose rate at 5 cm from the body surface between male and female groups 10 min after medication ( t=4.52, P < 0.05). There was no significant difference in other groups ( P > 0.05). Conclusions:During the 131I treatment, patients had high level of radiation around them, so it is necessary to strengthen the protection and management of patients and reduce unnecessary exposure to the public.

To ascertain the background and frequency of diagnostic medical X-ray procedures in China and provide the basis for regulatory oversight of such applications,a total of 557 medical institutions in 25 provinces or municipalities were surveyed by means of the optimally designed questionnaires and through stratified quota sampling.The numbers of procedures were calculated in terms of the type of procedures and the sex and age of examined patients.As a result,the frequencies of diagnostic X-ray procedures for 2016 in the country were derived using multiple linear regression analysis.The frequency of X-ray diagnosis in 10 provinces of China in 2016 was estimated to be 379-1 228 examinations per 1 000 population.Diagnostic X-ray applications have shown a rapid expansion in 2016 as compared with the period of "9th Five-Year Plan".It is very important to strengthen the regulation of medical diagnostic X-ray applications.

Objective To validate and discuss the time response correction formula for four types of dosimeters (6150AD6 + 6150AD-b,FH40G + FHZ672E-10,451P ionization chamber and AT1123).Methods The ambient dose equivalent rates shown by survey meters were recorded separately when X-ray emission time was 500,200,100 and 50 ms.The corrected values were obtained by the formula of circuit having a capacitance C and asistance R in series.Results Therewas no correlation between the value measured by AT1123 dosimeter and the time of irradiation.The values by other three kinds of dosimeters obviously varied with the time of irradiation.Conclusions It is not required to make the time response correction for the measured value of ATl123 dosemeter,whereas the values measured by the other three dosimeters could be corrected by the time response correction formula.

Objective To measure the circuit time constants of 4 kinds of radiation survey meters (451P ionization chamber dosimeter,6150AD6 + 6150AD-b dose meter,FH40G + FHZ672E-10 dose meter and AT1123 dose meter) and,to discuss the formula of time response correction and its application.Methods In the condition of continuous exposure of X-ray machine,the ambient dose equivalent rates shown by survey meters were recorded.In order to get the circuits time constant,the least squares fittingmethod was used to fit the data using the time response formula of circuit having a capacitance C and a resistance R in series.Results The relative uncertainty of fitted circuit time constants was higher than 20％ except for 6150AD6 + 6150AD-b dose meter.The relative uncertainty of fitted r was 8％ for 6150AD6 + 6150AD-b dose meter.Conclusions The time required to stabilize the dosimeter readings was 8,5,3 and 2 s,respectively,for the 451P ionization chamber dosimeter,6150AD6 +6150AD-b dose meter,FH40G + FHZ672E-10 dose meter and AT1123 dose meter.The rising trend of their measured values was not fully accordance with the RC circuit time response correction formula.

Objective To use Monte Carlo method to build a shielding calculation model for the proton and heavy ion treatment room,and to provide a reliable calculation method for shielding design.Methods A Monte Carlo-based FLUKA code was adopted to build the shielding calculation model for the proton and heavy ion treatment room,and to simulate the radiation field distribution in the proton and heavy ion treatment room.The calculation model was verified through the radiation detection around the proton and heavy ions treatment room.Results The FLUKA code-based simulation results were consistent with the radiation detection.Conclusions The shielding calculation model based on FLUKA code can simulate the radiation field from proton and heavy ions.Among the secondary particles,secondary neutrons are the dominant component and the main concern of accelerator shielding design.In shielding calculation,the emphasis should be put on both beam intensity and energy.

Objective To investigate the current situation of radiation protection in nuclear medicine diagnosis workplace.Methods The study was performed in 3 hospitals in northeast,north and central of China from February to December in 2013.The γ dose rate instrument was used to detect the workplace ambient dose equivalent rate of medicine preparation,leaching,packing,injection and imaging.Individual effective dose and equivalent dose were evaluated by photoluminescent dosimeter.Results The ambient dose equivalent rate was up to 1.92 mSv/h at repacking place and 1.2 mSv/h at injection place.The ambient dose equivalent rate of patients after injection was 5.36-240 μ,Sv/h.The hand equivalent dose was 0.01-0.02 mGy.Moreover,there were problems of staff route intersection,as well as the patients after injection staying in the public area.Conclusions Radiation workers should pay more attention to individual protection,and improve the operation proficiency to shorten the operation time.Furthermore,in order to protect public from unnecessary irradiation,there should be some changes in staff route and patients administration.