Identification of an ANCA-Associated Vasculitis Cohort Using Deep Learning and Electronic Health Records.

Background: ANCA-associated vasculitis (AAV) is a rare but serious disease. Traditional case-identification methods using claims data can be time-intensive and may miss important subgroups. We hypothesized that a deep learning model analyzing electronic health records (EHR) can more accurately identify AAV cases. Methods: We examined the Mass General Brigham (MGB) repository of clinical documentation from 12/1/1979 to 5/11/2021, using expert-curated keywords and ICD codes to identify a large cohort of potential AAV cases. Three labeled datasets (I, II, III) were created, each containing note sections. We trained and evaluated a range of machine learning and deep learning algorithms for note-level classification, using metrics like positive predictive value (PPV), sensitivity, F-score, area under the receiver operating characteristic curve (AUROC), and area under the precision and recall curve (AUPRC). The deep learning model was further evaluated for its ability to classify AAV cases at the patient-level, compared with rule-based algorithms in 2,000 randomly chosen samples. Results: Datasets I, II, and III comprised 6,000, 3,008, and 7,500 note sections, respectively. Deep learning achieved the highest AUROC in all three datasets, with scores of 0.983, 0.991, and 0.991. The deep learning approach also had among the highest PPVs across the three datasets (0.941, 0.954, and 0.800, respectively). In a test cohort of 2,000 cases, the deep learning model achieved a PPV of 0.262 and an estimated sensitivity of 0.975. Compared to the best rule-based algorithm, the deep learning model identified six additional AAV cases, representing 13% of the total. Conclusion: The deep learning model effectively classifies clinical note sections for AAV diagnosis. Its application to EHR notes can potentially uncover additional cases missed by traditional rule-based methods.

Unmasking Bias in AI: A Systematic Review of Bias Detection and Mitigation Strategies in Electronic Health Record-based Models.

OBJECTIVES: Leveraging artificial intelligence (AI) in conjunction with electronic health records (EHRs) holds transformative potential to improve healthcare. Yet, addressing bias in AI, which risks worsening healthcare disparities, cannot be overlooked. This study reviews methods to detect and mitigate diverse forms of bias in AI models developed using EHR data. METHODS: We conducted a systematic review following the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) guidelines, analyzing articles from PubMed, Web of Science, and IEEE published between January 1, 2010, and Dec 17, 2023. The review identified key biases, outlined strategies for detecting and mitigating bias throughout the AI model development process, and analyzed metrics for bias assessment. RESULTS: Of the 450 articles retrieved, 20 met our criteria, revealing six major bias types: algorithmic, confounding, implicit, measurement, selection, and temporal. The AI models were primarily developed for predictive tasks in healthcare settings. Four studies concentrated on the detection of implicit and algorithmic biases employing fairness metrics like statistical parity, equal opportunity, and predictive equity. Sixty proposed various strategies for mitigating biases, especially targeting implicit and selection biases. These strategies, evaluated through both performance (e.g., accuracy, AUROC) and fairness metrics, predominantly involved data collection and preprocessing techniques like resampling, reweighting, and transformation. DISCUSSION: This review highlights the varied and evolving nature of strategies to address bias in EHR-based AI models, emphasizing the urgent needs for the establishment of standardized, generalizable, and interpretable methodologies to foster the creation of ethical AI systems that promote fairness and equity in healthcare.

Designing biomimetic two-dimensional channels for uranium separation from seawater.

Efficient separation of uranium from seawater stands as a pivotal challenge. This study unveils an approach focusing on the ingenious design of biomimetic two-dimensional (2D) membranes tailored explicitly for this purpose. Leveraging the unique interplay of DNA strands housing U aptamers, pH-responsive i-motifs, and poly A(10) segments ingeniously embedded within graphene oxide membranes, a distinctive biomimetic 2D channel is engineered. The strategic integration of these bio-inspired elements enables dynamic adjustment of interlayer spacing, augmenting both the permeability of the membrane and the selectivity of the aptamer for uranyl ions. During the separation process, the encounter between uranyl ions and the enhanced aptamer within the interlayers initiates a crucial interaction, triggering a specific concentration polarization mechanism. This mechanism stands as the cornerstone for achieving a highly selective separation of uranyl ions from the vast and complex matrix of seawater. The membrane exhibits excellent performance in real seawater, with a rejection rate of uranyl ions of ≈100% and sustained selectivity of uranyl ions over ten cycles. Importantly, the selectivity of uranium and vanadium can reach 14.66. The significance of this research lies not only in the effective separation of uranyl ions but also in showcasing the broader applicability of 2D membrane design in chemical engineering.

Frailty Is More Predictive of Mortality than Age in Patients With Hip Fractures.

OBJECTIVES: To investigate the association between the Comprehensive Geriatric Assessment-based Frailty Index and adverse outcomes in older adult patients undergoing hip fracture surgery. DESIGN: Retrospective cohort study. SETTING: Academic Level 1 Trauma Center. PATIENTS: All patients aged 65 or older who underwent surgical repair of a hip fracture between May 2018 and August 2020 were identified through institutional database review. OUTCOME MEASURES AND COMPARISONS: Data including demographics, FI, injury presentation, and hospital course were collected. Patients were grouped by FI as nonfrail (FI < 0.21), frail (0.21 ≤ FI < 0.45), and severely frail (FI > 0.45). Adverse outcomes of these groups were compared using Kaplan Meier survival analysis. Risk factors for 1-year rehospitalization and 2-year mortality were evaluated using Cox hazard regression. RESULTS: Three hundred sixteen patients were included, with 62 nonfrail, 185 frail, and 69 severely frail patients. The total population was on average 83.8 years old, predominantly white (88.0%), and majority female (69.9%) with an average FI of 0.33 (SD: 0.14). The nonfrail cohort was on average 78.8 years old, 93.6% white, and 80.7% female; the frail cohort was on average 84.5 years old, 92.4% white, and 71.9% female; and the severely frail cohort was on average 86.4 years old, 71.0% white, and 55.1% female. Rate of 1-year readmission increased with frailty level, with a rate of 38% in nonfrail patients, 55.6% in frail patients, and 74.2% in severely frail patients (P = 0.001). The same pattern was seen in 2-year mortality rates, with a rate of 2.8% in nonfrail patients, 36.7% in frail patients, and 77.5% in severely frail patients (P < 0.0001). Being classified as frail or severely frail exhibited greater association with mortality within 2 years than age, with hazard ratio of 17.81 for frail patients and 56.81 for severely frail patients compared with 1.19 per 5 years of age. CONCLUSIONS: Increased frailty as measured by the Frailty Index is significantly associated with increased 2-year mortality and 1-year hospital readmission rates after hip fracture surgery. Degree of frailty predicts mortality more strongly than age alone. Assessing frailty with the Frailty Index can identify higher-risk surgical candidates, facilitate clinical decision making, and guide discussions about goals of care with family members, surgeons, and geriatricians. LEVEL OF EVIDENCE: Prognostic Level III. See Instructions for Authors for a complete description of levels of evidence.

Assessing fairness in machine learning models: A study of racial bias using matched counterparts in mortality prediction for patients with chronic diseases.

OBJECTIVE: Existing approaches to fairness evaluation often overlook systematic differences in the social determinants of health, like demographics and socioeconomics, among comparison groups, potentially leading to inaccurate or even contradictory conclusions. This study aims to evaluate racial disparities in predicting mortality among patients with chronic diseases using a fairness detection method that considers systematic differences. METHODS: We created five datasets from Mass General Brigham's electronic health records (EHR), each focusing on a different chronic condition: congestive heart failure (CHF), chronic kidney disease (CKD), chronic obstructive pulmonary disease (COPD), chronic liver disease (CLD), and dementia. For each dataset, we developed separate machine learning models to predict 1-year mortality and examined racial disparities by comparing prediction performances between Black and White individuals. We compared racial fairness evaluation between the overall Black and White individuals versus their counterparts who were Black and matched White individuals identified by propensity score matching, where the systematic differences were mitigated. RESULTS: We identified significant differences between Black and White individuals in age, gender, marital status, education level, smoking status, health insurance type, body mass index, and Charlson comorbidity index (p-value < 0.001). When examining matched Black and White subpopulations identified through propensity score matching, significant differences between particular covariates existed. We observed weaker significance levels in the CHF cohort for insurance type (p = 0.043), in the CKD cohort for insurance type (p = 0.005) and education level (p = 0.016), and in the dementia cohort for body mass index (p = 0.041); with no significant differences for other covariates. When examining mortality prediction models across the five study cohorts, we conducted a comparison of fairness evaluations before and after mitigating systematic differences. We revealed significant differences in the CHF cohort with p-values of 0.021 and 0.001 in terms of F1 measure and Sensitivity for the AdaBoost model, and p-values of 0.014 and 0.003 in terms of F1 measure and Sensitivity for the MLP model, respectively. DISCUSSION AND CONCLUSION: This study contributes to research on fairness assessment by focusing on the examination of systematic disparities and underscores the potential for revealing racial bias in machine learning models used in clinical settings.

A proof-of-concept study of sequential treatment with the HDAC inhibitor vorinostat following BRAF and MEK inhibitors in BRAFV600mutated melanoma.

PURPOSE: Development of resistance limits the clinical benefit of BRAF and MEK inhibitors (BRAFi/MEKi) in BRAFV600 mutated melanoma. It has been shown that short-term treatment (14 days) with vorinostat was able to initiate apoptosis of the resistant tumor cells. We aimed to assess the anti-tumor activity of sequential treatment with vorinostat following BRAFi/MEKi in patients with BRAFV600 melanoma who progressed after initial response to BRAFi/MEKi. PATIENTS AND METHODS: Patients with BRAFi/MEKi resistant BRAFV600 melanoma were treated with vorinostat 360 mg QD for 14 days followed by BRAFi/MEKi. The primary endpoint was an objective response rate of progressive lesions of at least 30% according to RECIST 1.1. Secondary endpoints included progression-free survival (PFS), overall survival (OS), safety, pharmacokinetics of vorinostat and translational molecular analyses using ctDNA and tumor biopsies. RESULTS: Twenty-six patients with progressive BRAFi/MEKi resistant BRAFV600 mutated melanoma received treatment with vorinostat. Twenty-two patients were evaluable for response. The ORR was 9% (one complete response for 31.2 months and one partial response for 14.9 months. Median PFS and OS were 1.4 and 5.4 months, respectively. Common adverse events were fatigue (23%) and nausea (19%). ctDNA analysis showed emerging secondary mutations in NRAS and MEK in eight patients at time of BRAFi/MEKi resistance. Elimination of these mutations by vorinostat treatment was observed in three patients. CONCLUSIONS: Intermittent treatment with vorinostat in patients with resistant BRAFV600mutated melanoma is well tolerated. Although the primary endpoint of this study was not met, durable anti-tumor responses were observed in a minority of patients (9%).

Imageable Brachytherapy with Chelator-Free Radiolabeling Hydrogel.

Brachytherapy stands as an essential clinical approach for combating locally advanced tumors. Here, an injectable brachytherapy hydrogel is developed for the treatment of both local and metastatic tumor. Fe-tannins nanoparticles are efficiently and stably radiolabeled with clinical used therapeutic radionuclides (such as 131I, 90Y, 177Lu, and 225Ac) without a chelator, and then chemically cross-linked with 4-armPEG-SH to form brachytherapy hydrogel. Upon intratumoral administration, magnetic resonance imaging (MRI) signal from ferric ions embedded within the hydrogel directly correlates with the retention dosage of radionuclides, which can real-time monitor radionuclides emitting short-range rays in vivo without penetration limitation during brachytherapy. The hydrogel's design ensures the long-term tumor retention of therapeutic radionuclides, leading to the effective eradication of local tumor. Furthermore, the radiolabeled hydrogel is integrated with an adjuvant to synergize with immune checkpoint blocking therapy, thereby activating potent anti-tumor immune responses and inhibiting metastatic tumor growth. Therefore, this work presents an imageable brachytherapy hydrogel for real-time monitoring therapeutic process, and expands the indications of brachytherapy from treatment of localized tumors to metastatic tumors.

Danshensu reduces neuronal excitability by enhancing potassium currents in bushy cells in the mouse cochlear nucleus.

Danshensu, also known as salvianic acid A, is a primary active compound extracted from a traditional Chinese herb Danshen (Salvia miltiorrhiza). While its antioxidative and neuroprotective effects are well-documented, the underlying mechanisms are poorly understood. In this study, we sought out to investigate if and how Danshensu modulates neuronal excitability and voltage-gated ionic currents in the central nervous system. We prepared brain slices of the mouse brainstem and performed patch-clamp recording in bushy cells in the anteroventral cochlear nucleus, with or without Danshensu incubation for 1 h. QX-314 was used internally to block Na+ current, while tetraethylammonium and 4-aminopyridine were used to isolate different subtypes of K+ current. We found that Danshensu of 100 µm decreased the input resistance of bushy cells by approximately 60% and shifted the voltage threshold of spiking positively by approximately 7 mV, resulting in significantly reduced excitability. Furthermore, we found this reduced excitability by Danshensu was caused by enhanced voltage-gated K+ currents in these neurons, including both low voltage-activated IK,A, by approximately 100%, and high voltage-activated IK,dr, by approximately 30%. Lastly, we found that the effect of Danshensu on K+ currents was dose-dependent in that no enhancement was found for Danshensu of 50 µm and Danshensu of 200 µm failed to cause significantly more enhancement on K+ currents when compared to that of 100 µm. We found that Danshensu reduced neuronal excitability in the central nervous system by enhancing voltage-gated K+ currents, providing mechanistic support for its neuroprotective effect widely seen in vivo.

Identification and bioinformatics analysis of lncRNAs in serum of patients with ankylosing spondylitis.

OBJECTIVES: The aim of this study was to explore the long non-coding RNA (lncRNA) expression profiles in serum of patients with ankylosing spondylitis (AS). The role of these lncRNAs in this complex autoimmune situation needs to be evaluated. METHODS: We used high-throughput whole-transcriptome sequencing to generate sequencing data from three patients with AS and three normal controls (NC). Then, we performed bioinformatics analyses to identify the functional and biological processes associated with differentially expressed lncRNAs (DElncRNAs). We confirmed the validity of our RNA-seq data by assessing the expression of eight lncRNAs via quantitative reverse transcription polymerase chain reaction (qRT-PCR) in 20 AS and 20 NC samples. We measured the correlation between the expression levels of lncRNAs and patient clinical index values using the Spearman correlation test. RESULTS: We identified 72 significantly upregulated and 73 significantly downregulated lncRNAs in AS patients compared to NC. qRT-PCR was performed to validate the expression of selected DElncRNAs; the results demonstrated that the expression levels of MALAT1:24, NBR2:9, lnc-DLK1-35:13, lnc-LARP1-1:1, lnc-AIPL1-1:7, and lnc-SLC12A7-1:16 were consistent with the sequencing analysis results. Enrichment analysis showed that DElncRNAs mainly participated in the immune and inflammatory responses pathways, such as regulation of protein ubiquitination, major histocompatibility complex class I-mediated antigen processing and presentation, MAPkinase activation, and interleukin-17 signaling pathways. In addition, a competing endogenous RNA network was constructed to determine the interaction among the lncRNAs, microRNAs, and mRNAs based on the confirmed lncRNAs (MALAT1:24 and NBR2:9). We further found the expression of MALAT1:24 and NBR2:9 to be positively correlated with disease severity. CONCLUSION: Taken together, our study presents a comprehensive overview of lncRNAs in the serum of AS patients, thereby contributing novel perspectives on the underlying pathogenic mechanisms of this condition. In addition, our study predicted MALAT1 has the potential to be deeply involved in the pathogenesis of AS.

QDPR deficiency drives immune suppression in pancreatic cancer.

The relevance of biopterin metabolism in resistance to immune checkpoint blockade (ICB) therapy remains unknown. We demonstrate that the deficiency of quinoid dihydropteridine reductase (QDPR), a critical enzyme regulating biopterin metabolism, causes metabolite dihydrobiopterin (BH2) accumulation and decreases the ratio of tetrahydrobiopterin (BH4) to BH2 in pancreatic ductal adenocarcinomas (PDACs). The reduced BH4/BH2 ratio leads to an increase in reactive oxygen species (ROS) generation and a decrease in the distribution of H3K27me3 at CXCL1 promoter. Consequently, myeloid-derived suppressor cells are recruited to tumor microenvironment via CXCR2 causing resistance to ICB therapy. We discovered that BH4 supplementation is capable to restore the BH4/BH2 ratio, enhance anti-tumor immunity, and overcome ICB resistance in QDPR-deficient PDACs. Tumors with lower QDPR expression show decreased responsiveness to ICB therapy. These findings offer a novel strategy for selecting patient and combining therapies to improve the effectiveness of ICB therapy in PDAC.

Enhancing Early Detection of Cognitive Decline in the Elderly: A Comparative Study Utilizing Large Language Models in Clinical Notes.

Background: Large language models (LLMs) have shown promising performance in various healthcare domains, but their effectiveness in identifying specific clinical conditions in real medical records is less explored. This study evaluates LLMs for detecting signs of cognitive decline in real electronic health record (EHR) clinical notes, comparing their error profiles with traditional models. The insights gained will inform strategies for performance enhancement. Methods: This study, conducted at Mass General Brigham in Boston, MA, analyzed clinical notes from the four years prior to a 2019 diagnosis of mild cognitive impairment in patients aged 50 and older. We used a randomly annotated sample of 4,949 note sections, filtered with keywords related to cognitive functions, for model development. For testing, a random annotated sample of 1,996 note sections without keyword filtering was utilized. We developed prompts for two LLMs, Llama 2 and GPT-4, on HIPAA-compliant cloud-computing platforms using multiple approaches (e.g., both hard and soft prompting and error analysis-based instructions) to select the optimal LLM-based method. Baseline models included a hierarchical attention-based neural network and XGBoost. Subsequently, we constructed an ensemble of the three models using a majority vote approach. Results: GPT-4 demonstrated superior accuracy and efficiency compared to Llama 2, but did not outperform traditional models. The ensemble model outperformed the individual models, achieving a precision of 90.3%, a recall of 94.2%, and an F1-score of 92.2%. Notably, the ensemble model showed a significant improvement in precision, increasing from a range of 70%-79% to above 90%, compared to the best-performing single model. Error analysis revealed that 63 samples were incorrectly predicted by at least one model; however, only 2 cases (3.2%) were mutual errors across all models, indicating diverse error profiles among them. Conclusions: LLMs and traditional machine learning models trained using local EHR data exhibited diverse error profiles. The ensemble of these models was found to be complementary, enhancing diagnostic performance. Future research should investigate integrating LLMs with smaller, localized models and incorporating medical data and domain knowledge to enhance performance on specific tasks.

Retrospective analysis of predictive factors for AVF dysfunction in patients undergoing MHD.

To construct an early clinical prediction model for AVF dysfunction in patients undergoing Maintenance Hemodialysis (MHD) and perform internal and external verifications. We retrospectively examined clinical data from 150 patients diagnosed with MHD at Hefei Third People's Hospital from January 2014 to June 2023. Depending on arteriovenous fistula (AVF) functionality, patients were categorized into dysfunctional (n = 62) and functional (n = 88) cohorts. Using the least absolute shrinkage and selection operator(LASSO) regression model, variables potentially influencing AVF functionality were filtered using selected variables that underwent multifactorial logistic regression analysis. The Nomogram model was constructed using the R software, and the Area Under Curve(AUC) value was calculated. The model's accuracy was appraised through the calibration curve and Hosmer-Lemeshow test, with the model undergoing internal validation using the bootstrap method. There were 11 factors exhibiting differences between the group of patients with AVF dysfunction and the group with normal AVF function, including age, sex, course of renal failure, diabetes, hyperlipidemia, Platelet count (PLT), Calcium (Ca), Phosphorus, D-dimer (D-D), Fibrinogen (Fib), and Anastomotic width. These identified factors are included as candidate predictive variables in the LASSO regression analysis. LASSO regression identified age, sex, diabetes, hyperlipidemia, anastomotic diameter, blood phosphorus, and serum D-D levels as 7 predictive factors. Unconditional binary logistic regression analysis revealed that advanced age (OR = 4.358, 95% CI: 1.454-13.062), diabetes (OR = 4.158, 95% CI: 1.243-13.907), hyperlipidemia (OR = 3.651, 95% CI: 1.066-12.499), D-D (OR = 1.311, 95% CI: 1.063-1.616), and hyperphosphatemia (OR = 4.986, 95% CI: 2.513-9.892) emerged as independent risk factors for AVF dysfunction in MHD patients. The AUC of the predictive model was 0.934 (95% CI: 0.897-0.971). The Hosmer-Lemeshow test showed high consistency between the model's predictive results and actual clinical observations (χ2 = 1.553, P = .092). Internal validation revealed an AUC of 0.911 (95% CI: 0.866-0.956), with the Calibration calibration curve nearing the ideal curve. Advanced age, coexisting diabetes, hyperlipidemia, blood D-D levels, and hyperphosphatemia are independent risk factors for AVF dysfunction in patients undergoing MHD.

Developmental pleiotropy of SDP1 from seedling to mature stages in B. napus.

Rapeseed, an important oil crop, relies on robust seedling emergence for optimal yields. Seedling emergence in the field is vulnerable to various factors, among which inadequate self-supply of energy is crucial to limiting seedling growth in early stage. SUGAR-DEPENDENT1 (SDP1) initiates triacylglycerol (TAG) degradation, yet its detailed function has not been determined in B. napus. Here, we focused on the effects of plant growth during whole growth stages and energy mobilization during seedling establishment by mutation in BnSDP1. Protein sequence alignment and haplotypic analysis revealed the conservation of SDP1 among species, with a favorable haplotype enhancing oil content. Investigation of agronomic traits indicated bnsdp1 had a minor impact on vegetative growth and no obvious developmental defects when compared with wild type (WT) across growth stages. The seed oil content was improved by 2.0-2.37% in bnsdp1 lines, with slight reductions in silique length and seed number per silique. Furthermore, bnsdp1 resulted in lower seedling emergence, characterized by a shrunken hypocotyl and poor photosynthetic capacity in the early stages. Additionally, impaired seedling growth, especially in yellow seedlings, was not fully rescued in medium supplemented with exogenous sucrose. The limited lipid turnover in bnsdp1 was accompanied by induced amino acid degradation and PPDK-dependent gluconeogenesis pathway. Analysis of the metabolites in cotyledons revealed active amino acid metabolism and suppressed lipid degradation, consistent with the RNA-seq results. Finally, we proposed strategies for applying BnSDP1 in molecular breeding. Our study provides theoretical guidance for understanding trade-off between oil accumulation and seedling energy mobilization in B. napus.

Looking Beyond Mortality Prediction: Primary Care Physician Views of Patients' Palliative Care Needs Predicted by a Machine Learning Tool.

OBJECTIVES: To assess primary care physicians' (PCPs) perception of the need for serious illness conversations (SIC) or other palliative care interventions in patients flagged by a machine learning tool for high 1-year mortality risk. METHODS: We surveyed PCPs from four Brigham and Women's Hospital primary care practice sites. Multiple mortality prediction algorithms were ensembled to assess adult patients of these PCPs who were either enrolled in the hospital's integrated care management program or had one of several chronic conditions. The patients were classified as high or low risk of 1-year mortality. A blinded survey had PCPs evaluate these patients for palliative care needs. We measured PCP and machine learning tool agreement regarding patients' need for an SIC/elevated risk of mortality. RESULTS: Of 66 PCPs, 20 (30.3%) participated in the survey. Out of 312 patients evaluated, 60.6% were female, with a mean (standard deviation [SD]) age of 69.3 (17.5) years, and a mean (SD) Charlson Comorbidity Index of 2.80 (2.89). The machine learning tool identified 162 (51.9%) patients as high risk. Excluding deceased or unfamiliar patients, PCPs felt that an SIC was appropriate for 179 patients; the machine learning tool flagged 123 of these patients as high risk (68.7% concordance). For 105 patients whom PCPs deemed SIC unnecessary, the tool classified 83 as low risk (79.1% concordance). There was substantial agreement between PCPs and the tool (Gwet's agreement coefficient of 0.640). CONCLUSIONS: A machine learning mortality prediction tool offers promise as a clinical decision aid, helping clinicians pinpoint patients needing palliative care interventions.

Targeting of vulnerabilities of drug-tolerant persisters identified through functional genetics delays tumor relapse.

Drug-tolerant persisters (DTPs) are a rare subpopulation of cells within a tumor that can survive therapy through nongenetic adaptive mechanisms to develop relapse and repopulate the tumor following drug withdrawal. Using a cancer cell line with an engineered suicide switch to kill proliferating cells, we perform both genetic screens and compound screens to identify the inhibition of bromodomain and extraterminal domain (BET) proteins as a selective vulnerability of DTPs. BET inhibitors are especially detrimental to DTPs that have reentered the cell cycle (DTEPs) in a broad spectrum of cancer types. Mechanistically, BET inhibition induces lethal levels of ROS through the suppression of redox-regulating genes highly expressed in DTPs, including GPX2, ALDH3A1, and MGST1. In vivo BET inhibitor treatment delays tumor relapse in both melanoma and lung cancer. Our study suggests that combining standard of care therapy with BET inhibitors to eliminate residual persister cells is a promising therapeutic strategy.

Unmasking bias in artificial intelligence: a systematic review of bias detection and mitigation strategies in electronic health record-based models.

OBJECTIVES: Leveraging artificial intelligence (AI) in conjunction with electronic health records (EHRs) holds transformative potential to improve healthcare. However, addressing bias in AI, which risks worsening healthcare disparities, cannot be overlooked. This study reviews methods to handle various biases in AI models developed using EHR data. MATERIALS AND METHODS: We conducted a systematic review following the Preferred Reporting Items for Systematic Reviews and Meta-analyses guidelines, analyzing articles from PubMed, Web of Science, and IEEE published between January 01, 2010 and December 17, 2023. The review identified key biases, outlined strategies for detecting and mitigating bias throughout the AI model development, and analyzed metrics for bias assessment. RESULTS: Of the 450 articles retrieved, 20 met our criteria, revealing 6 major bias types: algorithmic, confounding, implicit, measurement, selection, and temporal. The AI models were primarily developed for predictive tasks, yet none have been deployed in real-world healthcare settings. Five studies concentrated on the detection of implicit and algorithmic biases employing fairness metrics like statistical parity, equal opportunity, and predictive equity. Fifteen studies proposed strategies for mitigating biases, especially targeting implicit and selection biases. These strategies, evaluated through both performance and fairness metrics, predominantly involved data collection and preprocessing techniques like resampling and reweighting. DISCUSSION: This review highlights evolving strategies to mitigate bias in EHR-based AI models, emphasizing the urgent need for both standardized and detailed reporting of the methodologies and systematic real-world testing and evaluation. Such measures are essential for gauging models' practical impact and fostering ethical AI that ensures fairness and equity in healthcare.

Adsorption of humic acid from different organic solid waste compost to phenanthrene, is fluorescence excitation or quenching?

Humic acid (HA) from different organic solid waste (OSW) compost has been shown good adsorption properties for phenanthrene. However, the raw material of HA can affect its structure, resulting in differences in adsorption capacity. Therefore, this study focused on the adsorption characteristics of phenanthrene by HA from different OSW compost. In this work, chicken manure (CM), rice straw (RS) and lawn waste (LW) were selected as sources of composted HA. The adsorption mechanism of HA from different OSW compost were revealed through analytical techniques including three-dimensional fluorescence spectroscopy (EEM), two-dimensional correlation spectroscopy (2DCOS), and Fourier-transform infrared spectroscopy (FTIR). The results suggested that HA from LW compost had a better adsorption affinity for phenanthrene because of its more complex fluorescent component, where C1 as a simple component determined the adsorption process specifically. Furthermore, after HA from LW compost adsorbed phenanthrene, the increase in aromatic -COOH and -NH was the main reason for fluorescence quenching. These results indicated that HA from LW compost had better adsorption effect for phenanthrene. The results of this study were expected to provide a selection scheme for the control of phenanthrene pollution and environmental remediation.

Exploration of the Healthy Donor Effect Among 0.6 Million Blood Donors in China: Longitudinal Study.

BACKGROUND: The World Health Organization emphasizes the importance of completely voluntary blood donation to maintain safe and sustainable blood supplies. However, the benefits of blood donation for donors, such as reducing the risk of disease, remain a topic of debate due to the existence of the healthy donor effect (HDE). This effect arises because of inherent health differences between blood donors and the general population, and it is also considered a methodological issue. OBJECTIVE: This study aims to generate a more detailed health profile of blood donors from a donor cohort study to mitigate and quantify the HDE and properly interpret the association between blood donation and disease outcomes among blood donors. METHODS: A retrospective cohort study was conducted between January 2012 and December 2018 among donors before their first donation. One-to-one propensity score matching was conducted through a random selection of individuals without any history of blood donation, as reported from their electronic health records. We conducted a Poisson regression between blood donors and non-blood donors before the first donation to estimate the adjusted incidence rate ratio (AIRR) of selected blood donation-related diseases, as defined by 13 categories of International Classification of Diseases, Tenth Revision (ICD-10) codes. RESULTS: Of the 0.6 million blood donors, 15,115 had an inpatient record before their first donation, whereas 17,356 non-blood donors had an inpatient record. For the comparison between blood donors and the matched non-blood donors, the HDE (the disease incidence rate ratio between non-blood donors and blood donors) was an AIRR of 1.152 (95% CI 1.127-1.178; P<.001). Among disease categories not recommended for blood donation in China, the strongest HDE was observed in the ICD-10 D50-D89 codes, which pertain to diseases of the blood and blood-forming organs as well as certain disorders involving the immune mechanism (AIRR 3.225, 95% CI 2.402-4.330; P<.001). After age stratification, we found that people who had their first blood donation between 46-55 years old had the strongest HDE (AIRR 1.816, 95% CI 1.707-1.932; P<.001). Both male and female donors had significant HDE (AIRR 1.082, 95% CI 1.05-1.116; P=.003; and AIRR 1.236, 95% CI 1.196-1.277; P<.001, respectively) compared with matched non-blood donors. CONCLUSIONS: : Our research findings suggest that the HDE is present among blood donors, particularly among female donors and those who first donated blood between the ages of 46 and 55 years. TRIAL REGISTRATION: Chinese Clinical Trial Registry ChiCTR2200055983; https://www.chictr.org.cn/showproj.html?proj=51760.

Characteristics of MDR E. coli strains isolated from Pet Dogs with clinic diarrhea: A pool of antibiotic resistance genes and virulence-associated genes.

The increasing number of multi-drug resistant (MDR) bacteria in companion animals poses a threat to both pet treatment and public health. To investigate the characteristics of MDR Escherichia coli (E. coli) from dogs, we detected the antimicrobial resistance (AMR) of 135 E. coli isolates from diarrheal pet dogs by disc diffusion method (K-B method), and screened antibiotic resistance genes (ARGs), virulence-associated genes (VAGs), and population structure (phylogenetic groups and MLST) by polymerase chain reaction (PCR) for 74 MDR strains, then further analyzed the association between AMRs and ARGs or VAGs. Our results showed that 135 isolates exhibited high resistance to AMP (71.11%, 96/135), TET (62.22%, 84/135), and SXT (59.26%, 80/135). Additionally, 54.81% (74/135) of the isolates were identified as MDR E. coli. In 74 MDR strains, a total of 12 ARGs in 6 categories and 14 VAGs in 4 categories were observed, of which tetA (95.95%, 71/74) and fimC (100%, 74/74) were the most prevalent. Further analysis of associations between ARGs and AMRs or VAGs in MDR strains revealed 23 significant positive associated pairs were observed between ARGs and AMRs, while only 5 associated pairs were observed between ARGs and VAGs (3 positive associated pairs and 2 negative associated pairs). Results of population structure analysis showed that B2 and D groups were the prevalent phylogroups (90.54%, 67/74), and 74 MDR strains belonged to 42 STs (6 clonal complexes and 23 singletons), of which ST10 was the dominant lineage. Our findings indicated that MDR E. coli from pet dogs carry a high diversity of ARGs and VAGs, and were mostly belong to B2/D groups and ST10. Measures should be taken to prevent the transmission of MDR E. coli between companion animals and humans, as the fecal shedding of MDR E. coli from pet dogs may pose a threat to humans.

Study on the electrodeposition of uranium in chloride molten salt.

This study focuses on the recovery of UO2 from oxide spent fuel using electrodeposition. U3O8 was used as the initial material and dissolved in NaCl-2CsCl using NH4Cl at high temperatures by means of chlorination reaction. The electrolysis process was conducted using a three-electrode system to investigate the effects of cathode material and diameter, electrolysis temperature, electrolysis time, electrolysis voltage, and uranium concentration in the molten salt on the electrolysis reaction. By optimizing the electrolysis conditions, pure UO2 with a recovery efficiency of 97% was obtained, and the products were characterized using XRD, SEM-EDS, ICP-AES and XPS. It was found that within the scope of this experiment, increasing the cathode diameter, extending the electrolysis time, and increasing the reduction voltage appropriately all led to an improvement in the recovery efficiency of the electrolysis reaction, while other conditions had minimal effect on the reaction. Furthermore, doping of the electrolyte system was performed by adding La, Ce and Nd elements, while the removal of La showed good purification effects, with a maximum decontamination factor of 119. Furthermore, the system showed good purification effects for Nd, with a decontamination factor of 57.