External Validation Research Articles

Public evidence on AI products for digital pathology

Novel products applying artificial intelligence (AI)-based methods to digital pathology images are touted to have many uses and benefits. However, publicly available information for products can be variable, with few sources of independent evidence. This review aimed to identify public evidence for AI-based products for digital pathology. Key features of products on the European Economic Area/Great Britain (EEA/GB) markets were examined, including their regulatory approval, intended use, and published validation studies. There were 26 AI-based products that met the inclusion criteria and, of these, 24 had received regulatory approval via the self-certification route as General in vitro diagnostic (IVD) medical devices. Only 10 of the products (38%) had peer-reviewed internal validation studies and 11 products (42%) had peer-reviewed external validation studies. To support transparency an online register was developed using identified public evidence (https://osf.io/gb84r/), which we anticipate will provide an accessible resource on novel devices and support decision making.

Identification of lncRNA dual targeting PD-L1 and PD-L2 as a novel prognostic predictor for gastric cancer

BackgroundAlthough breakthroughs have been achieved in gastric cancer (GC) therapy with immune checkpoint inhibitors (ICIs) targeting programmed death-1 (PD-1) and programmed death-ligand 1 (PD-L1), the acquisition of high response rate remains a huge challenge for clinicians. It is imperative to identify novel biomarkers for predicting response to immunotherapy and explore alternative therapeutic strategy for GC.MethodsThe transcriptomic profiles and clinical information of GC patients from The Cancer Genome Atlas (TCGA)-stomach adenocarcinoma (STAD) database was used to screen differentially expressed lncRNAs between the tumor specimens and the paracancerous tissues. The TargetScan, miRDB and miRcode database were then utilized to construct competing endogenous RNA (ceRNA) networks and identify pivotal lncRNAs. An independent dataset from GEO (GSE70880) and 23 pairs of GC specimens of our cohort were subsequently performed for external validity. The relationship between clinical variables and gene expression were evaluated by Kruskal–wallis test and Wilcoxon signed-rank. The prognostic value of the candidate genes was assessed using Kaplan-Meier analysis and Cox regression models. CIBERSORT and Gene set enrichment analysis (GSEA) were used to determine immune cell infiltration. Gastric adenocarcinoma AGS cells and human embryonic kidney 293T (HEK293T) cells with knockdown of LINC01094 were generated by siRNA transfection, followed by detecting the alteration of the target miRNA and PD-L1/PD-L2 by RT-qPCR. Besides, the interaction between lncRNA and the miRNA–PD-L1/PD-L2 axis were verified by dual luciferase reporter assay.ResultsTwenty-two intersecting lncRNAs were identified to be PD-L1/PD-L2-related lncRNAs and LINC01094–miR-17-5p–PD-L1/PD-L2 was constructed as a potential ceRNA network. LINC01094 was increased in tumor specimens than adjacent normal samples and was positively associated with advanced tumor stages and EBV and MSI status. Furthermore, LINC01094 expression was an independent risk factor for poor overall survival (OS) in GC patients. CD8+ T cell exhaustion-related genes were enriched in high-LINC01094 tissues and high-PD-L2 group. A strong positive association of LINC01094 expression was established with M2 macrophages, IL-10+ TAM, as well as PD-L1 and PD-L2 levels, therefore a LINC01094–miR-17-5p–IL-10 network was proposed in macrophages. Using the exoRBase database, LINC01094 was assumed in blood exosomes of GC patients The results of knockdown experiments and luciferase reporter assays revealed that LINC01094 interacted with miR-17-5p and served as a miRNA sponge to regulate the expression of PD-L1 and PD-L2.ConclusionLINC01094 dually regulates the expression of PD-L1 and PD-L2 and shapes the immunosuppressive tumor microenvironment via sponging miR-17-5p. LINC01094 may serve as a potential prognostic predictor and therapeutic target in GC.

Prognostic nomograms for young breast cancer: A retrospective study based on the SEER and METABRIC databases.

Young breast cancer (YBC) is a subset of breast cancer that is often more aggressive, but less is known about its prognosis. In this study, we aimed to generate nomograms to predict the overall survival (OS) and breast cancer-specific survival (BCSS) of YBC patients. Data of women diagnosed with YBC between 2010 and 2020 were obtained from the Surveillance, Epidemiology, and End Results (SEER) database. The patients were randomly allocated into a training cohort (n = 15,227) and internal validation cohort (n = 6,526) at a 7:3 ratio. With the Cox regression models, significant prognostic factors were identified and used to construct 3-, 5-, and 10-year nomograms of OS and BCSS. Data from the Molecular Taxonomy of Breast Cancer International Consortium (METABRIC) database were used as an external validation cohort (n = 90). We constructed nomograms incorporating 10 prognostic factors for OS and BCSS. These nomograms demonstrated strong predictive accuracy for OS and BCSS in the training cohort, with C-indexes of 0.806 and 0.813, respectively. The calibration curves verified that the nomograms have good prediction accuracy. Decision curve analysis demonstrated their practical clinical value for predicting YBC patient survival rates. Additionally, we provided dynamic nomograms to improve the operability of the results. The risk stratification ability assessment also showed that the OS and BCSS rates of the low-risk group were significantly better than those of the high-risk group. Here, we generated and validated more comprehensive and accurate OS and BCSS nomograms than models previously developed for YBC. These nomograms can help clinicians evaluate patient prognosis and make clinical decisions.

ACCI-HBV-ACLF: A Novel Predictive Model for Hepatitis B Virus-Related Acute-On-Chronic Liver Failure.

Early identification of hepatitis B virus-related acute-on-chronic liver failure (HBV-ACLF) holds crucial importance in guiding clinical management and reducing mortality. However, existing scoring systems often overlook patient's underlying clinical condition, which significantly impacts prognosis. Use the age-adjusted Charlson comorbidity index (aCCI) to evaluate the patient's complications to develop a more precise model for predicting transplant-free mortality in HBV-ACLF patients. Nine hundred and six patients were included for investigation and were segregated into a training cohort and a temporal validation cohort according to the chronological order of admission in a ratio of 7:3. In the training cohort, univariate analysis, logistic regression analysis and LASSO regression analysis were used to construct a prognostic model and it was subsequently validated in a temporal validation cohort and an external validation cohort. We found total bilirubin, neutrophils, international normalised ratio and aCCI exhibited significant associations with 28-day transplant-free mortality and established a novel prognostic model, named aCCI-HBV-ACLF. The model demonstrated strong predictive performance, with area under the receiver operating characteristic curve (ROC) values of 0.859 for 28-day mortality, 0.822 for 90-day mortality. In the temporal validation cohort, aCCI-HBV-ACLF achieved area under the ROC values of 0.869 for 28-day mortality and 0.850 for 90-day mortality. In the external validation cohort, aCCI-HBV-ACLF had area under the ROC values of 0.868 for 28-day mortality and 0.888 for 90-day mortality. This study proposes a new prognostic model, which achieved excellent predictive ability for 28-/90-day transplant-free mortality rates among patients with HBV-ACLF.

Predicting Biochemical Recurrence of Prostate Cancer Post-Prostatectomy Using Artificial Intelligence: A Systematic Review

Background/Objectives: Biochemical recurrence (BCR) after radical prostatectomy (RP) is a significant predictor of distal metastases and mortality in prostate cancer (PCa) patients. This systematic review aims to evaluate the accuracy of artificial intelligence (AI) in predicting BCR post-RP. Methods: Adhering to PRISMA guidelines, a comprehensive literature search was conducted across Medline, Embase, Web of Science, and IEEE Xplore. Studies were included if they utilised AI to predict BCR in patients post-RP. Studies involving patients who underwent radiotherapy or salvage RP were excluded. This systematic review was registered on PROSPERO (International prospective register of systematic reviews) under the ID CRD42023482392. Results: After screening 9764 articles, 24 met the inclusion criteria. The included studies involved 27,216 patients, of whom 7267 developed BCR. AI algorithms developed using radiological parameters demonstrated higher predictive accuracy (median AUROC of 0.90) compared to algorithms based solely on pathological variables (median AUROC of 0.74) or clinicopathological variables (median AUROC of 0.81). According to the Prediction Model Risk of Bias Assessment Tool (PROBAST), the overall risk of bias was unclear in three studies due to ambiguous inclusion criteria and the exclusion of many patients because of missing follow-up data. In seven studies, the developed AI outperformed or was at least equivocal to traditional methods of BCR prediction. Conclusions: AI shows promise in predicting BCR post-RP, particularly when radiological data were used in its development. However, the significant variability in AI performance and study methodologies highlights the need for larger, standardised prospective studies with external validation prior to clinical application.

A Multi-Isotopic Chemometric Approach for Tracing Hazelnut Origins

High-value products, such as hazelnuts, are particularly vulnerable to fraud due to their price dependence on geographical origin. Guaranteeing hazelnuts’ authenticity is essential for consumer trust and safety. Stable isotope analysis has become a reference method for origin authentication as it is reliable, robust, and easily transferable across laboratories. However, multiple isotopic markers coupled with chemometric techniques are often needed to authenticate food provenance accurately. In this study, we focused on assessing the potential of bulk δ18O, along with δ2H and δ13C of the main fatty acids, as hazelnut-origin authenticity markers. PLS-DA classification models were developed to differentiate samples (n = 207) according to their region of origin. This multi-isotopic approach provided promising external validation results, achieving a 94% global correct classification rate in discriminating hazelnuts from regions with distinct geographical and environmental conditions. This study lays the groundwork for further model development and evaluation across additional production areas and harvest years.

Single-cell RNA sequencing highlights the role of proinflammatory fibroblasts, vascular endothelial cells, and immune cells in the keloid immune microenvironment.

Keloids, characterized by an aberrant wound-healing process and a highly complex immune microenvironment, pose significant challenges for clinical management. Fibroblasts and vascular endothelial cells (VEC) were identified as the leading cells of keloid development. However, their roles in the keloid immune landscape have yet to be thoroughly elucidated. To explore the functional state of cells in the immune landscape of keloids, we conducted a single-cell RNA sequencing analysis on the tissue from three keloid lesions and two specimens of healthy skin. We simultaneously utilized available keloid data from the public database for external validation. Specific subsets, such as proinflammatory fibroblasts (piF) and VEC, were markedly elevated in lesional skincompared to normal skin. Subsequent differential gene expression and Gene Ontology analyses indicated that these subsets may be involved in shaping the microenvironment. In keloids, there is an increased expression of immune-associated genes (P < 0.05), including TNFRSF6B, HGF, and TGFB3, alongside a decreased expression of inflammatory chemokines in the piF. Moreover, the significant upregulation of immune suppressive genes (P < 0.05), including CD39, CD73, and HIF1A, suggested the potential involvement of VEC as a conditional immune subpopulation in the keloid microenvironment. Immune cell communication analysis revealed preferential enrichment of macrophages and Tregs, highlighting intensified macrophage-centered interactions within the keloid microenvironment. Our study highlighted the role of piF and VEC in the immune microenvironment of keloids for the first time, providing potential targets for therapeutic development.

First-Trimester Prediction Models Based on Maternal Characteristics for Adverse Pregnancy Outcomes: A Systematic Review and Meta-Analysis.

Early risk stratification can facilitate timely interventions for adverse pregnancy outcomes, including preeclampsia (PE), small-for-gestational-age neonates (SGA), spontaneous preterm birth (sPTB) and gestational diabetes mellitus (GDM). To perform a systematic review and meta-analysis of first-trimester prediction models for adverse pregnancy outcomes. The PubMed database was searched until 6 June 2024. First-trimester prediction models based on maternal characteristics were included. Articles reporting on prediction models that comprised biochemical or ultrasound markers were excluded. Two authors identified articles, extracted data and assessed risk of bias and applicability using PROBAST. A total of 77 articles were included, comprising 30 developed models for PE, 15 for SGA, 11 for sPTB and 35 for GDM. Discriminatory performance in terms of median area under the curve (AUC) of these models was 0.75 [IQR 0.69-0.78] for PE models, 0.62 [0.60-0.71] for SGA models of nulliparous women, 0.74 [0.72-0.74] for SGA models of multiparous women, 0.65 [0.61-0.67] for sPTB models of nulliparous women, 0.71 [0.68-0.74] for sPTB models of multiparous women and 0.71 [0.67-0.76] for GDM models. Internal validation was performed in 40/91 (43.9%) of the models. Model calibration was reported in 21/91 (23.1%) models. External validation was performed a total of 96 times in 45/91 (49.5%) of the models. High risk of bias was observed in 94.5% of the developed models and in 58.3% of the external validations. Multiple first-trimester prediction models are available, but almost all suffer from high risk of bias, and internal and external validations were often not performed. Hence, methodological quality improvement and assessment of the clinical utility are needed.

Transdiagnostic modeling of clinician-rated symptoms in affective and nonaffective psychotic disorders.

Prevailing factor models of psychosis are centered on schizophrenia-related disorders defined by the Diagnostic and Statistical Manual of Mental Disorders and International Classification of Diseases, restricting generalizability to other clinical presentations featuring psychosis, even though affective psychoses are more common. This study aims to bridge this gap by conducting exploratory and confirmatory factor analyses, utilizing clinical ratings collected from patients with either affective or nonaffective psychoses (n = 1,042). Drawing from established clinical instruments, such as the Positive and Negative Syndrome Scale, Young Mania Rating Scale, and Montgomery-Åsberg Depression Rating Scale, a broad spectrum of core psychotic symptoms was considered for the model development. Among the candidate models considered, including correlated factors and multifactor models, a model with seven correlated factors encompassing positive symptoms, negative symptoms, depression, mania, disorganization, hostility, and anxiety was most interpretable with acceptable fit. The seven factors exhibited expected associations with external validators, were replicable through cross-validation, and were generalizable across affective and nonaffective psychoses. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

Performance of instrumental activities of daily living in patients with haemophilic arthropathy. A cross-sectional cohort study.

The development of haemophilic arthropathy causes joint damage that leads to functional impairment that limits the performance of activities in patients with haemophilia. The aim was to identify the best predictive model for performing instrumental activities of daily living in adult patients with haemophilia arthropathy. Cross-sectional cohort study. 102 patients were recruited. The dependent variable was the performance of instrumental activities of daily living (Lawton and Brody scale). The dependence on the performance of activities of daily living was the dependent endpoint (Barthel scale). The secondary variables were joint damage (Hemophilia Joint Health Score), pain intensity, and clinical, anthropometric, and sociodemographic variables. The degree of dependence, joint damage, pain intensity, and marital status (Cp=5.60) were the variables that best explain the variability in the performance of instrumental activities of daily living (R2 adj=0.51). Loss of predictive capacity is acceptable with good mean internal (R2 mean=0.40) and external (R2-r2=0.09) validation. According to the predictive pattern obtained, patients with haemophilia, who were married, without joint pain or damage, and independent in their day-to-day lives, had a score of 7.91 points (95% CI: 7.42; 8.39) in the performance of instrumental activities of daily living. The predictive model for the functional capacity of instrumental activities of daily living in haemophilia patients encompasses factors such as level of autonomy, joint impairment, pain severity, and marital status. Notably, despite the presence of joint damage, individuals with haemophilia exhibit a significant level of independence in carrying out both basic daily tasks and instrumental activities of daily living. Id NCT04715100.

A nomogram for predicting postpartum post-traumatic stress disorder: a prospective cohort study

BackgroundPostpartumpost-traumatic stress disorder (PTSD), as a psychological stress disorder, has long-term and widespread harm. Still, compared with other postpartum psychiatric disorders, postpartum PTSD has received relatively littleattention in China. This study aims to investigate the risk factors of postpartum PTSD and to develop a convenient and rapid nomogram screening tool to help clinical staff identify high-risk pregnant womenin time and take preventative and management measures.MethodsRecruited pregnant women hospitalized for delivery in Qingdao Municipal Hospital and Jinzhou Maternal and Child Health Hospital from November 2022 to October 2023 as convenient samples for the questionnaire survey. Telephone follow-up was conducted 42 days after delivery. After univariate analysis, multicollinearity analysis, and logistic regression analysis, the risk factors of postnatal PTSD were obtained, a prediction model was established, and a nomogram was drawn by R software. G*power3.1.9.7 calculated the effectiveness of the test. The model was validated internally using the Bootstrap approach, and external validation was carried out using a verification group. The accuracy of the model's predictions and its clinical application value were evaluated by the area under the curve, calibration plot, and decision curve analysis.ResultsA total of 602 women were recruited in this study, and the incidence of postpartum PTSD was 11.1% (67/602). Multifactorial logistic regression analysis showed that poor self-assessment of sleep status in late pregnancy (OR = 5.336), cesarean section (OR = 2.825), instrumental delivery (OR = 5.994), having fear of labor (OR = 4.857), and a high score of Five Factors Inventory Neuroticism subscale (OR = 1.244) were independent risk factors for developing postpartum PTSD. A high Quality of Relationship Index score (OR = 0.891) was a protective factor for postpartum PTSD. In the training and validation sets, the nomogram model's area under the ROC curve was 0.928 and 0.907, respectively. The calibration curves showed that the nomogram model was well-fitted, and the Decision Curve Analysis indicated that the nomogram model had good value for clinical application.ConclusionsWith its strong predictive capacity, the prediction model built using postpartum PTSD risk factors can help clinical caregivers identify high-risk pregnant women early on and implement timely preventive intervention strategies.

Artificial intelligence-enabled reconstruction of the right ventricular pressure curve using the peak pressure value: a proof-of-concept study

Abstract Aims Conventional echocardiographic parameters of right ventricular (RV) function are afterload-dependent. Therefore, incorporating RV pressures may enable the formulation of new parameters that reflect intrinsic RV function accurately. Accordingly, we sought to develop an artificial intelligence–based method to reconstruct the RV pressure curve based on the peak RV pressure. Methods and Results We invasively acquired RV pressure in 29 heart failure patients before and after implanting a left ventricular (LV) assist device. Using these tracings, we trained various machine learning models to reconstruct the RV pressure curve of the entire cardiac cycle based on the peak value of the curve. The best-performing model was compared with two other methods that estimated RV pressures based on a reference LV and RV pressure curve, respectively. Seventeen consecutive patients from another centre who underwent right heart catheterization and simultaneous echocardiography served as an external validation cohort. Among the evaluated algorithms, multilayer perceptron (MLP) achieved the best performance with an R2 of 0.887 (0.834–0.941). The RV and LV reference curve–based methods achieved R2 values of 0.879 (0.815–0.943) and 0.636 (0.500–0.771), respectively. During external validation, MLP exhibited similarly good performance [R2 0.911 (0.873–0.948)], which decreased only modestly if the echocardiography-derived peak RV pressure was used instead of the invasively measured peak RV pressure [R2 0.802 (0.694–0.909)]. Conclusions The proposed method enables the reconstruction of the RV pressure curve using only the peak value as input. Thus, it may serve as a fundamental component for developing new echocardiographic tools targeting the afterload-adjusted assessment of RV function.

Automated quantification of cerebral microbleeds in susceptibility-weighted MRI: association with vascular risk factors, white matter hyperintensity burden, and cognitive function.

To train and validate a deep learning (DL)-based segmentation model for cerebral microbleeds (CMB) on susceptibility-weighted MRI; and to find associations between CMB, cognitive impairment, and vascular risk factors. Participants in this single-institution retrospective study underwent brain MRI to evaluate cognitive impairment between January-September 2023. For training the DL model, the nnU-Net framework was used without modifications. The DL model's performance was evaluated on independent internal and external validation datasets. Linear regression analysis was used to find associations between log-transformed CMB numbers, cognitive function (mini-mental status examination [MMSE]), white matter hyperintensity (WMH) burden, and clinical vascular risk factors (age, sex, hypertension, diabetes, lipid profiles, and body mass index). Training of the DL model (n = 287) resulted in a robust segmentation performance with an average dice score of 0.73 (95% CI, 0.67-0.79) in an internal validation set, (n = 67) and modest performance in an external validation set (dice score = 0.46, 95% CI, 0.33-0.59, n = 68). In a temporally independent clinical dataset (n = 448), older age, hypertension, and WMH burden were significantly associated with CMB numbers in all distributions (total, lobar, deep, and cerebellar; all P <.01). MMSE was significantly associated with hyperlipidemia (β = 1.88, 95% CI, 0.96-2.81, P <.001), WMH burden (β = -0.17 per 1% WMH burden, 95% CI, -0.27-0.08, P <.001), and total CMB number (β = -0.01 per 1 CMB, 95% CI, -0.02-0.001, P = .04) after adjusting for age and sex. The DL model showed a robust segmentation performance for CMB. In all distributions, CMB had significant positive associations with WMH burden. Increased WMH burden and CMB numbers were associated with decreased cognitive function. CMB = cerebral microbleed; DL = deep learning, DSC = dice similarity coefficient; MMSE = mini-mental status examination; SVD = small vessel disease; SWI = susceptibility-weighted image; WMH = white matter hyperintensity.

Comprehensive single-cell and bulk transcriptomic analyses to develop an NK cell-derived gene signature for prognostic assessment and precision medicine in breast cancer

BackgroundNatural killer (NK) cells play crucial roles in mediating anti-cancer activity in breast cancer (BRCA). However, the potential of NK cell-related molecules in predicting BRCA outcomes and guiding personalized therapy remains largely unexplored. This study focused on developing a prognostic and therapeutic prediction model for BRCA by incorporating NK cell-related genes.MethodsThe data analyzed primarily originated from the TCGA and GEO databases. The prognostic role of NK cells was evaluated, and marker genes of NK cells were identified via single-cell analysis. Module genes closely associated with immunotherapy resistance were identified by bulk transcriptome-based weighted correlation network analysis (WGCNA). Following taking intersection and LASSO regression, NK-related genes (NKRGs) relevant to BRCA prognosis were screened, and the NK-related prognostic signature was subsequently constructed. Analyses were further expanded to clinicopathological relevance, GSEA, tumor microenvironment (TME) analysis, immune function, immunotherapy responsiveness, and chemotherapeutics. Key NKRGs were screened by machine learning and validated by spatial transcriptomics (ST) and immunohistochemistry (IHC).ResultsTumor-infiltrating NK cells are a favorable prognostic factor in BRCA. By combining scRNA-seq and bulk transcriptomic analyses, we identified 7 NK-related prognostic NKRGs (CCL5, EFHD2, KLRB1, C1S, SOCS3, IRF1, and CCND2) and developed an NK-related risk scoring (NKRS) system. The prognostic reliability of NKRS was verified through survival and clinical relevance analyses across multiple cohorts. NKRS also demonstrated robust predictive power in various aspects, including TME landscape, immune functions, immunotherapy responses, and chemotherapeutic sensitivity. Additionally, KLRB1 and CCND2 emerged as key prognostic NKRGs identified through machine learning and external validation, with their expression correlation with NK cells confirmed in BRCA specimens by ST and IHC.ConclusionsWe developed a novel NK-related gene signature that has proven valuable for evaluating prognosis and treatment response in BRCA, expecting to advance precision medicine of BRCA.

Antibiotic-induced loss of gut microbiome metabolic output correlates with clinical responses to CAR-T cell therapy

Antibiotic-induced microbiome dysbiosis is widespread in oncology, adversely affecting outcomes and side effects of various cancer treatments, including immune checkpoint inhibitors and chimeric antigen receptor T (CAR-T) cell therapies. In this study, we observed that prior exposure to broad-spectrum ABX with extended anaerobic coverage like piperacillin-tazobactam and meropenem was associated with worsened anti-CD19 CAR-T therapy survival outcomes in large B-cell lymphoma patients (n=422), compared to other ABX classes. In a discovery subset of these patients (n=67), we found that the use of these ABX was in turn associated with substantial dysbiosis of gut microbiome function, resulting in significant alterations of the gut and blood metabolome, including microbial effectors such as short-chain fatty acids (SCFAs) and other anionic metabolites, findings that were largely reproduced in an external validation cohort (n=58). Broader evaluation of circulating microbial metabolites revealed reductions in indole and cresol derivatives, as well as trimethylamine N-oxide, in patients who received ABX treatment (discovery n=40, validation n=28). These findings were recapitulated in an immune-competent CAR-T mouse model, where meropenem-induced dysbiosis led to a systemic dysmetabolome and decreased murine anti-CD19 CAR-T efficacy. Furthermore, we demonstrate that SCFAs can enhance the metabolic fitness of CAR-T cells, leading to improved tumor killing capacity. Together, these results suggest that broad-spectrum ABX deplete metabolically active commensals whose metabolites are essential for enhancing CAR-T efficacy, shedding light on the intricate relationship between ABX exposure, microbiome function and their impact on CAR-T cell efficacy. This highlights the potential for modulating the microbiome to augment CAR-T immunotherapy.

Partnering in construction re-visited: gauging progress in industry practice and prospects for advances in academic research

Despite the ubiquity of partnering and alliancing in industry discourse and academic research, questions remain about the extent of transformational change within the sector towards more collaborative working. Revisiting earlier work that highlighted issues, problems and dilemmas of partnering related to definitional ambiguities, conflicting (commercial) orientations and cultural reach and readiness, this paper highlights continuing problems of definition, formalization, translation and performance. Attention is directed towards the lack of external validation and institutionalization, as well as the need for more comparative analysis, situated understanding, awareness of organizational pluralism and recognition of relational dynamics. From this critical review, a framework is presented that embraces the variety and indeterminacy in the many definitions, pathways to collaboration, realizations in practice and evaluative recipes used. Partnering is presented instead as being constituted through complex and interacting bundles of practices that cut across levels of interaction, and which reflect competing (and contested) institutional influences, situated practices, outcomes/effects and performance evaluations. This practice-based approach is more attuned to the diversity and fluidity of the institutional contexts, organizational processes, and project/programme settings wherein partnering is situated and through which it is instantiated, and thus affords new avenues of research into its nature and effects.

Machine learning-based prediction of sarcopenia in community-dwelling middle-aged and older adults: findings from the CHARLS.

Sarcopenia is a prominent issue among aging populations and associated with poor health outcomes. This study aimed to examine the predictive value of questionnaire and biomarker data for sarcopenia, and to further develop a user-friendly calculator for community-dwelling middle-aged and older adults. We used two waves (2011 and 2013) of the China Health and Retirement Longitudinal Study (CHARLS) to predict sarcopenia, defined by the Asian Working Group for Sarcopenia 2019 criteria. We restricted the analytical sample to adults aged 45 or above (N = 2934). Five machine learning models were used to construct Q-based (only questionnaire variables), Bio-based (only biomarker variables), and combined (questionnaire plus biomarker variables) models. Area under the receiver operating characteristic curve (AUROC) was used for performance assessment. Temporal external validation was performed based on two datasets from CHARLS. Important predictors were identified by Shapley values and coefficients. Extreme gradient boosting (XGBoost), considering both questionnaire and biomarker characteristics, emerged as the optimal model, and its AUROC was 0.759 (95% CI: 0.747-0.771) at a decision threshold of 0.20 on the test set. Models also performed well on the external datasets. We found that cognitive function was the most important predictor in both Q-based and combined models, and blood urea nitrogen was the most important predictor in the Bio-based model. Other key predictors included education, haematocrit, total cholesterol, drinking, number of chronic diseases, and instrumental activities of daily living score. Our findings offer a potential for early screening and targeted prevention of sarcopenia among middle-aged and older adults in the community setting.

Improved Ensemble Model for Insecticide Recognition by Incorporating Insect Toxicity Data.

Pesticide molecules, such as insecticides, play a critical role in modern agricultural production. Traditional pesticide development methods are often inefficient and expensive, while data-driven artificial intelligence (AI) techniques have emerged as a useful tool to facilitate drug discovery. However, currently available commercial pesticide data is limited, which makes the trained models unsatisfactory in terms of performance and generalization. From a domain knowledge perspective, insect toxicity data were incorporated to improve the insecticide recognition of AI models. Compared to the models trained with the original data set, the new models performed better in the external validation, and their generalization was more desirable. In addition, by integrating different types of individual models, we obtained an ensemble model with better performance. Based on this, an online platform was developed to provide researchers with free access to insecticide screening (https://dpai.ccnu.edu.cn/InsectiVS/). Finally, two potential insecticide molecules with insecticidal activity against Plutella xylostella were successfully identified in a real-world scenario. In conclusion, this idea connects the fields of AI and agricultural chemistry and is expected to have wide application in pesticide research.

Development and validation of a cardiovascular risk prediction model for Sri Lankans using machine learning.

Sri Lankans do not have a specific cardiovascular (CV) risk prediction model and therefore, World Health Organization(WHO) risk charts developed for the Southeast Asia Region are being used. We aimed to develop a CV risk prediction model specific for Sri Lankans using machine learning (ML) of data of a population-based, randomly selected cohort of Sri Lankans followed up for 10 years and to validate it in an external cohort. The cohort consisted of 2596 individuals between 40-65 years of age in 2007, who were followed up for 10 years. Of them, 179 developed hard CV diseases (CVD) by 2017. We developed three CV risk prediction models named model 1, 2 and 3 using ML. We compared predictive performances between models and the WHO risk charts using receiver operating characteristic curves (ROC). The most predictive and practical model for use in primary care, model 3 was named "SLCVD score" which used age, sex, smoking status, systolic blood pressure, history of diabetes, and total cholesterol level in the calculation. We developed an online platform to calculate the SLCVD score. Predictions of SLCVD score were validated in an external hospital-based cohort. Model 1, 2, SLCVD score and the WHO risk charts predicted 173, 162, 169 and 10 of 179 observed events and the area under the ROC (AUC) were 0.98, 0.98, 0.98 and 0.52 respectively. During external validation, the SLCVD score and WHO risk charts predicted 56 and 18 respectively of 119 total events and AUCs were 0.64 and 0.54 respectively. SLCVD score is the first and only CV risk prediction model specific for Sri Lankans. It predicts the 10-year risk of developing a hard CVD in Sri Lankans. SLCVD score was more effective in predicting Sri Lankans at high CV risk than WHO risk charts.

Computational Study of Coumarin Compounds as Potential Inhibitors of Casein Kinase 2: DFT, 2D-QSAR, ADMET and Molecular Docking Investigations

Casein kinase 2 (CK2) is an ubiquitous, essential, and highly pleiotropic protein kinase whose abnormally high constitutive activity is suspected to underlie its pathogenic potential in neoplasia and other diseases. Recently, there has been a notable increase in interest in the use of casein kinase 2 (CK2) inhibitors to improve the treatment of a specific form of cancer while minimizing the risk of undesirable side effects. Recently, using different virtual screening approaches, we have identified several novel CK2 inhibitors. In particular, we have discovered that the coumarin moiety can be considered an attractive CK2 inhibitor scaffold. In the present work, quantitative structure-activity relationship (QSAR) analysis has been employed to envisage the inhibitory effects of 32 coumarin derivatives on the CK2 protein. The most efficient model is found by using multiple linear regression (MLR). Its capability is considered by the external and internal validation values found (R2 = 0.884, Q2cv = 0.822, R2pred = 0.821, and R2p = 0.811), which aligned well with Tropsha and Golbraikh’s approach. The highest docking score founded for the newly designed coumarins is −7.50 kcal mol-1, which indicates that candidates can bind to the CK2 receptor with greater affinity. Based on the results of the ADMET properties and drug similarity analyses, a DFT investigation was conducted to confirm the stability of the newly explored compounds. It appears that the most stable complexes are those of compound with the highest binding affinity with a lower risk of toxicité.