OBJECTIVE:Based on different algorithms of machine learning,the prediction model of lumbar disc herniation has become a trend and hot spot in the development of precision medicine.However,there is limited evidence on the reporting quality and methodological quality of prediction models of lumbar disc herniation outcomes using machine learning.This article is aimed to explore the performance of machine learning algorithms in predicting the prognosis of lumbar disc herniation by comprehensively analyzing the report quality and risk of bias of previous studies that developed and validated prognosis prediction models based on machine learning through a comprehensive literature search,in order to explore the performance of machine learning algorithms in predicting the prognosis of lumbar disc herniation.METHODS:The databases of CNKI,WanFang,VIP,SinOMED,PubMed,Web of Science,Embase,and The Cochrane Library were searched by computer.Studies on the use of machine learning to develop(and/or validate)prognostic prediction models for lumbar disc herniation were collected from the inception of the database to December 31,2023.Two researchers independently screened the literature,extracted data,and assessed the risk of bias of the included studies.The reporting quality and risk of bias of the included studies were assessed by the Multivariable Transparent Reporting of Predictive Models(TRIPOD)statement and the Predictive Model Risk of Bias Assessment Tool(PROBAST).The results of the evaluation were analyzed using descriptive statistics and visual charts.RESULTS:(1)A total of 23 articles were included,and the TRIPOD compliance of each study ranged from 11％to 87％,with a median compliance of 54％.The quality of reporting of titles,detailed descriptions of treatment measures,blinding of predictors,handling of missing data,details of risk stratification,specific procedures for enrollment,model interpretation,and model performance was mostly poor,with TRIPOD adherence rates ranging from 4％to 35％.(2)Of all included studies,61％had a high risk of bias and 39％had an unclear overall risk of bias.The area under the curve,accuracy,sensitivity and specificity were used to evaluate the performance of the model.The areas under the curve of 20 models were reported,ranging from 0.561 to 0.999.Three models reported the accuracy of the model,ranging from 82.07％to 89.65％.(3)Among all included studies,the statistical analysis domain was most often assessed as having a high risk of bias,mainly due to the small number of valid samples,the selection of predictors based on univariate analysis and the lack of calibration and discrimination assessment of the model in the study.CONCLUSION:These results indicate that machine learning can achieve good predictive ability in the development and validation of prognostic models for lumbar disc herniation.The commonly used algorithms include regression algorithm,support vector machine,decision tree,random forest,artificial neural network,naive Bayes and other algorithms.Reasonable algorithms combined with clinical practice can improve the accuracy of prognosis prediction of lumbar disc herniation.However,the reporting and methodological quality of prognosis prediction models based on machine learning are poor,the prediction performance of different models varies greatly,and the generalization and extrapolation of research models are unclear.There is an urgent need to improve the design,implementation and reporting of such studies.To promote the application of machine learning in the clinical practice of lumbar disc herniation prediction models,it is necessary to comprehensively consider various predictors related to the prognosis of the disease before modeling,and strictly follow the relevant standards of PROBAST tool during modeling.

ObjectiveTo analyze the epidemiological and clinical characteristics of the 21 confirmed monkeypox cases in a district of Chengdu City, and to provide scientific guidance for the prevention and control of subsequent monkeypox epidemics. MethodsData of confirmed monkeypox cases residing in this district were collected from the Disease Control and Prevention Information System of China. A retrospective descriptive epidemiological analysis was used to analyze the demographic, distributional and behavioral characteristics of the cases. ResultsThe first confirmed case of monkeypox was reported on July 5, 2023. Up to April 30, 2025, a total of 21 confirmed cases of monkeypox have been reported. All cases were male, with a mean age of (30.9±6.2) years. The highest proportion of cases(47.62%) was in the 30‒40 years age group. The majority were men who have sex with men (MSM) population (90.48%, 19/21). The results showed that 19.05% of cases were co-infected with HIV, and 19.05% had a history of syphilis infection. Within 21 days prior to symptom onset, 19 cases (90.48%) self-reported engaging in male-to-male sexual contact, among whom 10 cases (52.63%) reported having taken protective measures, while 9 cases (47.37%) did not take safety precautions. Thirteen cases (61.90%) had no travel history to areas with reported monkeypox cases during the 21 days before symptom onset. The predominant manifestation was exanthem (100%, 21/21), followed by fever (57.14%, 12/21) and lymphadenectasis (47.62%, 10/21). Among febrile cases, 50.00% (6/12) had low-grade fever (37.3‒38.0 ℃). All cases were identified through active medical consultation. The median interval from symptom onset to the first medical visit was 3 (2, 6) days, with a maximum interval of 14 days. The median interval from symptom onset to laboratory confirmation was 7 (5, 9) days. Six cases (28.57%) had two or more visits to the hospital, with bacterial infection being the primary initial diagnosis. ConclusionMonkeypox prevention and control efforts in a district of Chengdu City should prioritize MSM population and young and middle-aged adults aged 30 to <40 years. It is recommended to establish an integrated monkeypox epidemic prevention and control network by leveraging existing HIV/AIDS prevention and control network. Concurrently, accelerating the deployment of the national intelligent infectious disease monitoring and early warning front-end software will strengthen early detection capabilities and be beneficial for the overall effectiveness of epidemic prevention and control efforts.

OBJECTIVE To explore the improvement effects of Suting pingchuan decoction （STPC） on bronchial asthma in rats and its potential mechanism. METHODS Male SD rats were randomly divided into blank group， model group， STPC low-， medium- and high-dose groups （4.14， 8.28， and 16.56 g/kg， respectively， based on the crude drug dosage）， and dexamethasone group （positive control， 1 mg/kg）， with 8 rats in each group. Except for the blank group， the other groups were sensitized by intraperitoneal injection of ovalbumin and challenged by nebulized inhalation of ovalbumin to establish a bronchial asthma model. From the day of nebulization challenge， rats of each group were administered the corresponding drug solution or normal saline by gavage 1 hour before aerosolization， once a day， for 7 consecutive days. During the experiment， behavioral changes in rats of each group were observed. After the last administration， the levels of inflammatory factors （interleukin-1β， interleukin-18） in bronchoalveolar lavage fluid （BALF）， and oxidative stress indexes [malondialdehyde （MDA）， superoxide dismutase （SOD）] in lung tissue were determined； pathological changes in lung tissue were observed； cell apoptosis in lung tissue， and the mRNA expression of nucleotide-binding domain leucine-rich repeat and pyrin domain-containing receptor 3 （NLRP3） and the protein expressions of NLRP3， cleaved caspase-1， caspase-1， gasdermin D （GSDMD） and gasdermin D-N-terminal domain （GSDMD-N） in lung tissue were detected. RESULTS Compared with the blank group， rats in the model group showed symptoms such as scratching their ears and noses and frequent sneezing； the lung tissue structure was severely damaged， and there was obvious inflammatory cell infiltration. The levels of inflammatory factors in BALF， as well as the level of MDA， TUNEL-apoptotic cell rate， the mRNA expression of NLRP3， and the protein expressions of NLRP3， cleaved caspase-1， caspase-1， GSDMD-N and GSDMD in lung tissue were significantly increased or up-regulated， while the level of SOD in lung tissue was significantly decreased （ P ＜0.05 or P ＜0.01）. Compared with the model group， the above symptoms and pathological damages of lung tissue in each drug group were significantly improved， and all quantitative indicators （except for the protein expressions of GSDMD in the STPC groups， as well as the level of SOD， and the protein expressions of cleaved caspase-1， caspase-1 and GSDMD-N in the STPC low-dose group） were significantly reversed （ P ＜0.05 or P ＜0.01）. CONCLUSIONS STPC can improve airway inflammation and lung tissue damage in rats with bronchial asthma， reduce the level of oxidative stress， and decrease the release of inflammatory factors such as IL-1β and IL-18. Its mechanism of action may be related to the inhibition of pyroptosis activation mediated by the NLRP3/caspase-1/GSDMD signaling pathway.

ObjectiveTo construct a nomogram prediction model for non-valvular persistent atrial fibrillation （PeAF） patients with left ventricular hypertrophy （LVH） ， followed by prognostic analysis through follow-up. MethodsThis study retrospectively enrolled 949 patients with newly diagnosed and hospitalized non-valvular PeAF. Among them， 403 patients presented with LVH. The cohort was randomly stratified into a training set （n=665） and a validation set （n=284）. Univariate and multivariate Logistic regression analyses were employed to identify independent risk factors for PeAF complicated by LVH. A nomogram prediction model was subsequently constructed and evaluated for discriminative ability， calibration， and clinical utility using receiver operating characteristic （ROC） curve analysis， calibration plots， and decision curve analysis （DCA）. ResultsSeven independent risk factors were ultimately identified and included in the prediction model： female sex， hypertension， diabetes， red blood cell distribution width-SD （RDW-SD）， body mass index （BMI）， left atrial diameter （LAD）， and left ventricular ejection fraction （LVEF）. The area under the ROC curve （AUC） in the training set was 0.862 （95% CI： 0.834-0.890）， and in the validation set， it was 0.870 （95% CI： 0.829-0.911）， demonstrating excellent predictive performance. ConclusionIndependent risk factors for LVH in PeAF patients include female， hypertension， diabetes， RDW-SD， BMI， LAD， and LVEF. The prediction model built based on this can help early identification of PeAF patients with high risk of LVH. At the same time， the incidence of major adverse cardiovascular events （MACE） is higher in PeAF patients with LVH. Patients with atrial fibrillation combined with LVH may benefit from catheter ablation.

Complexity science and traditional Chinese medicine （TCM） share a profound intrinsic compatibility. To address the limitations of reductionist approaches in guiding the dynamically holistic theory and practice of TCM， this paper develops​ a novel research pathway under the guidance of complexity science， using the open complex giant systems （OCGS） theory as its practical framework， upholding the fundamental principles of TCM and breaking new ground of artificial intelligence （AI）. The core tasks of this pathway involve the objectification and structured diagnosis-treatment systems， and the construction of a complex， large-scale TCM model. The paper further delineates the data architecture design and reinforcement learning training pathway. Acknowledging the differences in information processing between the human brain and computers， it proposes a training strategy that minimizes manual labeling of "syndromes" and instead focuses on the training paradigm of "herb-human interaction → syndrome manifestation → dosage adjustment". Under the guidance of complexity science theory， TCM is poised for deep integration with AI technology， driving technological innovation powered by data and computational capabilities.

ObjectiveTo investigate the possible mechanism of Pibai Yucuo Formula （枇柏愈痤方， PYF） in treating acne from the perspective of skin barrier damage. MethodsThirty-two mice were randomly divided into blank group， model group， minocycline group， and PYF group， with 8 mice in each group. Except for the blank group， mice were induced by intradermal injection of Cutibacterium acnes （C.acnes） combined with topical application of artificial sebum to establish acne model. The blank group and model group received intragastric administration of 0.2 ml of distilled water， while the PYF group received intragastric administration of 22.75 g/（kg·d）of PYF， and the minocycline group received 0.013 g/（kg·d）of minocycline suspension， all once daily for 5 consecutive days. On day 0 and day 6 of the experiment， the body weight of mice in each group was recorded， and the absolute value of the body weight difference during the experiment was calculated. Skin conditions were assessed with multifunctional skin imaging system on the 2nd， 4th and 6th day of the experiment. Skin barrier function indicators including transepidermal water loss （TEWL）， and the water content of the stratum corneum and epidermis on day 0， 2， 4 and 6 of the experiment. Optical coherence tomography （OCT） was used to observe stratum corneum and skin thickness on the 1st， 3rd and 5th day of the experiment. Hematoxylin-eosin （HE） staining was performed to observe histopathological changes， while ELISA was used to detect interleukin-17A （IL-17A） levels， and immunofluorescence staining was used to assess skin barrier-related proteins filaggrin （FLG） and loricrin （LOR） levels of skin lesions on day 6 of the experiment. ResultsCompared to the blank group， the model group showed a decrease in body weight on day 6， and an increase in the absolute value of the difference in body weight before and after the experiment （P＜0.05）. On day 4 and 6， TEWL values increased， while water content in the skin stratum corneum and epidermis decreased （P＜0.05）， accompanied by elevated IL-17A level and reduced immunofluorescence intensity of FLG and LOR proteins （P＜0.05）. The model group mice showed papules or pustules at the skin modeling site with progressively worsening desquamation under multifunctional skin imaging system. OCT revealed focal epidermal protrusions， blurred epidermal-dermal boundaries， and disorganized structural layers. HE staining showed significant epidermal hyperkeratosis and incomplete keratinization in the skin， with keratin plug formation in hair follicles and glandular lumens， thickened stratum corneum， hyperplasia of the stratum spinosum， as well as dense dermal inflammatory cell infiltration， and capillary dilation. Compared to the model group， both the minocycline group and the PYF group showed a reduced difference in body weight before and after experiment （P＜0.05）. On day 4 and 6， the TEWL value decreased， and water content of the skin stratum corneum increased （P＜0.05）； on day 6， the IL-17A level in the skin lesions decreased and immunofluorescence intensity of FLG and LOR proteins increased （P＜0.05）. On day 4 and 6， the severity of the skin lesions and range of redness and swelling were lighter than those in the model group， with reverted epidermal thickness， smoother surface and clearer epidermis-dermis boundary. HE staining showed that the degree of skin keratinization was reduced， and the inflammatory infiltration and vascular dilation in the dermis were improved compared to the model group. The PYF group showed better results than the minocycline group in reducing TEWL value on day 4 （P＜0.05）. ConclusionPYF may improve inflammation and skin barrier damage by downregulating IL-17A levels in lesion tissue and increasing skin barrier-related proteins， which could be one of the potential mechanism of action on acne.

ObjectiveTo investigate the value of predictive models established based on machine learning methods in predicting the risk of metabolic associated fatty liver disease （MAFLD）， and to analyze its key risk factors. MethodsA retrospective analysis was performed for the 50 variables of 2 168 healthy individuals who underwent physical examination in Department of Health Assessment， Xiyuan Hospital， China Academy of Chinese Medical Sciences， from January 2021 to December 2024， including body composition， past history， and laboratory tests， and according to whether they were diagnosed with MAFLD or not， they were divided into MAFLD group with 265 individuals and non-MAFLD group with 1 903 individuals. The Mann-Whitney U test was used for comparison of continuous data between two groups， and the chi-square test was used for comparison of categorical data between two groups. Randomly split the research data into a training set and a validation set in a 70% to 30% ratio. Predictive factors were screened from the training set data using univariate analysis， LASSO regression， and multivariate Logistic regression analysis. Predictive models were then constructed using seven machine learning methods： Logistic regression， decision tree， random forest （RF）， eXtreme gradient boosting， light gradient boosting machine， support vector machine， and artificial neural network. Model performance was evaluated by plotting receiver operating characteristic curve for the validation set and calculating the area under the curve （AUC）， sensitivity， specificity， and Youden index for each model. Furthermore， the SHapley Additive exPlanation （SHAP） method was used to analyze the contribution of variables in the optimal model. ResultsThe prevalence rate of MAFLD among the 2 168 subjects was 12.22% （265/2 168）. Smoking， diastolic blood pressure， phase angle， visceral fat area， muscle fat ratio， waist-to-hip ratio， aspartate aminotransferase， non-HDL-C/HDL-C ratio， triglyceride-glucose index， and gallstones were independent risk factors for MAFLD （all P<0.05）. The seven predictive models of support vector machine， eXtreme gradient boosting， decision tree， light gradient boosting machine， artificial neural network， RF， and Logistic regression had an AUC of 0.738， 0.754， 0.757， 0.786， 0.795， 0.796， and 0.815， respectively， in the validation set， among which the RF model had the best discriminatory ability （AUC=0.796， 95% confidence interval： 0.754 — 0.839）， with a sensitivity of 81.01%， a specificity of 63.16%， and a Youden index of 44.17%. The SHAP analysis showed that visceral fat area， waist-to-hip ratio， and diastolic blood pressure were the top three predictive factors in terms of importance. ConclusionThe RF model， constructed based on body composition and clinical indicators， has a good performance in predicting the risk of MAFLD， and its interpretability can help to identify high-risk individuals in the early stage in clinical practice.

Background Amid global climate change, extreme environmental events are occurring more frequently, and it is imperative to investigate the impacts of combined exposure to fine particluate matter (PM_2.5) and cold spells (CS) on population mortality. Objective To analyze the association between sequential extreme PM_2.5-cold spell (EP-CS) events and non-accidental mortality among residents in Zigong City from 2016 to 2021. Methods Using time-series study design, meteorological data in Zigong were collected from the Zigong Meteorological Bureau for the period from January 1, 2016 to December 31, 2021, while daily non-accidental mortality data were obtained from the mortality surveillance system of the Zigong Center for Disease Control and Prevention. We adopted the percentile method to define extreme PM_2.5 events and cold spells. We analyzed the risk effect of EP-CS events on non-accidental mortality among residents in this city and explored the potential amplification of damage resulting from different patterns of consecutive extreme events by using distributed lag nonlinear model (DLNM). We also conducted stratified analyses based on age, gender, education level, and marital status. Results The EP-CS events demonstrated a significant impact on non-accidental mortality among the local residents, exhibiting a certain lagged effect. The effects on the overall residents lasted from lag0 (RR=1.030, 95%CI: 1.013, 1.048) to lag14 (RR=1.035, 95%CI: 1.019, 1.052). Notably, the effects were more pronounced among females, individuals aged 65 years and above, and those who were never married, divorced, or widowed. Different patterns of EP-CS events all associated with adverse effects, the health impact of EP-CS events was significantly greater than that of individual PM_2.5 pollution or CS events. The analysis of lag effects across different event patterns revealed that the overall effect of EP-CS events with shorter intervals (0–7 d) had a stronger effect compared to EP-CS with longer intervals (8–14 d), and the RR values of lag14 were 1.034 (95%CI: 1.015, 1.054) and 1.017 (95%CI: 1.007, 1.027), suggesting that the damaging effect of compound events occurring in the short term was more significant. Conclusion All sequential extreme EP-CS events have an impact on non-accidental mortality among residents in this city, with compound events demonstrating a stronger effect. Females, individuals aged ≥65 years, and those who were never married, divorced, or widowed are more sensitive to EP-CS events.

Objective To analyze the spatiotemporal distribution characteristics of and trends in the disease burden of dengue fever in China from 2005 to 2024, so as to provide insights into formulation of dengue fever control strategies. Methods Data pertaining to dengue fever cases in China from 2005 to 2024 were retrieved from the Infectious Disease Reporting Information System of Chinese Center for Disease Control and Prevention, and city population, gross domestic product (GDP), GDP per capita, and consumer price index in China were captured from the China Statistical Yearbook, National Bureau of Statistics of China, the China City Statistical Yearbook, and bureaus of statistics in each city. The disability-adjusted life years (DALYs), years of life lost (YLLs), and years lived with disability (YLDs) due to dengue fever were calculated in China from 2005 to 2024. The direct and indirect economic burdens of dengue fever were estimated to calculate the total economic burden. The trends in the disease burden of dengue fever were estimated in China from 2005 to 2024 using a Joinpoint regression model with the software Joinpoint 4.9.0.0, and the average annual percent change (AAPC) and its 95% confidence interval (CI) were calculated. In addition, the DALYs rate and economic burden of dengue fever in China were subjected to global and local spatial autocorrelation analyses using the software ArcGIS 10.8. Results The gross DALYs due to dengue fever were 5 558 person-years in China from 2005 to 2024, and the DALYs of dengue fever increased from 36 person-years in 2005 to 899 person-years in 2024, with an increase of 23.97 folds. The average annual DALYs rate of dengue fever was 0.02 person-years/10⁵ in China during the 20-year study period from 2005 to 2024, and the DALYs rate peaked in 2014 (0.13 person-years/10⁵) and reduced during the COVID-19 pandemic from 2020 to 2022. YLDs were the main contributor of DALYs due to dengue fever in China from 2005 to 2024, with a total of 5 354 person-years, accounting for 96.33% (5 354 person-years/5 558 person-years) of the gross DALYs. The gross DALYs of dengue fever were 2 982 person-years among men (53.66%) and 2 575 person-years among women (46.34%) in China from 2005 to 2024, and high DALYs of dengue fever were measured among residents at ages of 15 to 30 years (1 639 person-years), 30 to 45 years (1 857 person-years), and 45 to 60 years (1 204 person-years), respectively, accounting for 84.56% (4 700 person-years/5 558 person-years) of total DALYs due to dengue fever in China. The total economic burden of dengue fever was estimated to be 612 million Yuan in China from 2005 to 2024, with an average annual economic burden of 30.584 million Yuan. The economic burden of dengue fever increased from 196 000 Yuan in 2005 to 121 million Yuan in 2024 in China, with an increase of 616.35 folds, and the per capita economic burden increased from 3 322.21 Yuan in 2005 to 4 940.01 Yuan in 2024, with an increase of 48.70%. Dengue fever cases were reported in 274 cities (counties) across 31 provinces (autonomous regions, municipalities) in China from 2005 to 2024, with relatively higher DALYs in Guangdong Province and Yunnan Province. Spatial autocorrelation analysis revealed that the disease burden of dengue fever appeared positive aggregation in Chinese cities (counties) from 2005 to 2024 (global Moran’s I = 0.045, Z = 2.24, P < 0.05), with high-high clusters mainly concentrated in the Pearl River Delta region in Guangdong Province and Xishuangbanna Dai Autonomous Prefecture and Pu’er City in Yunnan Province, and the total economic burden (global Moran’s I = 0.032, Z = 9.55, P < 0.001), per capita economic burden (global Moran’s I = 0.208, Z = 27.34, P < 0.001), and the proportion of total economic burdens in GDP in 2024 (global Moran’s I = 0.017, Z = 5.91, P < 0.001) all presented positive aggregation, with relatively higher total economic burdens mainly concentrated in Guangdong Province and Yunnan Province. Joinpoint regression analysis showed that the gross DALYs rates of dengue fever appeared an overall tendency towards a rise in China from 2005 to 2024 (AAPC = 16.24%, P = 0.029), and the DALYs rate presented an overall tendency towards a rise among both men (AAPC = 14.75%, P = 0.028) and women (AAPC = 14.93%, P = 0.037) during the study period. The per capita direct economic burden appeared an overall tendency towards a rise among dengue fever patients in China from 2005 to 2024 (AAPC = 2.16%, P = 0.012); however, there was no significant difference in the trends in the per capita indirect economic burden (AAPC = 0.46%, P = 0.470). In addition, the DALYs rate of dengue fever appeared a tendency towards a rise in 84.67% (232/274) of cities (counties) in China from 2005 to 2024, and the per capita economic burden appeared a tendency towards a rise in 85.40% (234/274) of cities (counties), while the DALYs rate and per capita economic burden of dengue fever appeared a tendency towards a rise in 77.01% (211/274) of cities (counties). Conclusions The disease burden of dengue fever significantly increased in China from 2005 to 2024. It is recommended to reinforce integrated dengue fever control in high-risk areas and among high-risk populations, and to improve the surveillance of imported dengue fever cases and vector control.

OBJECTIVE:Based on different algorithms of machine learning,the prediction model of lumbar disc herniation has become a trend and hot spot in the development of precision medicine.However,there is limited evidence on the reporting quality and methodological quality of prediction models of lumbar disc herniation outcomes using machine learning.This article is aimed to explore the performance of machine learning algorithms in predicting the prognosis of lumbar disc herniation by comprehensively analyzing the report quality and risk of bias of previous studies that developed and validated prognosis prediction models based on machine learning through a comprehensive literature search,in order to explore the performance of machine learning algorithms in predicting the prognosis of lumbar disc herniation.METHODS:The databases of CNKI,WanFang,VIP,SinOMED,PubMed,Web of Science,Embase,and The Cochrane Library were searched by computer.Studies on the use of machine learning to develop(and/or validate)prognostic prediction models for lumbar disc herniation were collected from the inception of the database to December 31,2023.Two researchers independently screened the literature,extracted data,and assessed the risk of bias of the included studies.The reporting quality and risk of bias of the included studies were assessed by the Multivariable Transparent Reporting of Predictive Models(TRIPOD)statement and the Predictive Model Risk of Bias Assessment Tool(PROBAST).The results of the evaluation were analyzed using descriptive statistics and visual charts.RESULTS:(1)A total of 23 articles were included,and the TRIPOD compliance of each study ranged from 11％to 87％,with a median compliance of 54％.The quality of reporting of titles,detailed descriptions of treatment measures,blinding of predictors,handling of missing data,details of risk stratification,specific procedures for enrollment,model interpretation,and model performance was mostly poor,with TRIPOD adherence rates ranging from 4％to 35％.(2)Of all included studies,61％had a high risk of bias and 39％had an unclear overall risk of bias.The area under the curve,accuracy,sensitivity and specificity were used to evaluate the performance of the model.The areas under the curve of 20 models were reported,ranging from 0.561 to 0.999.Three models reported the accuracy of the model,ranging from 82.07％to 89.65％.(3)Among all included studies,the statistical analysis domain was most often assessed as having a high risk of bias,mainly due to the small number of valid samples,the selection of predictors based on univariate analysis and the lack of calibration and discrimination assessment of the model in the study.CONCLUSION:These results indicate that machine learning can achieve good predictive ability in the development and validation of prognostic models for lumbar disc herniation.The commonly used algorithms include regression algorithm,support vector machine,decision tree,random forest,artificial neural network,naive Bayes and other algorithms.Reasonable algorithms combined with clinical practice can improve the accuracy of prognosis prediction of lumbar disc herniation.However,the reporting and methodological quality of prognosis prediction models based on machine learning are poor,the prediction performance of different models varies greatly,and the generalization and extrapolation of research models are unclear.There is an urgent need to improve the design,implementation and reporting of such studies.To promote the application of machine learning in the clinical practice of lumbar disc herniation prediction models,it is necessary to comprehensively consider various predictors related to the prognosis of the disease before modeling,and strictly follow the relevant standards of PROBAST tool during modeling.