ObjectiveThe widespread adoption of portable fundus cameras for primary care and community screening is hindered by limitations in current autofocus(AF) technologies. Image-based methods relying on sharpness evaluation require iterative searches, resulting in slow convergence, while projection-based techniques are susceptible to optical artifacts and calibration errors. To address these challenges, this study introduces a novel AF system based on direct wavefront sensing, designed to deliver simultaneous high speed, high precision, and operational robustness within the compact form factor essential for portable ophthalmic devices. MethodsOur approach fundamentally reimagines the AF process by directly measuring the ocular wavefront aberration. We developed a custom portable fundus camera integrating a miniaturized Shack-Hartmann wavefront sensor (SHWS) into the optical path. An 850 nm laser diode projects a point source onto the retina via oblique illumination to minimize corneal reflections. Light scattered from this spot carries the eye’s refractive error through the imaging optics and is directed to the SHWS, positioned at a plane optically conjugate to the primary color CMOS imaging sensor. A microlens array within the SHWS samples the incident wavefront, generating a pattern of focal spots on a CCD. Real-time centroid analysis of these spots provides a map of local wavefront slopes. These measurements are processed through a singular value decomposition (SVD) algorithm to fit a Zernike polynomial basis set, enabling real-time reconstruction of the wavefront phase. The defocus component (S) is extracted from the second-order Zernike coefficients, providing a direct, quantitative measure of the refractive error in diopters. This value serves as a precise error signal in a closed-loop control system, which commands a voice-coil actuated focusing lens to its null position in a single, deterministic step, eliminating the need for iterative search algorithms. ResultsComprehensive evaluation demonstrated the system’s high performance. Testing on a calibrated model eye (OEMI-7) established a highly linear relationship between the computed defocus S and the focusing lens position across a ±20 Diopter (D) compensation range, achievable within a 5 mm mechanical travel. The system achieved a focusing precision of 0.08 D, corresponding to an 18-fold improvement over a conventional projection spot-size method tested under identical conditions. The total focus acquisition time, encompassing wavefront measurement, computation, and lens actuation, averaged under 0.5 s. Clinical validation with 25 human volunteers (50 eyes, refractive range -15 D to +10 D) confirmed practical efficacy. The wavefront-sensing AF succeeded in 92% of attempts with a mean time of 0.5 s, substantially outperforming a projection-based benchmark which achieved only a 32% success rate with an average time of 4.25 s. The system provided instantaneous directional guidance and maintained stability during minor ocular movements. Objective assessment of image quality, via amplitude contrast of retinal vasculature, showed consistent and significant enhancement following AF correction across the entire tested diopter range. ConclusionThis work successfully implements and validates a direct wavefront-sensing autofocus paradigm for portable fundus cameras. By directly quantifying and compensating for the optical defocus aberration, this method bypasses the fundamental limitations of image-processing and projection-based techniques, enabling rapid, precise, and deterministic diopter compensation. The developed system delivers an exceptional combination of a wide operational range (±20 D), high accuracy (0.08 D), fast convergence (0.5 s), and a compact physical footprint. This technology provides a practical and high-performance focusing solution capable of enhancing the reliability, throughput, and diagnostic utility of portable retinal imaging in large-scale screening applications. Future efforts will be directed towards system cost optimization and performance adaptation for diverse ocular conditions.

ObjectivePost-ischemic acute inflammation and the subsequent persistent dysregulation of the immune microenvironment represent major pathological drivers that aggravate neuronal injury and severely restrict functional recovery following ischemic stroke. Although current reperfusion therapies partially restore blood flow, they fail to effectively modulate the secondary inflammatory cascade and oxidative stress, which remain critical barriers to neurological restoration. To address this challenge, this study aimed to engineer and systematically evaluate a biomimetic nanosystem composed of transforming growth factor-β1 (TGF-β1)-loaded platelet membrane-camouflaged lipid nanoparticles (PLP). This nanosystem was designed to achieve dual lesion-targeted delivery and immune microenvironment remodeling. By verifying its spatiotemporal accumulation, anti-inflammatory activity, and neuroprotective efficacy, we sought to establish an integrated therapeutic strategy that simultaneously enables lesion targeting, immune regulation, and functional recovery after ischemic injury. MethodsThe physicochemical properties of PLP, including hydrodynamic particle size, zeta potential, structural stability, and morphology, were characterized using dynamic light scattering, zeta potential analysis, and transmission electron microscopy. The preservation of platelet membrane-derived adhesion and immunoregulatory proteins was confirmed by SDS-PAGE through comparative analysis of protein band profiles between PLP and native platelet membranes. The in vitro biological activities of PLP were evaluated using two complementary cellular models. LPS-induced M1-polarized RAW264.7 macrophages were employed to assess inflammatory modulation, while oxygen glucose deprivation/reperfusion (OGD/R)-induced BV2 microglial cells and SH-SY5Y neuronal cells were utilized to investigate neuroinflammatory regulation and neuronal protection. For in vivo validation, a transient middle cerebral artery occlusion (tMCAO) mouse model was established to mimic ischemia-reperfusion injury. The spatiotemporal biodistribution and lesion-targeting capability of the PLP were monitored through live fluorescence imaging. Therapeutic efficacy was comprehensively evaluated by triphenyltetrazolium chloride (TTC) staining, glial fibrillary acidic protein (GFAP) immunofluorescence analysis, body weight monitoring, and neurological severity score (NSS) assessment. ResultsPLP nanoparticles displayed a uniform spherical morphology, nanoscale particle size distribution, and stable negative surface charge, indicating favorable colloidal stability and circulation potential. SDS-PAGE results confirmed the effective retention of key platelet membrane proteins associated with endothelial adhesion, immune evasion, and inflammatory regulation, demonstrating the successful biomimetic construction. Optimal therapeutic concentrations were determined in OGD/R-induced BV2 cells, where PLP exhibited excellent cytocompatibility and anti-inflammatory activity.In vitro experiments demonstrated that PLP significantly inhibited the polarization of RAW264.7 macrophages toward the pro-inflammatory M1 phenotype and markedly reduced neuronal apoptosis under ischemia-reperfusion conditions. In vivo fluorescence imaging revealed that PLP rapidly accumulated in the ischemic brain hemisphere and maintained prolonged retention for up to 7 d, suggesting enhanced lesion-specific targeting and sustained drug release. Compared with control group, PLP treatment significantly reduced cerebral infarct volume, attenuated reactive astrogliosis, improved weight recovery, and accelerated neurological functional restoration, as reflected by significantly improved NSS scores. ConclusionThis study establishes a multifunctional biomimetic nanoplatform that integrates platelet membrane-mediated active targeting with the anti-inflammatory, antioxidative, and neuroprotective properties of TGF-β1. The PLP system enables rapid lesion homing and long-term retention while synergistically regulating the post-stroke inflammatory microenvironment by suppressing pro-inflammatory immune activation, reducing neuronal apoptosis, and limiting excessive astrocyte reactivity. Importantly, this study proposes a conceptually therapeutic paradigm that combines targeted delivery with immune microenvironment remodeling to achieve comprehensive neurovascular protection. These findings provide strong experimental evidence supporting the translational potential of biomimetic nanotherapeutics as next-generation precision interventions for ischemic stroke.

Objective To explore the prevalence of depressive symptoms in postoperative patients with ovarian cancer and to analyze its influencing factors from multiple dimensions, including clinical characteristics, psychological factors, and laboratory indicators. Methods A cross-sectional study was conducted, which enrolled 235 postoperative patients with ovarian cancer. Depressive status was assessed using the patient health questionnaire, and the demographic, pathological, and medical record data of the patients were collected using the generalized anxiety disorder scale, Pittsburgh sleep quality index, European organization for research and treatment of cancer quality of life questionnaire core 30, and ECOG performance status score. Peripheral blood tumor marker (CA125), routine blood test, lymphocyte subsets, and serum cytokine levels were measured. Univariate and multivariate binary logistic regression analysis were used for statistical analysis. Results The prevalence of depression in postoperative patients with ovarian cancer was 39.15% (92/235). Univariate analysis showed that ECOG score ≥ 2 points, pain, anxiety, poor sleep quality, low quality of life, low life satisfaction, tumor recurrence, six or more cycles of chemotherapy, as well as higher levels of CA125, NLR, and NAR, and lower hemoglobin levels were significantly associated with depression (all P<0.05). Multivariate binary Logistic regression analysis showed that anxiety (OR=1.975, 95%CI: 1.231-3.170), sleep efficiency (OR=4.181, 95%CI: 1.211-14.43), sleep latency (OR=34.806, 95%CI: 4.258-284.542), ECOG performance status score, cognitive function (OR=0.918, 95%CI: 0.868-0.97), and life satisfaction were independent risk factors for depression (all P<0.05). Laboratory indicators were not independent influencing factors in the multivariate Logistic regression model. Conclusion Depression in postoperative patients with ovarian cancer is influenced by physiological, psychological, and social factors. Clinical management should focus on patients with anxiety, sleep disorders, poor physical condition, and low life satisfaction, and a comprehensive prevention and treatment strategy centered on psychological intervention and taking into account symptom management and social support should be implemented.

Objective To explore the safety and feasibility of 5G remote robot-assisted thoracoscopic anatomic lobectomy and segmentectomy. Methods A retrospective analysis was conducted on patients who underwent 5G remote robot-assisted thoracoscopic anatomic lobectomy or segmentectomy between July and September 2024. The surgeries were performed collaboratively by Shanghai Chest Hospital, Affiliated Hospital of Yangzhou University, and The First Affiliated Hospital of Nanchang University. Perioperative outcomes were recorded and analyzed. Results Ten patients were included, comprising 4 anatomic lobectomies and 6 segmentectomies. The median age was 61.0 (56.0, 72.0) years. The median intraoperative bi-directional network latency was 52.5 (39.0, 54.0) ms. There were no instances of network interruption, robot-related adverse events, or conversions to open thoracotomy. The median postoperative chest tube drainage volume was 500.0 (375.0, 600.0) mL, and the median hospital stay was 5.5 (4.0, 6.0) days. No complications of Clavien-Dindo grade Ⅱor higher occurred. All patients were recurrence-free and alive during the 6-month postoperative follow-up period. Conclusion 5G remote robot-assisted thoracoscopic surgery demonstrates high safety and operational stability. This technology shows promising potential for clinical application and is worthy of further development and utilization.

Objective To analyze the postoperative complications and survival outcomes of elderly patients with renal cell carcinoma undergoing laparoscopic surgery, and to identify the influencing factors, so as to provide reference for improving the safety of laparoscopic surgery in such patients. Methods A retrospective analysis was conducted on the clinical, surgical, and pathological data of 153 patients aged ≥75 years who underwent laparoscopic nephrectomy for renal cell carcinoma at our hospital during 2012 and 2025.Logistic regression analyses were performed to identify factors influencing postoperative complications, and a nomogram was constructed and validated to predict the complications.Kaplan-Meier method was used to draw the survival curve, and the effects of different surgical approaches on the prognosis and the survival difference in the age subgroups (75-＜80 years, ≥80 years) were analyzed.Cox regression analyses were conducted to determine factors affecting survival outcomes. Results Among the 153 patients, 15 developed postoperative complications, including 5 severe cases. Multivariate logistic regression identified Charlson Comorbidity Index (CCI) ≥3 and prolonged operation time as independent risk factors for postoperative complications (P＜0.05).The C-index of the logistic regression model constructed based on these factors (age, CCI≥3 and operation time) was 0.77 (95%CI:0.64-0.90), with an average absolute error of 1.6%.Subgroup analysis revealed superior overall survival in patients aged 75-＜80 years undergoing partial nephrectomy compared to radical nephrectomy (P=0.005 6).No statistically significant difference in overall survival was observed among patients ≥80 years who received different surgical approaches (P=0.062).Multivariate Cox regression analysis identified preoperative low albumin levels, preoperative renal insufficiency, tumor diameter＞4 cm, and pathological stage as independent risk factors for overall survival (P＜0.05). Conclusion Laparoscopic surgery is safe and effective for elderly patients with renal cell carcinoma, with survival outcomes correlated to nutritional status, renal function, tumor size and pathological staging.Patients aged ≥80 years still derive surgical benefit, and no survival differences exist between different surgical approaches in this age group.

BACKGROUND:Spleen has the functions of blood storage,hematopoiesis,and immunity.With the increase of age,the structural degeneration and functional decline of spleen lead to the impairment of immune system function,thus accelerating the aging process of the body.The treatment of spleen aging in tree shrews with highly active umbilical cord mesenchymal stem cells has not been reported. OBJECTIVE:To explore the intervention effect and mechanism of highly active umbilical cord mesenchymal stem cells on spleen aging in tree shrews. METHODS:Highly active umbilical cord mesenchymal stem cells were isolated,cultured,and obtained from the umbilical cord tissue of newborn tree shrews by caesarean section.The differentiation abilities of adipogenesis,osteogenesis,and chondrogenesis were detected by three-line differentiation kit.Cell cycle and surface markers were detected by flow cytometry.The second generation of highly active umbilical cord mesenchymal stem cells were transfected with Genechem Green Fluorescent Protein with infection complex values of 100,120,140,160,180,and 200,respectively,to screen the best transfection conditions.After transfection,the fourth generation of highly active umbilical cord mesenchymal stem cells was injected into the tail vein of tree shrews in the elderly treatment group.The young control group and the aged model group were not given special treatment.After 4 months of treatment,the spleen tissue was taken and the structure of the spleen was observed by hematoxylin-eosin staining.β-Galactosidase staining was used to detect the activity of aging-related galactosidase.Immunohistochemical staining was used to detect the expression levels of p21 and p53 proteins.Ki67 and PCNA immunofluorescence staining was used to detect cell proliferation activity.Immunofluorescence staining was used to detect the expression levels of spleen autophagy protein molecules Beclin 1 and APG5L/ATG5.Reactive oxygen species fluorescence staining was used to detect the content of reactive oxygen species in spleen tissue.CD3 immunofluorescence staining was used to detect the change of the proportion of total T lymphocytes.The secretion levels of interleukin 1β and transforming growth factor β1 in spleen were detected by enzyme linked immunosorbent assay.The distribution of highly active umbilical cord mesenchymal stem cells labeled with green fluorescent protein in spleen tissue was observed by DAPI double staining of nucleus. RESULTS AND CONCLUSION:(1)Highly active umbilical cord mesenchymal stem cells grew in a short spindle shape with fish-like growth,with a large proportion of G0/G1 phase,and had the potential to differentiate into adipogenesis,osteogenesis,and chondrogenesis.(2)Multiplicity of infection=140 and transfection for 72 hours were the best conditions for labeling tree shrews highly active umbilical cord mesenchymal stem cells with Genechem Green Fluorescent Protein.(3)Compared with the aged model group,in the aged treatment group,the spleen tissue cells of tree shrews were arranged closely,and the area of white pulp was increased(P<0.01);the boundary between red pulp and white pulp was clear;the proportion of germinal centers did not show statistically significant difference(P>0.05).The activity level of galactosidase related to spleen tissue aging was decreased(P<0.001),and the expression levels of aging protein molecules p21 and p53 were down-regulated(P<0.001).The expression levels of proliferation-related molecules Ki67 and PCNA were up-regulated(P<0.001,P<0.05);expression levels of autophagy-related molecules Beclin 1 and APG5L/ATG5 were up-regulated(P<0.001),and the content of reactive oxygen species decreased(P<0.001),and the proportion of CD3+T cells increased(P<0.05).The secretion level of interleukin 1β in the aging-related secretion phenotype decreased(P<0.001);no significant difference was found in transforming growth factor β1 level(P>0.05).Compared with the young control group,the above indexes were significantly different in the elderly treatment group(P<0.05).(4)Green fluorescent cells labeled with green fluorescent protein were observed in spleen tissue of tree shrews the elderly treatment group by frozen tissue section observation.The results show that intravenous infusion of highly active umbilical cord mesenchymal stem cells can migrate to spleen tissue,inhibit the production of reactive oxygen species,down-regulate the expression of aging-related proteins,induce autophagy,promote cell proliferation,reduce chronic inflammation,and then improve the structure and function of spleen tissue.

ObjectiveBased on ultra performance liquid chromatography-quadrupole-electrostatic field orbital trap high-resolution mass spectrometry（UPLC-Q-Orbitrap-MS）， to identify the in vivo and in vitro chemical components of total phenolic acids in Gei Herba（TPAGH）， and to clarify the pharmacological effects and potential mechanisms of the effective part in promoting acute wound healing and inhibiting scar formation. MethodsUPLC-Q-Orbitrap-MS was used to identify the chemical components of TPAGH and ingredients absorbed in vivo after topical administration. A total of 120 ICR mice were randomly divided into the model group， recombinant human epidermal growth factor（rhEGF） group（4 mg·kg^-1）， and low， medium， and high dose groups of TPAGH（3.5， 7， 14 mg·kg^-1）， with 24 mice in each group. A full-thickness skin excision model was constructed， and each administration group was coated with the drug at the wound site， and the model group was treated with an equal volume of normal saline， the treatment was continued for 30 days， during which 8 mice from each group were sacrificed on days 6， 12， and 30. The healing of the wounds in the mice was observed， and histopathological changes in the skin tissues were dynamically observed by hematoxylin-eosin（HE）， Masson， and Sirius red staining， and enzyme-linked immunosorbent assay（ELISA） was used to dynamically measure the contents of interleukin-6（IL-6）， tumor necrosis factor-α（TNF-α）， vascular endothelial growth factor A（VEGFA）， matrix metalloproteinase（MMP）-3 and MMP-9 in skin tissues. Network pharmacology was used to predict the targets related to the promotion of acute wound healing and the inhibition of scar formation by TPAGH， and molecular docking of key components and targets was performed. Gene Ontology（GO） biological process analysis and Kyoto Encyclopedia of Genes and Genomes（KEGG） pathway enrichment analysis were carried out for the related targets， so as to construct a network diagram of herbal material-compound-target-pathway-pharmacological effect-disease for further exploring its potential mechanisms. ResultsA total of 146 compounds were identified in TPAGH， including 28 phenylpropanoids， 31 tannins， 23 triterpenes， 49 flavonoids， and 15 others， and 16 prototype components were found in the serum of mice. Pharmacodynamic results showed that， compared with the model group， the TPAGH groups showed a significant increase in relative wound healing rate and relative scar inhibition rate（P<0.05）， and the number of new capillaries， number of fibroblasts， number of new skin appendages， epidermal regeneration rate， collagen deposition ratio， and Ⅲ/Ⅰ collagen ratio in the tissue were significantly improved（P<0.05， 0.01）， the levels of IL-6， TNF-α， MMP-3 and MMP-9 in the skin tissues were reduced to different degrees， while the level of VEGFA was increased. Network pharmacology analysis screened 10 core targets， including tumor protein 53（TP53）， sarcoma receptor coactivator（SRC）， protein kinase B（Akt）1， signal transducer and activator of transcription 3（STAT3）， epidermal growth factor receptor（EGFR） and so on， participating in 75 signaling pathways such as advanced glycation end-products（AGE）-receptor for AGE（AGE/RAGE） signaling pathway， phosphatidylinositol 3-kinase（PI3K）/Akt signaling pathway， mitogen-activated protein kinase（MAPK） signaling pathway. Molecular docking confirmed that the key components genistein， geraniin， and casuariin had good binding ability to TP53， SRC， Akt1， STAT3 and EGFR. ConclusionThis study comprehensively reflects the chemical composition of TPAGH and the absorbed components after topical administration through UPLC-Q-Orbitrap-MS. TPAGH significantly regulates key indicators of skin healing and tissue reconstruction， thereby clarifying its role in promoting acute wound healing and inhibiting scar formation. By combining in vitro and in vivo component identification with network pharmacology， the study explores how key components may bind to targets such as TP53， Akt1 and EGFR， exerting therapeutic effects through related pathways such as immune inflammation and vascular regeneration.

Background Per- and polyfluoroalkyl substances (PFAS) are a class of persistent organic pollutants widely used in various products, leading to population exposure and long-term accumulation. At present, there is a lack of research on the relationships between pre-pregnancy PFAS and menstrual characteristics among women undergoing assisted reproductive technology (ART) in China. Objective To explore the relationships between pre-pregnancy PFAS exposure among women undergoing ART and menstrual characteristics prior to assisted reproductive treatment. Methods This study employed a cross-sectional research design, recruiting women undergoing ART treatment at the Reproductive Clinic of the International Peace Maternity & Child Health Hospital, School of Medicine, Shanghai Jiao Tong University, from 2017 to 2020 as study participants. Ultra-high-performance liquid chromatography-tandem mass spectrometry was used to detect 42 types of PFAS in pre-pregnancy serum samples. Questionnaires were administered to collect information on demographic characteristics, lifestyle habits, and menstrual characteristics (average menstrual cycle length, average menstrual period length, menstrual irregularities, and menstrual bleeding volume) of women undergoing ART. Multiple linear regression, binary logistic regression, and multinomial logistic regression analyses were performed to investigate the relationships between individual PFAS exposure before pregnancy and menstrual characteristics among ART women. Additionally, weighted quantile sum (WQS) model was applied to analyze the association between PFAS mixtures and menstrual characteristics. Results In the pre-pregnancy serum samples of the study population, 15 PFAS were detected in more than 60% of the samples, including perfluoroheptanoic acid (PFHpA), perfluorooctanoic acid (PFOA), perfluorononanoic acid (PFNA), perfluorodecanoic acid (PFDA), perfluoroundecanoic acid (PFUnDA), perfluorododecanoic acid (PFDoDA), perfluorobutanesulfonic acid (PFBS), perfluorohexanesulfonic acid (PFHxS), perfluoroheptanesulfonic acid (PFHpS), perfluorooctanesulfonic acid (PFOS), 6:2 chlorinated polyfluorinated ether sulfonate (6:2 Cl-PFESA), 8:2 chlorinated polyfluorinated ether sulfonate (8:2 Cl-PFESA), perfluoro-2-propoxypropanoic acid (HFPO-DA), perfluoro-2-methoxyacetic acid (PFMOAA), and perfluoro-(3,5,7,9,11-pentaoxadodecanoic) acid (PFO5DoDA). Among them, PFOA had the highest median concentration of 9.160 ng·mL⁻¹. The single PFAS exposure analysis revealed a positive correlation between PFAS and irregular menstrual cycles. Specifically, for every natural-log unit (e) increase in PFOA, PFBS, or PFHxS level, the incidence of irregular menstrual cycles increased by 57%, 42%, or 39%, respectively. Most PFAS were positively correlated with the average number of menstrual cycle days, such as PFHpA (b=1.08, 95%CI: 0.11, 2.05), PFOA (b=1.69, 95%CI: 0.39, 3.00), PFBS (b=1.23, 95%CI: 0.25, 2.22), PFHxS (b=1.47, 95%CI: 0.61, 2.32), PFHpS (b=1.48, 95%CI: 0.35, 2.61), and 6:2 Cl-PFESA (b=0.90, 95%CI: 0.08, 1.72). Furthermore, levels of PFHpA (OR=1.39, 95%CI: 1.06, 1.82), PFOA (OR=1.58, 95%CI: 1.09, 2.30), PFBS (OR=1.37, 95%CI: 1.04, 1.80), PFHxS (OR=1.34, 95%CI: 1.05, 1.71), PFHpS (OR=1.53, 95%CI: 1.10, 2.14), and 6:2 Cl-PFESA (OR=1.34, 95%CI: 1.06, 1.70) were positively correlated with low menstrual blood volume, while PFOA (OR=0.40, 95%CI: 0.23, 0.71), PFHpS (OR=0.45, 95%CI: 0.29, 0.71), and HFPO-DA (OR=0.68, 95%CI: 0.48, 0.97) were negatively correlated with high menstrual blood volume. The mixed exposure model showed that PFAS mixtures were positively correlated with the average number of menstrual cycle days (b=1.60, 95%CI: 0.49, 2.71), irregular menstrual cycles (OR=1.77, 95%CI: 1.19, 2.63), and low menstrual blood volume (OR=1.59, 95%CI: 1.08, 2.35), but negatively correlated with high menstrual blood volume (OR=0.40, 95%CI: 0.22, 0.73). Conclusion Women undergoing ART in Shanghai are widely exposed to PFAS prior to conception. Exposure to PFAS before pregnancy may be related to menstrual characteristics among women seeking ART before undergoing fertility treatments, but additional data from larger populations are required to validate the findings of this study.

Objective: Machine learning (ML) has been reported to have better predictive capability than traditional statistical techniques. The aim of this study was to assess the efficacy of ML algorithms and logistic regression (LR) for predicting depressive symptoms during the COVID-19 pandemic. Methods: Analyses were carried out in a national cross-sectional study involving 21,916 participants. The ML algorithms in this study included random forest (RF), support vector machine (SVM), neural network (NN), and gradient boosting machine (GBM) methods. The performance indices were sensitivity, specificity, accuracy, precision, F1-score, and area under the receiver operating characteristic curve (AUC). Results: LR and NN had the best performance in terms of AUCs. The risk of overfitting was found to be negligible for most ML models except for RF, and GBM obtained the highest sensitivity, specificity, accuracy, precision, and F1-score. Therefore, LR, NN, and GBM models ranked among the best models. Conclusion Compared with ML models, LR model performed comparably to ML models in predicting depressive symptoms and identifying potential risk factors while also exhibiting a lower risk of overfitting.

Objective: Machine learning (ML) has been reported to have better predictive capability than traditional statistical techniques. The aim of this study was to assess the efficacy of ML algorithms and logistic regression (LR) for predicting depressive symptoms during the COVID-19 pandemic. Methods: Analyses were carried out in a national cross-sectional study involving 21,916 participants. The ML algorithms in this study included random forest (RF), support vector machine (SVM), neural network (NN), and gradient boosting machine (GBM) methods. The performance indices were sensitivity, specificity, accuracy, precision, F1-score, and area under the receiver operating characteristic curve (AUC). Results: LR and NN had the best performance in terms of AUCs. The risk of overfitting was found to be negligible for most ML models except for RF, and GBM obtained the highest sensitivity, specificity, accuracy, precision, and F1-score. Therefore, LR, NN, and GBM models ranked among the best models. Conclusion Compared with ML models, LR model performed comparably to ML models in predicting depressive symptoms and identifying potential risk factors while also exhibiting a lower risk of overfitting.