Serum iron level is independently associated with sarcopenia: a retrospective study.

Sarcopenia greatly reduces the quality of life of the elderly, and iron metabolism plays an important role in muscle loss. This study aimed to investigate the association between iron status and sarcopenia. A total of 286 adult patients hospitalized between 2019 and 2021 were included in this study, of which 117 were diagnosed with sarcopenia. Serum iron, total iron binding capacity (TIBC), transferrin, and transferrin saturation levels were compared between groups with and without sarcopenia and were included in the logistic analyses, with significant variables further included in the logistic regression model for the prediction of sarcopenia. Serum iron, TIBC, and transferrin levels decreased significantly in the sarcopenia group (p < 0.05), and were negatively associated with handgrip strength, relative skeletal muscle index, and multiple test performances (p < 0.05). Multivariate logistic analysis showed that sex, age, body mass index (BMI), and serum iron level were independent risk factors for sarcopenia. In the final logistic regression model, male sex (odds ratio [OR] 3.65, 95% confidence interval [CI] 1.67-7.98), age > 65 years (OR 5.40, 95% CI 2.25-12.95), BMI < 24 kg/m2 (OR 0.17, 95% CI 0.08-0.36), and serum iron < 10.95 µmol/L (OR 0.39, 95% CI 0.16-0.93) were included. Our study supported the impact of iron metabolism on muscle strength and performance.

Research on Human Posture Estimation Algorithm Based on YOLO-Pose.

In response to the numerous challenges faced by traditional human pose recognition methods in practical applications, such as dense targets, severe edge occlusion, limited application scenarios, complex backgrounds, and poor recognition accuracy when targets are occluded, this paper proposes a YOLO-Pose algorithm for human pose estimation. The specific improvements are divided into four parts. Firstly, in the Backbone section of the YOLO-Pose model, lightweight GhostNet modules are introduced to reduce the model's parameter count and computational requirements, making it suitable for deployment on unmanned aerial vehicles (UAVs). Secondly, the ACmix attention mechanism is integrated into the Neck section to improve detection speed during object judgment and localization. Furthermore, in the Head section, key points are optimized using coordinate attention mechanisms, significantly enhancing key point localization accuracy. Lastly, the paper improves the loss function and confidence function to enhance the model's robustness. Experimental results demonstrate that the improved model achieves a 95.58% improvement in mAP50 and a 69.54% improvement in mAP50-95 compared to the original model, with a reduction of 14.6 M parameters. The model achieves a detection speed of 19.9 ms per image, optimized by 30% and 39.5% compared to the original model. Comparisons with other algorithms such as Faster R-CNN, SSD, YOLOv4, and YOLOv7 demonstrate varying degrees of performance improvement.

Efficacy of Endoscopic Tissue Adhesive in Patients with Gastrointestinal Tumor Bleeding.

BACKGROUND: Gastrointestinal tumors bleeding remains a significantly clinical challenge due to its resistance to conventional endoscopic hemostasis methods. While the efficacy of endoscopic tissue adhesives (ETA) in variceal bleeding has been established, its role in gastrointestinal tumor bleeding (GITB) remains ambiguous. AIMS: This study aims to assess the feasibility and effectiveness of ETA in the treatment of GITB. METHODS: The study enrolled 30 patients with GITB who underwent hemostasis through Histoacryl® tissue glue injection. Hemostasis success rates, ETA-related adverse events, and re-bleeding rates were evaluated. RESULTS: ETA application achieved successful hemostasis at all tumor bleeding sites, with immediate hemostasis observed in all 30 (100.0%) patients. Among the initially hemostasis cases, 5 patients (17.0%) experienced re-bleeding within 30 days, and the 60 day re-bleeding rate was 20.0% (6/30). Expect for one case of vascular embolism, no adverse events related with ETA application were reported. The 6 month survival was 93%. CONCLUSION: ETA demonstrated excellent immediate hemostasis success rate in GITB cases and showed promising outcomes in prevention re-bleeding.

Development of a distributed group control strategy for pumping well groups connected by multisource DC microgrids.

Due to the alternating loads on pumping units and the integration of new energy sources, multisource DC microgrid pumping unit well groups experience increased fluctuations in voltage and power as well as superimposed peak and valley values. This work presents a distributed control strategy for pumping unit well groups on a multisource DC microgrid based on the weighted moving average algorithm. A centralized control program is implanted in the RTU of the single-well controller of each pumping unit, and communication with each well is realized via SCADA and multicast communication, resulting in a distributed well group system. The real-time power values of the pumping well group are calculated by grouping the power values, and each group is weighted using the total power fluctuation threshold of the well group as the control target. Then, a weighted moving average algorithm is used to predict the next power value and form a table of predicted real-time power spectra. According to the power values in the community power spectrum table, the inverter frequency is proportionally adjusted downwards to reach the power peak before deceleration; after the power peak is crossed, the frequency is increased in the same way to reach the power valley before acceleration. Finally, the peak and valley power values of the bus system level off and further learn to reach the set impulse; ultimately, a stable impulse is formed. In laboratory testing and field application in the Shengli Oilfield XIN-11 block, the group control software module effectively suppressed the active power peak and valley values and voltage fluctuations of the bus system, the active power fluctuation rate range decreased by more than 70%, and the DC bus voltage fluctuation range decreased by more than 80%; moreover, the active power decreased by approximately 6% without additional hardware costs.

Flexible cellulose nanofibers/MXene composite films for UV-shielding packaging.

Cellulose nanofibers (CNF) based films are promising packaging materials, but the lack of special functions (especially UV-shielding property) usually restrict their further applications. In this work, MXene was incorporated into the CNF film by a direct solvent volatilization induced film forming method to study its UV-shielding property for the first time, which avoided the using of a vacuum filtration equipment. The composite films containing glycerin could be folded repeatedly without breaking, showing good flexibility. The structure and properties of MXene/CNF composite films (CMF) were characterized systematically. The results showed that MXene distributed uniformly in the CNF film matrix and there was strong hydrogen bonding interaction between CNF and MXene. The tensile strength and Young's modulus of the composite films could reach 117.5 MPa and 2.23 GPa, which was 54.1 % and 59.2 % higher than those of pure CNF film, respectively. With the increase of MXene content, both the UVA and UVB shielding percentages increased significantly from 17.2 % and 25.5 % to 100.0 %, showing excellent UV-shielding property. Moreover, CMF exhibited a low oxygen permeability (OP) value of 0.39 cc µm d-1 m-2 kPa-1, a low water vapor permeability (WVP) value of 5.13 × 10-11 g-1s-1Pa-1 and a high antibacterial rate against E. coli (94.1 % at 24 h), showing potential application in the packaging field.

Circ-EIF3I Promotes Hepatocellular Carcinoma Progression Through Modulating miR-361-3p/DUSP2 Axis.

Hepatocellular carcinoma (HCC) is one of the most common malignant cancers globally. Circular RNAs (circRNAs) have been implicated in the development of HCC. Previous studies have confirmed that circ-EIF3I plays an important role in the progress of lung cancer. Nevertheless, the biological functions of circ-EIF3I and the underlying mechanisms by which they regulate HCC progression remain unclear. In this study, the regulatory mechanism and targets were studied with bioinformatics analysis, luciferase reporting analysis, transwell migration, Cell Counting Kit-8, and 5-Ethynyl-2'-deoxyuridine analysis. In addition, in vivo tumorigenesis and metastasis assays were employed to evaluate the roles of circ-EIF3I in HCC. The result shows that the circ-EIF3I expression was increased in HCC cell line, which means that circ-EIF3I plays a role in the progression of HCC. Downregulation of circ-EIF3I suppressed HCC cells' proliferation and migration in both in vivo and in vitro experiments. Bioinformatics and luciferase report analysis confirmed that both miR-361-3p and Dual-specificity phosphatase 2 (DUSP2) were the downstream target of circ-EIF3I. The overexpression of DUSP2 or inhibition of miR-361-3p restored HCC cells' proliferation and migration ability after silence circ-EIF3I. Taken together, our study found that downregulation of circ-EIF3I suppressed the progression of HCC through miR-361-3p/DUSP2 Axis.

High serum total bilirubin as a potential protective factor for gestational diabetes mellitus: A retrospective cohort study of 92,885 Chinese pregnant women.

AIMS: Identifying physiological factors that could reduce pregnant women's risk for developing gestational diabetes mellitus (GDM) is crucial for early prevention and intervention. We aimed to examine whether higher serum levels of total bilirubin (TBIL) were associated with a decreased risk of GDM. METHODS: We conducted a retrospective cohort study in a tertiary care hospital in Shanghai, China. A total of 92,885 pregnant women were included. Serum TBIL levels were determined during the first antenatal visit before 24 weeks of gestation and GDM was diagnosed with a 75-g oral glucose tolerance test (OGTT) at 24-28 weeks of gestation. RESULTS: A total of 13,037 GDM cases were identified, a prevalence of 14.0 % (13,037/92,885). These women had a higher median TBIL concentration 7.9 versus 7.6 mmol/l (P < 0.001). For the 91,051 women with TBIL within the physiologically normal range (≤ 17.1 µmol/l), a one interquartile range increase in TBIL (3.4 µmol/l) was associated with a decreased risk of GDM: adjusted odds ratio (OR)=0.89 [95 % CI 0.87;0.92]. For these women, the adjusted ORs for GDM across TBIL quartiles were: 0.92 [0.88;0.97] for the second, 0.85 [0.81;0.90] for the third, and 0.78 [0.74;0.83] for the fourth quartile in comparison with the first quartile. CONCLUSION: Our study demonstrated that elevated serum TBIL levels were associated with decreased risk of GDM and supported its potential role in the prevention and early intervention of GDM.

THUMPD2 catalyzes the N2-methylation of U6 snRNA of the spliceosome catalytic center and regulates pre-mRNA splicing and retinal degeneration.

The mechanisms by which the relatively conserved spliceosome manages the enormously large number of splicing events that occur in humans (∼200 000 versus ∼300 in yeast) are poorly understood. Here, we show deposition of one RNA modification-N2-methylguanosine (m2G) on the G72 of U6 snRNA (the catalytic center of the spliceosome) promotes efficient pre-mRNA splicing activity in human cells. This modification was identified to be conserved among vertebrates. Further, THUMPD2 was demonstrated as the methyltransferase responsible for U6 m2G72 by explicitly recognizing the U6-specific sequences and structural elements. The knock-out of THUMPD2 eliminated U6 m2G72 and impaired the pre-mRNA splicing activity, resulting in thousands of changed alternative splicing events of endogenous pre-mRNAs in human cells. Notably, the aberrantly spliced pre-mRNA population elicited the nonsense-mediated mRNA decay pathway. We further show that THUMPD2 was associated with age-related macular degeneration and retinal function. Our study thus demonstrates how an RNA epigenetic modification of the major spliceosome regulates global pre-mRNA splicing and impacts physiology and disease.

Kinesin superfamily member KIFC2 as an independent prognostic biomarker of colon adenocarcinoma associated with poor immune response.

Clinical outcomes of colon adenocarcinoma (COAD) exhibit heterogeneity among different patients, highlighting the need for novel prognostic biomarkers. Kinesin superfamily members have been shown to play a crucial role in tumors and can predict cancer diagnosis and prognosis. However, the role of kinesin family member C2 (KIFC2) in tumors, particularly its prognostic value in COAD, remains poorly understood. Our bioinformatics analysis of the cancer genome atlas and GEO databases revealed significantly higher expression of KIFC2 in COAD, correlating with a worse prognosis in the cancer genome atlas-COAD and GSE17536 cohorts. Additionally, differentially expressed genes in COAD were enriched in immune-related pathways, and patients with higher KIFC2 expression showed fewer activated CD4 + T cells. These findings suggest KIFC2 as a potential prognostic biomarker for COAD, warranting further validation in clinical studies.

Impacts of acute and chronic suspended solids exposure on juvenile freshwater mussels.

Construction activities may affect adjacent water systems by introducing increased levels of suspended solids into the water body and may subsequently affect the survival and growth of freshwater mussels. We tested three sediment types from sites in Missouri, including Spring River sediment (SRS), Osage River bank clay soil (ORC), and quarried limestone from Columbia (LMT). We prepared series of suspensions of each sediment with total suspended solids concentrations ranging from 0 to 5000 mg/L. Juveniles from three mussel species, Fatmucket (Lampsilis siliquoidea), Arkansas Brokenray (Lampsilis reeveiana), and Washboard (Megalonaias nervosa) were exposed to these suspensions in both acute (96-h) and chronic (28-d) tests. No clear impact on survival was observed from the acute or chronic exposures, but chronic test showed that juvenile mussels' growth was strongly affected. Interestingly, growth was enhanced at lower levels of SRS and ORC (≤500 mg/L, p < 0.05), and the juvenile mussels exposed to 500 mg/L SRS exhibited approximately 60 % more dry weight than those reared in the control. LMT did not enhance growth. Growth was slowed by high concentrations (>1000 mg/L) of all three sediments, implying that high suspended solids levels could reduce survival in the long term. Our findings may help to inform regulations and guidelines for construction activities to minimize adverse effects on juvenile mussels.

The association of lipid metabolism and sarcopenia among older patients: a cross-sectional study.

Sarcopenia has become a heavy disease burden among the elderly. Lipid metabolism was reported to be involved in many degenerative diseases. This study aims to investigate the association between dysregulated lipid metabolism and sarcopenia in geriatric inpatients. This cross-sectional study included 303 patients aged ≥ 60, of which 151 were diagnosed with sarcopenia. The level of total cholesterol (TC), triglyceride (TG), high-density lipoprotein (HDL), low-density lipoprotein (LDL), homocysteine (HCY), BMI, and fat percentage, were compared between sarcopenia and non-sarcopenia patients. The Spearman correlation coefficient was used to estimate the association between sarcopenia and the level of lipid metabolism. To determine risk factors related to sarcopenia, a multivariate logistic regression analysis was carried out. Risk prediction models were constructed based on all possible data through principal component analysis (PCA), Logistic Regression (LR), Support Vector Machine (SVM), k-Nearest Neighbor (KNN), and eXtreme Gradient Boosting (XGboost). We observed rising prevalence of sarcopenia with increasing age, decreasing BMI, and fat percentage (p < 0.001, Cochran Armitage test). Multivariate logistic regression analysis revealed sarcopenia's risk factors, including older age, male sex, lower levels of BMI, TC, and TG, and higher levels of LDL and HCY (p < 0.05). The sarcopenia risk prediction model showed the risk prediction value of sarcopenia, with the highest area under the receiver operating curve (AUC) of 0.775. Our study provided thorough insight into the risk factors associated with sarcopenia. It demonstrated that an increase in lipid metabolism-related parameters (BMI, TG, TC), within normal reference ranges, may be protective against sarcopenia. The present study can illuminate the direction and significance of lipid metabolism-related factors in preventing sarcopenia.

Comprehensive Detection of PD-L1 Protein and mRNA in Tumor Cells and Extracellular Vesicles through a Real-Time qPCR Assay.

A growing number of studies have shown that tumor cells secrete extracellular vesicles (EVs) containing programmed death-ligand 1 (PD-L1) protein. These vesicles can travel to lymph nodes and remotely inactivate T cells, thereby evading immune system attack. Therefore, the simultaneous detection of PD-L1 protein expression in cells and EVs is of great significance in guiding immunotherapy. Herein, we developed a method based on qPCR for the simultaneous detection of PD-L1 protein and mRNA in EVs and their parental cells (PREC-qPCR assay). Lipid probes immobilized on magnetic beads were used to capture EVs directly from samples. For RNA assay, EVs were directly broken by heating and quantified with qPCR. As to protein assay, EVs were recognized and bound with specific probes (such as aptamers), which were used as templates in subsequent qPCR analysis. This method was used to analyze EVs of patient-derived tumor clusters (PTCs) and plasma samples from patients and healthy volunteers. The results revealed that the expression of exosomal PD-L1 in PTCs was correlated with tumor types and significantly higher in plasma-derived EVs from tumor patients than that of healthy individuals. When extended to cells and PD-L1 mRNAs, the results showed that the expression of PD-L1 protein was consistent with mRNA in cancer cell lines, while PTCs demonstrated significant heterogeneity. This comprehensive detection of PD-L1 at four levels (cell, EVs, protein, and mRNA) is believed to enhance our understanding of the relationship among PD-L1, tumors, and the immune system and to provide a promising tool for predicting the benefits of immunotherapy.

A single-center retrospective study of the COCO technique in the treatment of chronic internal carotid artery occlusion.

OBJECTIVE: To introduce a novel endovascular recanalization method and to investigate its success rate, periprocedural complications, and early outcomes in patients with chronic internal carotid artery occlusion (CICAO). As this novel technique was designed to treat CICAO with a full coaxial system, we named it the COCO technique. METHODS: Data from consecutive patients with symptomatic CICAO who underwent endovascular recanalization in our institution were retrospectively reviewed. The COCO technique allows extracranial angioplasty and stenting with occasional intracranial angioplasty and stenting as needed to be performed in a coaxial fashion. Patients' demographic and clinical information, morphologic characteristics, procedural results, complications, and follow-up outcomes were recorded. RESULTS: Forty-nine patients were enrolled in this study. The technical success rate was 89.8% (44/49). Four patients experienced intraoperative complications, two patients had a slight subarachnoid hemorrhage, and two patients had asymptomatic dissection. Distal embolization or carotid-cavernous arteriovenous fistula was not detected. In addition, three patients developed hemorrhagic complications and three developed postoperative ischemic complications. All these patients improved after conservative treatment and subsequent rehabilitation. During the median 6 (3-6) months of follow-up, one patient died of severe pneumonia and two patients experienced recurrent ischemic events. In patients with successful recanalization, modified Rankin Scale scores were lower at the 3-month follow-up than at baseline (1 (0-2) vs 2 (1-2), P=0.04). Restenosis was observed in six (15.8%) patients. CONCLUSIONS: Our study showed that the COCO technique is effective and safe for endovascular recanalization in patients with CICAO and has low periprocedural complications and favorable functional outcomes.

Temporal-spatial dynamics of electronic plasma in a femtosecond laser-induced sapphire microstructure.

In this study, the time-spatial evolution of single-pulse femtosecond laser-induced plasma in sapphire is studied by using femtosecond time-resolved pump-probe shadowgraphy. Laser-induced sapphire damage occurred when the pump light energy was increased to 20 µJ. Based on its shadowgraphy image, the threshold electron density can be estimated to be about 2.48×1020 c m -3. The evolution law of the transient peak electron density and its spatial position as femtosecond laser propagation in sapphire were researched. The transitions from single-focus to multi-focus as the laser focus shifted from the surface to a deeper part were observed from the transient shadowgraphy images. The focal point distance in multi-focus increased as the focal depth increased. The distributions of femtosecond laser-induced free electron plasma and the final microstructure were consistent with each other.

SynDLP is a dynamin-like protein of Synechocystis sp. PCC 6803 with eukaryotic features.

Dynamin-like proteins are membrane remodeling GTPases with well-understood functions in eukaryotic cells. However, bacterial dynamin-like proteins are still poorly investigated. SynDLP, the dynamin-like protein of the cyanobacterium Synechocystis sp. PCC 6803, forms ordered oligomers in solution. The 3.7 Å resolution cryo-EM structure of SynDLP oligomers reveals the presence of oligomeric stalk interfaces typical for eukaryotic dynamin-like proteins. The bundle signaling element domain shows distinct features, such as an intramolecular disulfide bridge that affects the GTPase activity, or an expanded intermolecular interface with the GTPase domain. In addition to typical GD-GD contacts, such atypical GTPase domain interfaces might be a GTPase activity regulating tool in oligomerized SynDLP. Furthermore, we show that SynDLP interacts with and intercalates into membranes containing negatively charged thylakoid membrane lipids independent of nucleotides. The structural characteristics of SynDLP oligomers suggest it to be the closest known bacterial ancestor of eukaryotic dynamin.

RNA-seq reveals Nup62 as a potential regulator for cell division after traumatic brain injury in mice hippocampus.

Background: Hippocampus impairment is a common condition encountered in the clinical diagnosis and treatment of traumatic brain injury (TBI). Several studies have investigated this phenomenon. However, its molecular mechanism remains unclear. Methods: In this study, Illumina RNA-seq technology was used to determine the gene expression profile in mice hippocampus after TBI. We then conducted bioinformatics analysis to identify the altered gene expression signatures and mechanisms related to TBI-induced pathology in the hippocampus. Real-time quantitative polymerase chain reaction and western blot were adopted to verify the sequencing results. Results: The controlled cortical impact was adopted as the TBI model. Hippocampal specimens were removed for sequencing. Bioinformatics analysis identified 27 upregulated and 17 downregulated differentially expressed genes (DEGs) in post-TBI mouse models. Potential biological functions of the genes were determined via Gene Set Enrichment Analysis (GSEA)-based Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses, which suggested a series of functional changes in the nervous system. Specifically, the nucleoporin 62 (Nup62) DEG was discussed and verified. Gene ontology biological process enriched analysis suggests that the cell division was upregulated significantly. The present study may be helpful for the treatment of impaired hippocampus after TBI in the future.

Targeted therapy for intractable cancer on the basis of molecular profiles: An open-label, phase II basket trial (Long March Pathway).

Purpose: We evaluated he effects of molecular guided-targeted therapy for intractable cancer. Also, the epidemiology of druggable gene alterations in Chinese population was investigated. Materials and methods: The Long March Pathway (ClinicalTrials.gov identifier: NCT03239015) is a non-randomized, open-label, phase II trial consisting of several basket studies examining the molecular profiles of intractable cancers in the Chinese population. The trial aimed to 1) evaluate the efficacy of targeted therapy for intractable cancer and 2) identify the molecular epidemiology of the tier II gene alterations among Chinese pan-cancer patients. Results: In the first stage, molecular profiles of 520 intractable pan-cancer patients were identified, and 115 patients were identified to have tier II gene alterations. Then, 27 of these 115 patients received targeted therapy based on molecular profiles. The overall response rate (ORR) was 29.6% (8/27), and the disease control rate (DCR) was 44.4% (12/27). The median duration of response (DOR) was 4.80 months (95% CI, 3.33-27.2), and median progression-free survival (PFS) was 4.67 months (95% CI, 2.33-9.50). In the second stage, molecular epidemiology of 17,841 Chinese pan-cancer patients demonstrated that the frequency of tier II gene alterations across cancer types is 17.7%. Bladder cancer had the most tier-II alterations (26.1%), followed by breast cancer (22.4%), and non-small cell lung cancer (NSCLC; 20.2%). Conclusion: The Long March Pathway trial demonstrated a significant clinical benefit for intractable cancer from molecular-guided targeted therapy in the Chinese population. The frequency of tier II gene alterations across cancer types supports the feasibility of molecular-guided targeted therapy under basket trials.

Efficacy and safety of berberine for several cardiovascular diseases: A systematic review and meta-analysis of randomized controlled trials.

BACKGROUND: Berberine has been widely used for the adjuvant therapy of several cardiovascular diseases (CVDs). However, evidence for its efficacy remains controversial. PURPOSE: This study aimed to evaluate the efficacy and safety of berberine in CVDs. STUDY DESIGN: A systematic review and meta-analysis of randomized controlled trials (RCTs). METHODS: We searched ten electronic databases for articles from inception to December 23, 2022. RCTs comparing berberine alone or combined with statins versus statins or routine for CVDs were included. Meta-analysis was performed according to the Cochrane Handbook. RESULTS: Forty-four RCTs were included with 4606 patients. There were no differences between berberine alone and routine or statins in improving total cholesterol (TC) (SMD, 0.43; 95% CI, -0.39 to 1.24; p = 0.30; I2 = 95%), triglyceride (TG) (SMD, -0.14; 95% CI, -0.49 to 0.21; p = 0.44; I2 = 76%), low-density lipoprotein cholesterol (LDL-C) (SMD, 0.69; 95% CI, -0.23 to 1.60; p = 0.14; I2 = 96%), high-density lipoprotein cholesterol (HDL-C) (SMD, 0.55; 95% CI, -0.48 to 1.57; p = 0.30; I2 = 96%), and Crouse score levels. Berberine alone significantly reduced National Institute of Health Stroke Scale (NIHSS) score, high-sensitivity C-reactive protein (hs-CRP), interleukin-6 (IL-6), tumor necrosis factor-α (TNF-α), and intima-media thickness (IMT) levels than routine therapy. Berberine plus statins significantly reduced TC, TG, LDL-C, NIHSS score, hs-CRP, TNF-α, IMT, Crouse score, and number of unstable plaques levels than routine or statins. However, no differences were found between groups in improving HDL-C and IL-6 levels. There were no significant differences between groups in the incidence of adverse reactions. CONCLUSION: This study suggests that berberine may be a promising alternative for CVDs with no serious adverse reactions. However, our results may be limited by the quality of existing research. High-quality RCTs are needed to provide more convinced evidence.

Using machine learning for mortality prediction and risk stratification in atezolizumab-treated cancer patients: Integrative analysis of eight clinical trials.

BACKGROUND: Few models exist to predict mortality in cancer patients receiving immunotherapy. Our aim was to build a machine learning-based risk stratification model for predicting mortality in atezolizumab-treated cancer patients. METHODS: Data from 2538 patients in eight atezolizumab-treated cancer clinical trials across three cancer types (non-small-cell lung cancer, bladder transitional cell carcinoma, and renal cell carcinoma) were included. The whole cohort was randomly split into development and validation cohorts in a 7:3 ratio. Machine-learning algorithms (extreme gradient boosting, random forest, logistic regression with lasso regularization, support vector machine, and K-nearest neighbor) were applied to develop prediction models. Model performance was mainly assessed by area under the receiver operating characteristic curve (AUC) value, calibration plot, and decision curve analysis. The probability of death risk was then stratified. RESULTS: One thousand and three hundred and seventy-nine (54.33%) patients died. The random forest (RF) model was overall the best in terms of predictive performance, with the AUC of 0.844 (95% confidence interval [CI]: 0.826-0.862) in the development cohort and 0.786 (95% CI: 0.754-0.818) in the validation cohort for predicting mortality. Twelve baseline variables contributing to mortality prediction in the RF model were C-reactive protein, PD-L1 level, cancer type, prior liver metastasis, derived neutrophil-to-lymphocyte ratio, alkaline phosphatase, albumin, hemoglobin, white blood cell count, number of metastatic sites, pulse rate, and Eastern Cooperative Oncology Group (ECOG) performance status. A total of 1782 (70.2%) patients were separated into the high-risk and 756 (29.8%) low-risk groups. Patients in the high-risk group were significantly more likely to die, experience disease progression, discontinue study, and discontinue treatment than patients in the low-risk group (all p values < 0.001). Risk groups were not associated with immune-related adverse events and grades 3-5 treatment-related adverse events (all p values > 0.05). CONCLUSION: RF model has good performance in mortality prediction and risk stratification for cancer patients receiving atezolizumab monotherapy.

Current Trends and Challenges in Drug-Likeness Prediction: Are They Generalizable and Interpretable?

Importance: Drug-likeness of a compound is an overall assessment of its potential to succeed in clinical trials, and is essential for economizing research expenditures by filtering compounds with unfavorable properties and poor development potential. To this end, a robust drug-likeness prediction method is indispensable. Various approaches, including discriminative rules, statistical models, and machine learning models, have been developed to predict drug-likeness based on physiochemical properties and structural features. Notably, recent advancements in novel deep learning techniques have significantly advanced drug-likeness prediction, especially in classification performance. Highlights: In this review, we addressed the evolving landscape of drug-likeness prediction, with emphasis on methods employing novel deep learning techniques, and highlighted the current challenges in drug-likeness prediction, specifically regarding the aspects of generalization and interpretability. Moreover, we explored potential remedies and outlined promising avenues for future research. Conclusion: Despite the hurdles of generalization and interpretability, novel deep learning techniques have great potential in drug-likeness prediction and are worthy of further research efforts.