Objective:To establish machine learning models to predict visual improvement and anatomical response after anti-vascular endothelial growth factor (VEGF) treatment in patients with diabetic macular edema (DME).Methods:A multi-algorithm machine learning predictive modeling study based on retrospective clinical data was conducted.A total of 225 patients with DME who received their first intravitreal anti-VEGF injection at Fuzhou University Affiliated Provincial Hospital were enrolled between January 2023 and April 2025.According to data completeness, 204 cases were included in the visual recovery prediction model and 201 cases were included in the anatomical response prediction model.Baseline data included optical coherence tomography (OCT) features and blood biomarkers.The primary outcomes were defined as an improvement of ≥1 line in visual acuity and a reduction of ≥20% in central retinal thickness (CRT) after anti-VEGF treatment.Feature selection was performed using univariate logistic regression and Lasso regression.Four machine learning algorithms, logistic regression (LR), decision tree, multilayer perceptron, and random forest, were trained and validated.Model performance was evaluated using accuracy, sensitivity, specificity, area under the receiver operating characteristic curve (AUC), and decision curve analysis.The best-performing model was further interpreted using SHAP analysis, and a nomogram was constructed for clinical application.This study adhered to the Declaration of Helsinki.The study protocol was approved by the Ethics Committee of Fuzhou University Affiliated Provincial Hospital (No.K2025-03-064).Written informed consent was obtained from each subject.Results:Among the 37 baseline variables, five key predictors were identified for the outcome of ≥20% CRT reduction: baseline CRT, baseline CRT ≥400 μm, presence of subretinal fluid (SRF), disorganization of the retinal inner layers (DRIL), and integrity of the ellipsoid zone (EZ).Among the four models, the LR model had the best performance, with an accuracy of 0.88, sensitivity of 0.94, specificity of 0.70, and an AUC of 0.94 (95% confidence interval [ CI]: 0.87-1.00).SHAP analysis showed that baseline CRT ≥400 μm, DRIL, SRF, and baseline CRT contributed positively to the outcome, while EZ integrity was a negative predictor for CRT reduction.For the outcome of ≥1-line visual improvement, two key predictors were identified: baseline best corrected visual acuity (BCVA) and EZ integrity.Both baseline BCVA and EZ integrity were negative predictors for ≥1-line visual improvement.The LR model also had the best performance in the internal validation cohort, with an accuracy of 0.71, sensitivity of 0.67, specificity of 0.75, and an AUC of 0.76 (95% confidence interval [ CI]: 0.61-0.91).A visual nomogram was developed based on the selected predictors and the best-performing model.By converting patient-specific clinical characteristics into scores, clinicians can calculate a total score and estimate the probability of achieving a reduction of ≥20% in CRT and a ≥1-line improvement in visual acuity after anti-VEGF therapy. Conclusions:Machine learning-based model building can effectively predict visual and anatomical response following anti-VEGF treatment in DME patients.Logistic regression shows robust predictive performance for both outcomes.Identification of key predictors, especially OCT features such as EZ integrity, SRF, and DRIL, may aid in guiding treatment expectation assessment and personalized intervention strategies.Nomogram constructed in this study shows good clinical applicability and may serve as a decision-support tool to improve the precision of DME management.

Objective:To establish machine learning models to predict visual improvement and anatomical response after anti-vascular endothelial growth factor (VEGF) treatment in patients with diabetic macular edema (DME).Methods:A multi-algorithm machine learning predictive modeling study based on retrospective clinical data was conducted.A total of 225 patients with DME who received their first intravitreal anti-VEGF injection at Fuzhou University Affiliated Provincial Hospital were enrolled between January 2023 and April 2025.According to data completeness, 204 cases were included in the visual recovery prediction model and 201 cases were included in the anatomical response prediction model.Baseline data included optical coherence tomography (OCT) features and blood biomarkers.The primary outcomes were defined as an improvement of ≥1 line in visual acuity and a reduction of ≥20% in central retinal thickness (CRT) after anti-VEGF treatment.Feature selection was performed using univariate logistic regression and Lasso regression.Four machine learning algorithms, logistic regression (LR), decision tree, multilayer perceptron, and random forest, were trained and validated.Model performance was evaluated using accuracy, sensitivity, specificity, area under the receiver operating characteristic curve (AUC), and decision curve analysis.The best-performing model was further interpreted using SHAP analysis, and a nomogram was constructed for clinical application.This study adhered to the Declaration of Helsinki.The study protocol was approved by the Ethics Committee of Fuzhou University Affiliated Provincial Hospital (No.K2025-03-064).Written informed consent was obtained from each subject.Results:Among the 37 baseline variables, five key predictors were identified for the outcome of ≥20% CRT reduction: baseline CRT, baseline CRT ≥400 μm, presence of subretinal fluid (SRF), disorganization of the retinal inner layers (DRIL), and integrity of the ellipsoid zone (EZ).Among the four models, the LR model had the best performance, with an accuracy of 0.88, sensitivity of 0.94, specificity of 0.70, and an AUC of 0.94 (95% confidence interval [ CI]: 0.87-1.00).SHAP analysis showed that baseline CRT ≥400 μm, DRIL, SRF, and baseline CRT contributed positively to the outcome, while EZ integrity was a negative predictor for CRT reduction.For the outcome of ≥1-line visual improvement, two key predictors were identified: baseline best corrected visual acuity (BCVA) and EZ integrity.Both baseline BCVA and EZ integrity were negative predictors for ≥1-line visual improvement.The LR model also had the best performance in the internal validation cohort, with an accuracy of 0.71, sensitivity of 0.67, specificity of 0.75, and an AUC of 0.76 (95% confidence interval [ CI]: 0.61-0.91).A visual nomogram was developed based on the selected predictors and the best-performing model.By converting patient-specific clinical characteristics into scores, clinicians can calculate a total score and estimate the probability of achieving a reduction of ≥20% in CRT and a ≥1-line improvement in visual acuity after anti-VEGF therapy. Conclusions:Machine learning-based model building can effectively predict visual and anatomical response following anti-VEGF treatment in DME patients.Logistic regression shows robust predictive performance for both outcomes.Identification of key predictors, especially OCT features such as EZ integrity, SRF, and DRIL, may aid in guiding treatment expectation assessment and personalized intervention strategies.Nomogram constructed in this study shows good clinical applicability and may serve as a decision-support tool to improve the precision of DME management.

Objective To establish an animal model of dampness-syndrome in mice (single model) and evaluate its characteristics of dampness-syndrome. The above-mentioned mice with dampness syndrome were used to construct mice model of ischemic stroke (double model) and observe the effect of dampness-pathogenic on the outcome of stroke. Methods Healthy C57BL/6J male mice were randomly divided into dampness-syndrome (including sham-surgery group and ischemic stroke group,with 10 mice in each group) and non dampness-syndrome groups (including sham-surgery group and ischemic stroke group,with 10 mice in each group). The dampness-syndrome group was fed with high-fat diet and the non dampness-syndrome group was fed with normal diet for 12 weeks. After the mice model of dampness-syndrome was successfully established,transient middle cerebral artery occlusion/reperfusion (tMCAO/R) surgery was used to replicate an ischemic stroke mice model. Evaluation indicators for dampness-syndrome mice model:the general status including body weight,morphology,posture,activity status,and physical characteristics,the histopathological observation of the aorta (oil red O staining,Masson-trichrome staining) and liver (HE staining,oil red O staining),electron microscopic observation of the tongue tissue (scanning electron microscopy,electron microscopy),blood lipid levels[total cholesterol(TC),triglycerides(TG)]and liver coefficient. Evaluation indicators for ischemic stroke mice model:neurological function score and the cerebral infarction volume ratio. Results Compared with the non dampness-syndrome group,the mice in the dampness-syndrome group showed an increased in body weight,poor hair color,sparse hair,fatigue and laziness,mental atrophy,anorexia and lethargy. It was observed that the aortic lumen was narrowed,the intima was significantly thickened,lipid plaque deposition was increased,and foam cells were visible. A large amount of red lipid droplets appeared in liver cells. There were obvious lipid infiltration and diffuse steatosis. Increased keratosis of the mucosal layer of tongue tissue,the thicker stratum corneum,lipofuscin,and bacteria on the tongue surface were found. Serum TG and TC levels significantly increased(P<0.01),and the liver coefficient significantly decreased (P<0.001). Compared with non dampness-syndrome group (sham-surgery group),neurological function score and the cerebral infarction volume ratio in dampness-syndrome ischemic stroke group obviously increased (P<0.001). Conclusion High-fat feeding for 12 weeks combined with tMCAO/R modeling can successfully establish a mice model with dampness-syndrome ischemic stroke,and the neurological function score and cerebral infarction volume in the dampness-syndrome ischemic stroke group was more severe than that in the non dampness-syndrome ischemic stroke group.

Objective To investigate the mechanism of retinal injury in rats caused by light-emitting diodes(LEDs)with different color rendering indexes(CRIs).Methods Totally 20 Sprague-Dawley rats were randomly divided into nor-mal control(NC)group(sunlight),low CRI(CRI-L)group(blue light),medium CRI(CRI-M)group(conventional LED),and high CRI(CRI-H)group(full-spectrum LED),with 5 rats in each group,exposed to light for 12 hours daily for 4 consecutive weeks.Hematoxylin & eosin staining was used to assess morphological changes in the retina.Dihydroethidi-um staining was employed to detect the levels of reactive oxygen species(ROS)in retinal tissues.The messenger ribonu-cleic acid(mRNA)expressions of NOD-like receptor family pyrin domain containing protein 3(NLRP3),Gasdermin D(GSDMD)and Caspase-1 were analyzed by real-time quantitative polymerase chain reaction(RT-qPCR),and their protein expressions were measured through immunohistochemical staining.Environmental light spectra were measured using a spectroradiometer.Results Rats in the CRI-L group showed the thinnest retina,followed by the CRI-M group and CRI-H group.The fluorescence intensity of ROS in the NC group,CRI-L group,CRI-M group and CRI-H group was 1.000±0.046,25.060±1.732,14.530±3.776 and 1.821±0.587,respectively.The ROS level in the CRI-H group was significantly lower than that in the CRI-L group and CRI-M group(both P＜0.05).RT-qPCR showed that the relative mRNA expression of NL-RP3 in the NC group,CRI-L group,CRI-M group and CRI-H group was 1.004±0.005,4.004±0.716,2.027±0.303 and 0.741±0.069,respectively;the relative mRNA expression of Caspase-1 was 1.010±0.006,4.337±0.345,2.268±0.058 and 0.713±0.021,respectively;the relative mRNA expression of GSDMD was 1.000±0.000,2.938±0.559,1.955±0.166 and 1.213±0.051,respectively.Compared with the NC group,the relative expressions of NLRP3,Caspase-1 and GSDMD in the CRI-L group and CRI-M group significantly increased(all P＜0.05).The immunohistochemical staining results showed that the fluorescence intensity of NLRP3 in the retina of rats in the NC group,CRI-L group,CRI-M group and CRI-H group was 0.379 4±0.002 2,0.400 7±0.011 4,0.379 0±0.006 9 and 0.377 0±0.007 5,respectively;the fluorescence intensity of Caspase-1 was 0.367 2±0.005 8,0.442 6±0.041 1,0.382 4±0.011 9 and 0.380 6±0.006 5,respectively;the fluorescence intensity of GSDMD was 0.159 5±0.013 4,0.167 5±0.011 9,0.397 6±0.014 3 and 0.377 2±0.022 8,respec-tively.Compared with the NC group,rats in the CRI-L group showed increased fluorescence intensity of NLRP3 and Caspase-1,and rats in the CRI-M and CRI-H showed increased fluorescence intensity of GSDMD(all P＜0.05).The spec-tral comparison revealed that the CRI-H group had a broader spectral coverage and a distribution closer to natural light spectra.Conclusion Conventional LED exposure can induce a decrease in retinal thickness,upregulate the ROS expres-sion in retinal tissues,and increase the expression levels of NLRP3,Caspase-1 and GSDMD.High CRI full-spectrum LEDs can mitigate pyroptosis through the ROS/NLRP3 pathway by optimizing their spectral distribution,offering better biosafety.