Recent Publications

Sage Open Aging. 2025 Jun 9;11:30495334251345091. doi: 10.1177/30495334251345091. eCollection 2025 Jan-Dec.

ABSTRACT

Polypharmacy is the concurrent use of many drugs increasingly prevalent in elderly populations worldwide, with over 50% of older adults taking more than five drugs. The rise in chronic conditions such as diabetes, hypertension, and dementia largely drives this trend. However, polypharmacy poses significant risks, including adverse drug events, falls, cognitive decline, and hospitalizations, particularly due to age-related physiological changes that alter drug metabolism and clearance. Vulnerable populations, especially those with comorbidities, face heightened risks of complications associated with polypharmacy, underscoring the need for effective management strategies. Despite extensive literature on polypharmacy, gaps remain in practical, evidence-based approaches to safely reduce medication burdens. Structured deprescribing protocols are emerging as a critical intervention for reducing inappropriate medication use, improving patient outcomes, and lowering healthcare costs. These protocols involve stepwise algorithms and tools for screening to find potentially inappropriate medications and clinical decision-making frameworks to prioritize deprescribing as it can reduce fall risk, hospitalizations, and healthcare costs while enhancing quality of life. This narrative review examines evidence from trials, cohort studies, and meta-analyses on deprescribing protocols, highlighting challenges like patient resistance and withdrawal effects. It emphasizes tailored approaches, patient engagement, and shared decision-making for safe, effective, patient-centered deprescribing across healthcare settings.

PMID:40611862 | PMC:PMC12220885 | DOI:10.1177/30495334251345091

Sci Rep. 2025 Jul 2;15(1):23673. doi: 10.1038/s41598-025-07436-1.

ABSTRACT

Integrating artificial intelligence (AI) into sustainable urban development presents an innovative pathway for addressing global environmental and socio-economic challenges. This study examines how AI technologies-such as machine learning, the Internet of Things (IoT), and big data analytics-can advance the three zeros method, a sustainability framework proposed by Nobel laureate Muhammad Yunus, which focuses on zero carbon emissions, zero poverty, and zero waste. By analyzing panel data across 50 countries and incorporating case studies, the research highlights AI's role in promoting carbon neutrality, economic inclusivity, and waste reduction. The findings reveal that AI-driven R&D innovation exerts the most decisive influence on sustainability, followed by AI-powered infrastructure, while market advantage plays a comparatively more minor role. Additionally, the study uncovers regional disparities in AI's impact, with the most significant benefits observed in countries at upper-middle levels of sustainable development. Moreover, urbanization serves as a threshold factor, altering AI's effects on sustainability. When urbanization is below a critical level, AI-driven innovation and infrastructure support sustainability, whereas the AI market advantage inhibits it. However, infrastructure may hinder sustainable development beyond this threshold while AI market mechanisms become more influential. These insights underscore the need for policymakers to tailor AI-driven sustainability strategies based on urbanization dynamics and regional development levels.

PMID:40603460 | DOI:10.1038/s41598-025-07436-1

PLoS One. 2025 Jul 1;20(7):e0327348. doi: 10.1371/journal.pone.0327348. eCollection 2025.

ABSTRACT

BACKGROUND: Tuberculosis (TB) regained its position as the leading cause of death globally from a single infectious disease agent in 2024. Delayed diagnosis and treatment hamper effective TB control. We investigated the duration of diagnostic and treatment delay along with the associated factors among people with pulmonary TB in Bangladesh.

METHODS: A mixed-method study was conducted between December'19 and March'21, at icddr,b TB Screening and Treatment Centres (TBSTCs), Dhaka. We interviewed people with TB (PWTB) seeking care at these TBSTCs using a structured questionnaire to collect data on socio-demographic, clinical and healthcare seeking behaviors. We used established frameworks to define stages of delay and associated factors. Qualitative interviews were conducted among a subset of participants to gain further insight into the factors associated with delay.

RESULTS: We enrolled 895 PWTB with mean (±SD) age 36.6 (±16.1) years; majority of participants were males (69.9%) and living in urban areas (82.3%). The median (IQR) patient delay estimated was 47 (29-72) days, with diagnostic delay 45 (30-70) days and treatment delay 2 (2-4) days. The predictors of delay were those with diabetes (OR 2.0, 95% CI - 1.11, 3.42), who initially self-treated (OR 2.1, 95% CI - 1.09, 3.88), and were bacteriologically diagnosed (OR 3.7, 95% CI - 1.31, 10.46). Qualitative approach supported the quantitative findings and revealed the practice of visiting formal physicians during worsening illness, neglecting to acknowledge signs or symptoms consistent with TB, lack of TB related knowledge, and financial insolvency as major reasons for delay.

CONCLUSION: Our findings showed that improper health-seeking behavior is one of the major drivers of patient delay. Thus, targeted programmatic intervention to raise community awareness on TB and its care services with a special focus on informal providers can help reduce this delay.

PMID:40591627 | DOI:10.1371/journal.pone.0327348

J Cardiothorac Vasc Anesth. 2025 May 8:S1053-0770(25)00363-5. doi: 10.1053/j.jvca.2025.05.006. Online ahead of print.

ABSTRACT

Genetic polymorphisms within the β1-adrenoceptor are common within the population. Although not directly causative of disease, accumulating evidence supports that they have significant molecular and clinical effects, including altering the response to inotropes and β-blockers, as well as altering exercise capacity. Here, we summarize current evidence as relevant to anesthetists who treat patients with heart failure.

PMID:40581538 | DOI:10.1053/j.jvca.2025.05.006

J Adv Vet Anim Res. 2025 Mar 24;12(1):80-89. doi: 10.5455/javar.2025.l874. eCollection 2025 Mar.

ABSTRACT

OBJECTIVE: The goal of this study is to describe the genome of Streptococcus agalactiae that was found in clinical mastitis in cattle in Bangladesh. This work will show how strong the bacteria are and how important they are for public health.

MATERIALS AND METHODS: Whole genome sequencing (WGS) was performed using the Illumina MiSeq platform, followed by comprehensive analysis with various bioinformatic tools to identify key genomic features.

RESULTS: WGS revealed that the isolates are closely related, belonging to sequence type ST4, a rare type previously identified in both human and animal hosts. The isolates possess 44 virulence-related genes linked to adherence, capsule biogenesis, enzyme production, immunoreactive antigens, protease, and cytolysin production. They also carry two pilus islands (PIs), PI-1 and PI-2b, which are often associated with invasive diseases. PI-2b proteins are key targets for vaccine development against Group B Streptococcus (GBS). The isolates belong to serotype Ia and carry the gbs2018-2 variant, indicating their adaptability to a wide range of hosts, including humans and animals. These virulence factors are critical for understanding S. agalactiae's pathogenicity and developing vaccines against its infections. Additionally, the isolates harbor antimicrobial resistance genes conferring resistance to glycopeptides (vanT, vanY), macrolides (mreA), peptides (mprF), penicillins and β-lactams (pbp), and aminoglycosides. Source tracking via the BacWGSTdb website identified these isolates as closely related to human pathogens, indicating their zoonotic potential.

CONCLUSION: These results suggest that S. agalactiae could be a zoonotic pathogen. This highlights the need for ongoing genomic surveillance to fully understand how it causes disease and come up with effective ways to control it.

PMID:40568491 | PMC:PMC12186781 | DOI:10.5455/javar.2025.l874

Adv Sci (Weinh). 2025 Jun 25:e03957. doi: 10.1002/advs.202503957. Online ahead of print.

ABSTRACT

Mutational variants of human lysozyme cause a rare but fatal hereditary systemic amyloidosis by populating an intermediate state that self-assembles into amyloid fibrils. However, this intermediate state is recalcitrant to detailed structural investigation, as it is only transiently and sparsely populated. Here, this work investigates the intermediate state of an amyloid-forming human lysozyme variant (I59T) using CEST and CPMG RD NMR at low pH. 15N CEST profiles probe the thermal unfolding of the native state into the denatured ensemble and reveal a distinct intermediate state. Global fitting of 15N CEST and CPMG data provides kinetic and thermodynamic parameters, characterizing the intermediate state populated at 0.6%. 1H CEST data further confirm the presence of the intermediate state displaying unusually high or low 1HN chemical shifts. To investigate the structural details of this intermediate state, this work uses molecular dynamics (MD) simulations, which recapitulate the experimentally observed folding pathway and free energy landscape. This work observes a high-energy intermediate state with a locally disordered β-domain and C-helix, stabilized by non-native hydrogen bonding and amide-π interactions, accounting for its anomalous 1H chemical shifts. Together, these NMR and MD data provide the first direct structural information on the intermediate state, offering insights into targeting lysozyme amyloidosis.

PMID:40557600 | DOI:10.1002/advs.202503957

Cureus. 2025 May 23;17(5):e84696. doi: 10.7759/cureus.84696. eCollection 2025 May.

ABSTRACT

Introduction Polymorphisms of the leptin receptor (LEPR) gene are associated with type 2 diabetes mellitus (T2DM), but the association varies among different geographic populations. The present study aims to observe the association of single-nucleotide polymorphisms (SNPs) of the LEPR gene (rs1137100 and rs1137101) with youth-onset T2DM in Bangladesh. Methods This case-control study encompassed 62 individuals with youth-onset T2DM (age range 18-29 years) and an equal number of age-matched controls with normal glucose tolerance (NGT). Genotyping was done by polymerase chain reaction-restriction fragment length polymorphism (PCR-RFLP). Genotypes of both LEPR SNPs were expressed as AA, AG, and GG, where G is considered a risk allele. Results The frequency of G-allele was higher in DM than in NGT for both rs1137100 (55.6% (69/124) vs. 42.7% (53/124); OR 1.7, 95% CI 1.02-2.78; p=0.042) and rs1137101 (59.7% (74/124) vs. 41.9% (52/124); OR 95% CI 1.24-3.40, p=0.005). In the codominant model, the GG genotype was associated with DM (GG vs. AA: rs1137100: OR 3.37; CI 1.11-10.19; p=0.032; rs1137101: OR 4.93; CI 1.62-14.99; p=0.005) but not the AG genotype (AG vs. AA). In the dominant model, the risk variants AG+GG (vs. AA) of rs1137100 did not have an association (p=0.289), but rs1137101 had (OR 2.60; CI 1.07-6.33; p=0.035). In the recessive model, risk variant GG (vs. AG+AA) of both SNPs had an association with DM (rs1137100: OR 2.98; CI 1.19-7.47; p=0.017; rs1137101: OR 3.02; CI 1.25-7.27; p=0.014). No association was significant in any models when adjusted for body mass index (BMI). Conclusion Although the LEPR gene SNPs rs1137100 and rs1137101 show a potential association with an increased risk of youth-onset T2DM in the Bangladeshi population, this association appears to be BMI-dependent.

PMID:40551902 | PMC:PMC12182985 | DOI:10.7759/cureus.84696

Medicine (Baltimore). 2025 Jun 20;104(25):e42858. doi: 10.1097/MD.0000000000042858.

ABSTRACT

Systemic lupus erythematosus is an autoimmune disease with diverse clinical manifestations, including lupus nephritis. Calcineurin inhibitors (CNIs) are a treatment option, but traditional CNIs have limitations. Voclosporin, a novel oral CNI, inhibits calcineurin to modulate T-cell activation and stabilize podocytes in lupus nephritis. This review assesses voclosporin's therapeutic potential in treating SLE (lupus nephritis), examining its mechanism of action, clinical efficacy, safety profile, and advantages over other CNIs. A broad search was conducted to identify studies published from 2009 to 2024 on voclosporin and other CNIs in lupus nephritis, using databases such as PUBMED, SCOPUS, Google Scholar and Cochrane Library. MeSH Keywords included "voclosporin," "lupus nephritis," "systemic lupus erythematosus," and "calcineurin inhibitors." Studies were included if they reported relevant clinical outcomes, evaluated voclosporin in lupus nephritis, or provided comparative data on voclosporin versus other CNIs, focusing on randomized controlled trials, systematic reviews, meta-analyses, retrospective studies and cohort studies. Voclosporin demonstrated higher renal response rates at 52 weeks than standard treatment alone (40.8% vs 22.5%). It has stable pharmacokinetics, reducing the need for individualized dose adjustments and frequent monitoring. Safety outcomes show a lower incidence of adverse effects like hypertension and hyperlipidemia compared to traditional CNIs. Voclosporin offers superior efficacy and safety compared to traditional CNIs for managing lupus nephritis, with predictable dosing and a favorable side effect profile. Continued research is needed to optimize voclosporin's use and support personalized medicine approaches.

PMID:40550096 | DOI:10.1097/MD.0000000000042858

iScience. 2025 Jun 6;28(6):112784. doi: 10.1016/j.isci.2025.112784. eCollection 2025 Jun 20.

ABSTRACT

[This corrects the article DOI: 10.1016/j.isci.2025.112353.].

PMID:40546946 | PMC:PMC12179605 | DOI:10.1016/j.isci.2025.112784

Oncologist. 2025 Jun 4;30(6):oyaf127. doi: 10.1093/oncolo/oyaf127.

ABSTRACT

BACKGROUND: Since 2021, glioblastomas have been classified into two subgroups: classic glioblastomas (histGB), defined as IDH wild-type grade 4 astrocytomas with necrosis and vascular proliferation, showing contrast enhancement (CE) on MRI; and molecular glioblastomas (molGB), characterized by specific alterations (7+/10-, EGFR amplification, TERT mutation). Although not always the case, molGB often lack CE and may mimic low-grade gliomas (LGG), hence complicating the diagnosis. Survival outcomes remain debated. This study aimed to evaluate the response of molGB to standard treatment and assess the ability of machine learning and deep learning to differentiate molGB without CE from LGG on MRI.

METHODS: We retrospectively studied 132 glioblastoma patients treated with radiotherapy and temozolomide, comparing the survival outcomes of histGB and molGB. Artificial intelligence (AI) models were trained using features from MRI FLAIR hypersignal segmentation to distinguish molGB without CE from LGG.

RESULTS: No significant difference in median overall survival (OS) (20.6 vs 18.4 months, P = .2) or progression-free survival (10.1 vs 9.3 months, P = .183) was observed between molGB and histGB. However, molGB without CE demonstrated improved median OS (31.2 vs 18 months, hazard ratios 0.45). Artificial intelligence models distinguished molGB without CE from LGG, achieving a best-performing ROC AUC of 0.85.

CONCLUSIONS: While patients with molGB and histGB have similar overall survival, patients with molGB without CE appear to have better outcomes. Artificial intelligence models effectively differentiate molGB from LGG, supporting their potential diagnostic utility.

PMID:40542584 | DOI:10.1093/oncolo/oyaf127

Nat Chem Biol. 2025 Jun 19. doi: 10.1038/s41589-025-01944-x. Online ahead of print.

ABSTRACT

Click chemistry is a powerful concept that refers to a set of covalent bond-forming reactions with highly favorable properties. In this Perspective, I outline the analogous concept of click biology as a set of reactions derived from the regular building blocks of living cells, rapidly forming covalent bonds to specific partners under cell-friendly conditions. Click biology using protein components employs canonical amino acids and may react close to the diffusion limit, with selectivity in living cells amid thousands of components generated from the same building blocks. I discuss how the criteria for click chemistry can be applied or modified to fit the extra constraints of click biology and achieve favorable characteristics for biological research. Existing reactions that may be described as click biology include split intein reconstitution, spontaneous isopeptide bond formation by SpyTag and SpyCatcher and suicide enzyme reaction with small-molecule ligands (HaloTag and SNAP-tag). I also describe how click biology has created new possibilities in fields including molecular imaging, mechanobiology, vaccines and engineering cellular intelligence.

PMID:40537536 | DOI:10.1038/s41589-025-01944-x

Spine Deform. 2025 Jun 19. doi: 10.1007/s43390-025-01122-y. Online ahead of print.

ABSTRACT

PURPOSE: The surgical management of severe spinal deformities (> 80°) in children and young adults remains challenging despite technical advances. Large, stiff curves with a short radius present a high complication rate, also in terms of acute or chronic spinal cord injury. There is a lack of consensus regarding the perioperative management of these curves, also due to the limited evidence available. This study aimed to review the strategies used in different hospitals for the surgical management of severe spinal deformities in young subjects and, in particular, to analyze the complication rate in a large series of patients.

METHODS: Multicentric, retrospective study on surgically treated patients younger than 25 with a coronal or sagittal deformity measuring at least 80°. Perioperative and radiographic data were collected. The rate of mechanical, neurologic, infectious, and other complications was analyzed, along with possible risk factors.

RESULTS: Data from 161 patients were analyzed. After surgery, there was a significant improvement of the deformity both on the coronal and sagittal plane. The overall complication rate was 25%. An increased angular ratio in the first postoperative X-ray was the main risk factor for the development of complications, along with a higher correction of the spinosacral angle. Performing a 3CO was associated with a higher risk of infections but not with a higher risk of mechanical complications. The use of an anterior approach did not increase the complication rate. A higher rate of non-mechanical and non-infectious complications was observed in patients who did not wear a brace or did not undergo halo traction before surgery.

CONCLUSION: While spinal fusion is an effective treatment for the management of severe deformities, the rate of complications is still high.

PMID:40536621 | DOI:10.1007/s43390-025-01122-y

J Am Chem Soc. 2025 Jun 18. doi: 10.1021/jacs.5c08017. Online ahead of print.

ABSTRACT

The electrochemical reduction of CO2 in acidic media has received considerable attention, as it can mitigate the carbonate formation issue. However, the available protons in acidic media can boost the hydrogen evolution reaction (HER), so alkali metal cations are generally employed to promote the CO2 reduction reaction (CO2RR) while suppressing the HER. Here we report that NH4+ cations are more effective promoters of CO2 electrolysis on Au in acidic media than Na+ and K+, achieving a 3-fold improvement in the CO2RR activity. The promotional effect of NH4+ cations is mainly attributed to their enhanced electrostatic stabilization of the CO2 adsorption, which is the rate-limiting step for the CO2RR on Au. An estimation of the local pH under relevant conditions indicates that NH4+ can mitigate the interfacial pH swing during the CO2RR compared to Na+ and K+. Further studies on Au nanocatalysts of varying sizes indicate that the cation effect is independent of the catalytic sites. Our work advances the understanding of the cation effect on CO2RR using nonmetal cations and demonstrates a viable cation strategy to enhance CO2 electrolysis.

PMID:40530857 | DOI:10.1021/jacs.5c08017

Front Comput Neurosci. 2025 Jun 3;19:1570979. doi: 10.3389/fncom.2025.1570979. eCollection 2025.

ABSTRACT

[This corrects the article DOI: 10.3389/fncom.2024.1418280.].

PMID:40529250 | PMC:PMC12170620 | DOI:10.3389/fncom.2025.1570979

Nucleic Acids Res. 2025 Jun 6;53(11):gkaf496. doi: 10.1093/nar/gkaf496.

ABSTRACT

HiBiT is an engineered luciferase's 11-amino-acid component that can be introduced as a tag at either terminus of a protein of interest. When the LgBiT component and a substrate are present, HiBiT and LgBiT dimerize forming a functional luciferase. The HiBiT technology has been extensively used for high-throughput protein turnover studies in cells. Here, we have adapted the use of the HiBiT technology to quantify messenger RNA (mRNA) translation temporally in vitro in the rabbit reticulocyte system and in cellulo in HEK293 cells constitutively expressing LgBiT. The assay system can uniquely detect differences in cap, 5'UTR, modified nucleotide composition, coding sequence optimization and poly(A) length, and their effects on mRNA translation over time. Importantly, using these assays we established the optimal mRNA composition varied depending on the encoded protein of interest, highlighting the importance of screening methods tailored to the protein of interest, and not reliant on reporter proteins. Our findings demonstrated that HiBiT can be easily and readily adapted to monitor real-time mRNA translation in live cells and offers a novel and highly favourable method for the development of mRNA-based therapeutics.

PMID:40521662 | DOI:10.1093/nar/gkaf496

Foods. 2025 May 28;14(11):1906. doi: 10.3390/foods14111906.

ABSTRACT

As the global demand for sustainable protein sources grows, valorizing side streams in plant protein processing has become crucial. This study revisits the conventional alkaline-isoelectric extraction of faba bean protein isolates, introducing an enhanced mass balance-driven approach to recover underutilized protein fractions from typically discarded side streams. Through strategic pH manipulation and centrifugation, four distinct protein fractions were recovered with purities ranging from 34.6% to 89.6%, collectively recapturing a significant portion of the 16% protein loss in standard processing. SDS-PAGE and FTIR analyses confirmed the structural diversity among the recovered fractions, with albumin-rich and globulin-rich profiles exhibiting unique spectral and electrophoretic signatures. Functionally, fractions B and D exhibited superior water- and oil-holding capacities, indicating their potential utility in food formulations requiring enhanced moisture and lipid retention. In contrast, fraction C, characterized by low water-holding capacity and high solubility, may be better suited to applications prioritizing emulsification performance, such as in dairy or meat analogs. This study not only highlights the feasibility of reclaiming high-quality protein from industrial byproducts but also underscores the potential of these recovered proteins in diverse food and non-food sectors, including pharmaceuticals and cosmetics. These findings contribute to circular economy strategies by transforming waste into value-added ingredients with functional and commercial significance.

PMID:40509434 | DOI:10.3390/foods14111906

Br J Anaesth. 2025 Jun 5:S0007-0912(25)00294-6. doi: 10.1016/j.bja.2025.04.044. Online ahead of print.

NO ABSTRACT

PMID:40480913 | DOI:10.1016/j.bja.2025.04.044

Emerg Microbes Infect. 2025 Jun 6:2511132. doi: 10.1080/22221751.2025.2511132. Online ahead of print.

ABSTRACT

The increasing spread of highly pathogenic avian influenza (HPAI) A/H5 viruses poses a pandemic threat. Circulating clade 2.3.4.4b viruses have demonstrated rapid transcontinental dissemination, extensive reassortment, epizootic spread and potential sustained mammal-to-mammal transmission, signifying a heightened risk of becoming a human pathogen of high consequence. A broadly protective, future-proof vaccine against multiple clades of H5 influenza is urgently needed for pandemic preparedness. Here, we combine two novel vaccine technologies to generate a Digitally Immune Optimised and Selected H5 antigen (DIOSvax-H5inter) displayed multivalently on the mi3 nanocage using the SpyTag003/SpyCatcher003 conjugation system. Mice immunised with DIOSvax-H5inter Homotypic Nanocages at low doses demonstrate potent, cross-clade neutralising antibody and T cell responses against diverse H5 strains. DIOSvax-H5inter Homotypic Nanocages provide a scalable vaccine candidate with the potential for pan-H5 protection against drifted or newly emergent H5 strains. This World Health Organization preferred characteristic is essential for prospective strategic stockpiling in the pre-pandemic phase.Trial registration: ClinicalTrials.gov identifier: NCT06145178..

PMID:40476519 | DOI:10.1080/22221751.2025.2511132

Open Biol. 2025 Jun;15(6):250052. doi: 10.1098/rsob.250052. Epub 2025 Jun 4.

ABSTRACT

Naked mole-rats (NMRs, Heterocephalus glaber) are highly unusual rodents exhibiting remarkable adaptations to their subterranean habitat and resistance to developing various age-related diseases such as those related to abnormal cell proliferation or cancer, neurodegeneration and inflammation. In other rodents, as well as humans, a ubiquitous Ca2+ influx pathway, namely the store-operated Ca2+ entry (SOCE), has been implicated in all these diseases. SOCE is triggered by intracellular Ca2+ store depletion resulting in interaction of Stim proteins with Orai proteins, the putative homologues of which appear to be present in the NMR genome, but no functional characterization of SOCE in NMRs has yet been conducted. In this study, we provide the first functional and pharmacological characterization of SOCE in NMR using both excitable and non-excitable cells.

PMID:40460874 | DOI:10.1098/rsob.250052

J Clin Med. 2025 May 16;14(10):3495. doi: 10.3390/jcm14103495.

ABSTRACT

Introduction: Sepsis leads to substantial global health burdens in terms of morbidity and mortality and is associated with numerous risk factors. It is crucial to identify sepsis at an early stage in order to limit its escalation and sequelae associated with the condition. The purpose of this research is to predict ICU mortality early and evaluate the predictive accuracy of machine learning algorithms for ICU mortality among septic patients. Methods: The study used a retrospective cohort from computerized ICU records accumulated from 280 hospitals between 2014 and 2015. Initially the sample size was 23.47K. Several machine learning models were trained, validated, and tested using five-fold cross-validation, and three sampling strategies (Under-Sampling, Over-Sampling, and Combination). Results: The under-sampled approach combined with augmentation for the Extra Trees model produced the best performance with Accuracy, Precision, Sensitivity, Specificity, F1-Score, and AUC of 90.99%, 84.16%, 94.89%, 88.48%, 89.20%, and 91.69%, respectively, with Top 30 features. For Over-Sampling, the Top 29 combined features showed the best performance with Accuracy, Precision, Sensitivity, Specificity, F1-Score, and AUC of 82.99%, 51.38%, 71.72%, 85.41%, 59.87%, and 78.56%, respectively. For Down-Sampling, the Top 31 combined features produced Accuracy, Precision, Sensitivity, Specificity, F1-Score, and AUC of 81.78%, 49.08%, 79.76%, 82.21%, 60.76%, and 80.98%, respectively. Conclusions: Machine learning models can reliably predict ICU mortality when suitable clinical predictors are utilized. The study showed that the proposed Extra Trees model can predict ICU mortality with an accuracy of 90.99% accuracy using only single-entry data. Incorporating longitudinal data could further enhance model performance.

PMID:40429489 | DOI:10.3390/jcm14103495