Abdom Radiol (NY). 2025 Apr 11. doi: 10.1007/s00261-025-04929-5. Online ahead of print.

ABSTRACT

The exploration of deep learning techniques for predicting various biological characteristics of endometrial cancer (EC) is of significant importance. The objective of this study was to develop an optimized radiomics scheme combining multiparametric magnetic resonance imaging (MRI), deep learning, and machine learning to predict biological features including myometrial invasion (MI), lymph-vascular space invasion (LVSI), histologic grade (HG), and estrogen receptor (ER). This retrospective study involved 201 EC patients, who were divided into four groups according to the specific tasks. The proposed radiomics scheme extracted quantitative imaging features and multidimensional deep learning features from multiparametric MRI. Several classifiers were employed to predict biological features. Model performance and interpretability were assessed using traditional classification metrics, Gradient-weighted Class Activation Mapping (Grad-CAM), and SHapley Additive exPlanation (SHAP) techniques. In the deep MI (DMI) prediction task, the proposed protocol achieved an area under the curve (AUC) value of 0.960 (95% CI 0.9005-1.0000) in the test cohort. In the LVSI prediction task, the AUC of the proposed scheme in the test cohort was 0.924 (95% CI 0.7760-1.0000). In the HG prediction task, the AUC value of the proposed scheme in the test cohort was 0.937 (95% CI 0.8561-1.0000). In the ER prediction task, the AUC value of the proposed scheme in the test cohort was 0.929 (95% CI 0.7991-1.0000). The proposed radiomics scheme outperformed the comparative scheme and effectively extracted imaging features related to the expression of EC biological characteristics, providing potential clinical significance for accurate diagnosis and treatment decision-making.

PMID:40214699 | DOI:10.1007/s00261-025-04929-5

Minerva Anestesiol. 2025 Apr 11. doi: 10.23736/S0375-9393.25.18746-4. Online ahead of print.

ABSTRACT

The integration of artificial intelligence (AI) and all new technologies (NTs) into enhanced recovery after surgery (ERAS) protocols offers significant opportunities to address implementation challenges and improve patient care. Despite the proven benefits of ERAS, limitations such as resistance to change, resource constraints, and poor interdepartmental communication persist. AI can play a crucial role in overcoming ERAS implementation barriers by simplifying clinical plans, ensuring high compliance, and creating patient-centered approaches. Advanced techniques like machine learning and deep learning can optimize preoperative management, intraoperative phases, and postoperative recovery pathways. AI integration in ERAS protocols has the potential to revolutionize perioperative medicine by enabling personalized patient care, enhancing monitoring strategies, and improving clinical decision-making. The technology can address common postoperative challenges by developing individualized ERAS plans based on patient risk factors and optimizing perioperative processes. While challenges remain, including the need for external validation and data security, the authors suggest that the combination of AI, NTs, and ERAS protocols should become an integral part of routine clinical practice. This integration ultimately leads to improved patient outcomes and satisfaction in surgical care, transforming the perioperative medicine landscape by tailoring pathways to patients' needs.

PMID:40214219 | DOI:10.23736/S0375-9393.25.18746-4

J Med Eng Technol. 2025 Apr 11:1-8. doi: 10.1080/03091902.2025.2488827. Online ahead of print.

ABSTRACT

Medical image fusion reduces the time required for medical diagnosis by creating a composite image from a set of images belonging to different modalities. This paper introduces a deep learning framework for medical image fusion, optimising the number of convolutional layers and selecting an appropriate activation function. The conducted experiments demonstrate that employing three convolution layers with a swish activation function for the intermediate layers is sufficient to extract the salient features of the input images. The tuned features are fused using element-wise fusion rules to prevent the loss of minute details crucial for medical images. The comprehensive fused image is then reconstructed from these features using another set of three convolutional layers. Experimental results confirm that the proposed methodology outperforms other conventional medical image fusion methods in terms of various metrics and the quality of the fused image.

PMID:40214199 | DOI:10.1080/03091902.2025.2488827

MAbs. 2025 Dec;17(1):2490788. doi: 10.1080/19420862.2025.2490788. Epub 2025 Apr 11.

ABSTRACT

The shift toward subcutaneous administration for biologic therapeutics has gained momentum due to its patient-friendly nature, convenience, reduced healthcare burden, and improved compliance compared to traditional intravenous infusions. However, a significant challenge associated with this transition is managing the viscosity of the administered solutions. High viscosity poses substantial development and manufacturability challenges, directly affecting patients by increasing injection time and causing pain at the injection site. Furthermore, high viscosity formulations can prolong residence time at the injection site, affecting absorption kinetics and potentially altering the intended pharmacological profile and therapeutic efficacy of the biologic candidate. Here, we report the application of a multimodal feature learning workflow for predicting the viscosity of antibodies in therapeutics discovery. It integrates multiple data sources including sequence, structural, physicochemical properties, as well as embeddings from a language model. This approach enables the model to learn from various underlying rules, such as physicochemical rules from molecular simulations and protein evolutionary patterns captured by large, pre-trained deep learning models. By comparing the effectiveness of this approach to other selected published viscosity prediction methods, this study provides insights into their intrinsic viscosity predictive potential and usability in early-stage therapeutics antibody development pipelines.

PMID:40214197 | DOI:10.1080/19420862.2025.2490788

Cells. 2025 Apr 3;14(7):536. doi: 10.3390/cells14070536.

ABSTRACT

Fibrosing disorders including idiopathic pulmonary fibrosis (IPF) are progressive irreversible diseases, often with poor prognoses, characterized by the accumulation of excessive scar tissue and extracellular matrix. Translational regulation has emerged as a critical aspect of gene expression control, and the dysregulation of key effectors is associated with disease pathogenesis. This review examines the current literature on translational regulators in IPF, focusing on microRNAs (miRNAs), long non-coding RNAs (lncRNAs), and RNA transcript modifications including alternative polyadenylation and chemical modification. Some of these translational regulators potentiate fibrosis, and some of the regulators inhibit fibrosis. In IPF, some of the profibrotic regulators are upregulated, and some of the antifibrotic regulators are downregulated. Correcting these defects in IPF-associated translational regulators could be an intriguing avenue for therapeutics.

PMID:40214489 | DOI:10.3390/cells14070536

Cells. 2025 Mar 26;14(7):497. doi: 10.3390/cells14070497.

ABSTRACT

This paper investigates the ability of our selective small molecule TG2 inhibitor 1-155 in reducing fibrosis in a bleomycin-induced pulmonary fibrosis mouse model. Formulated as a fine stable suspension, 1-155 was delivered intranasally (IN) at 3 mg/kg via IN delivery once daily. It significantly inhibited collagen deposition in the lungs in the bleomycin-challenged mice. Compared to its vehicle control treatment, a significant reduction in a key myofibroblast marker α smooth muscle actin and TG2 was also detected in the 1-155-treated animals. Most importantly, 1-155 treatment significantly improved several key lung function parameters, such as cord compliance, vital capacity, and dynamic compliance, which are comparable to that found for the positive control nintedanib at a much higher dosage of 60 mg/kg twice daily via oral delivery. The 1-155-treated mice showed a trend in improvement of average body weight. For the first time, our study demonstrates the effectiveness of a selective small molecule TG2 inhibitor in reducing pulmonary fibrosis in a pre-clinical model. Importantly, we were able to correlate this effect of 1-155 with the improvement of animal lung function showing the potential of the use of TG2 inhibitors as a therapeutic treatment for fibrotic lung conditions like IPF.

PMID:40214451 | DOI:10.3390/cells14070497

Plant Biotechnol J. 2025 Apr 11. doi: 10.1111/pbi.70054. Online ahead of print.

NO ABSTRACT

PMID:40214971 | DOI:10.1111/pbi.70054

Cells. 2025 Apr 4;14(7):548. doi: 10.3390/cells14070548.

ABSTRACT

Extracellular vesicles (EVs) have been explored as promising vehicles for drug delivery. One of the most valuable features of EVs is their ability to cross physiological barriers, particularly the blood-brain barrier (BBB). This significantly enhances the development of EV-based drug delivery systems for the treatment of CNS disorders. The present review focuses on the factors and techniques that contribute to the successful delivery of EV-based therapeutics to the brain. Here, we discuss the major methods of brain targeting which includes the utilization of different administration routes, capitalizing on the biological origins of EVs, and the modification of EVs through the addition of specific ligands on to the surface of EVs. Finally, we discuss the current challenges in large-scale EV production and drug loading while highlighting future perspectives regarding the application of EV-based therapeutics for brain delivery.

PMID:40214500 | DOI:10.3390/cells14070548

Clin Pharmacol Ther. 2025 Apr 11. doi: 10.1002/cpt.3674. Online ahead of print.

ABSTRACT

Patritumab deruxtecan (HER3-DXd, also known as MK-1022), an antibody-drug conjugate consisting of a human epidermal growth factor receptor 3 (HER3) antibody attached to a topoisomerase I inhibitor payload (DXd), has demonstrated efficacy in patients with metastatic breast cancer and non-small cell lung cancer (NSCLC). Exposure-efficacy was assessed in 446 patients with EGFR-mutated NSCLC; exposure-safety was assessed in 715 patients with NSCLC, breast cancer, or colorectal cancer. A range of HER3-DXd dosing regimens was evaluated, including fixed-dosing regimens (1.6-8 mg/kg every 3 weeks [Q3W]; 3.2-6.4 mg/kg Q3W), an up-titration dosing regimen, and an alternative Q2W/Q3W dosing regimen. Logistic regression or time-to-event models were used to test the relationships of each endpoint with pharmacokinetic analytes (anti-HER3-ac-DXd and DXd). Anti-HER3-ac-DXd exposure was positively associated with objective response rate, and bone metastasis was identified as a significant covariate. DXd exposure showed a stronger correlation with most safety endpoints compared with anti-HER3-ac-DXd exposure, except for grade ≥ 2 nausea/vomiting and any grade adjudicated drug-related interstitial lung disease (ILD). Dose response predictions verified a manageable safety profile for the 5.6 mg/kg Q3W regimen. This observation was supported by low predicted rates of adjudicated drug-related ILD and adverse events leading to treatment discontinuation with the 5.6 mg/kg Q3W regimen. Overall, these results support the selection of HER3-DXd 5.6 mg/kg Q3W as the recommended dosing regimen for patients with NSCLC, and these data inform the optimal dosing regimen for other tumor types.

PMID:40214010 | DOI:10.1002/cpt.3674

Front Plant Sci. 2025 Mar 27;16:1540722. doi: 10.3389/fpls.2025.1540722. eCollection 2025.

ABSTRACT

In this study, a target spraying decision and hysteresis algorithm is designed in conjunction with deep learning, which is deployed on a testbed for validation. The overall scheme of the target spraying control system is first proposed. Then YOLOv5s is lightweighted and improved. Based on this, a target spraying decision and hysteresis algorithm is designed, so that the target spraying system can precisely control the solenoid valve and differentiate spraying according to the distribution of weeds in different areas, and at the same time, successfully solve the operation hysteresis problem between the hardware. Finally, the algorithm was deployed on a testbed and simulated weeds and simulated tillering wheat were selected for bench experiments. Experiments on a dataset of realistic scenarios show that the improved model reduces the GFLOPs (computational complexity) and size by 52.2% and 42.4%, respectively, with mAP and F1 of 91.4% and 85.3%, which is an improvement of 0.2% and 0.8%, respectively, compared to the original model. The results of bench experiments showed that the spraying rate under the speed intervals of 0.3-0.4m/s, 0.4-0.5m/s and 0.5-0.6m/s reached 99.8%, 98.2% and 95.7%, respectively. Therefore, the algorithm can provide excellent spraying accuracy performance for the target spraying system, thus laying a theoretical foundation for the practical application of target spraying.

PMID:40212873 | PMC:PMC11983634 | DOI:10.3389/fpls.2025.1540722

Ann Med Surg (Lond). 2025 Feb 28;87(3):1506-1528. doi: 10.1097/MS9.0000000000003043. eCollection 2025 Mar.

ABSTRACT

Cancer is one of the leading causes of death worldwide. In cancer therapy, anti-cancer drugs are the current treatment-of-choice for patients with metastatic cancers, but these drugs present a major drawback: they destroy healthy cells along with cancerous cells. Unfortunately, the drug discovery process for de novo drugs is costly and time-consuming. To address this global problem, our research team has established the concept of "Chemotherapy with a molecular rational basis", which focuses on the identification of molecular targets in tumor cells, whose activation or inhibition induces apoptosis or sensitizes the tumor cells to apoptosis. Here we review the experimental and clinical evidence of pentoxifylline (PTX) in the setting of chemotherapy with a molecular rational basis. A search of the literature was conducted for articles published during the period from 2 January 2003 to 21 October 2024. Articles published in English or Spanish were included. The keywords "Pentoxifylline" OR "BL 191" OR "trental" AND "cancer" were used for in vitro, in vivo, and clinical studies. PTX is an approved, accessible, and relatively safe drug. Furthermore there is a large body of experimental and clinical evidence of the beneficial effects of PTX in cancer therapy, either alone or in combination with antitumor drugs, sometimes even more effective than traditional chemotherapy regimens. However, it is necessary to carry out larger clinical trials in cancer patients to identify the benefits, adverse effects and even pharmacological interactions of PTX with current chemotherapy regimens and thus achieve a new drug repositioning that benefits our patients.

PMID:40213176 | PMC:PMC11981314 | DOI:10.1097/MS9.0000000000003043

Eur J Med Res. 2025 Apr 11;30(1):267. doi: 10.1186/s40001-025-02443-4.

ABSTRACT

Current TB treatment regimens are hindered by drug resistance, numerous adverse effects, and long treatment durations, highlighting the need for 'me-better' treatment regimens. Host-directed therapy (HDT) has gained recognition as a promising approach in TB treatment. It allows the repurposing of existing drugs approved for other conditions and aims to enhance the effectiveness of existing anti-TB therapies, minimize drug resistance, decrease treatment duration, and adverse effects. By modulating the host immune response, HDT ameliorates immunopathological damage and improves overall outcomes by promoting autophagy, antimicrobial peptide production, and other mechanisms. It holds promise for addressing the challenges posed by multiple and extensively drug-resistant Mycobacterium tuberculosis strains, which are increasingly difficult to treat using conventional therapies. This article reviews various HDT candidates, including repurposed drugs, explores their underlying mechanisms such as autophagy promotion and inflammation reduction, while emphasizing their potential to improve TB treatment outcomes and outlining future research directions.

PMID:40211397 | DOI:10.1186/s40001-025-02443-4

Cell Commun Signal. 2025 Apr 10;23(1):179. doi: 10.1186/s12964-025-02185-0.

ABSTRACT

BACKGROUND: About 40% of relapsed or non-responder tumors exhibit therapeutic resistance in the absence of a clear genetic cause, suggesting a pivotal role of intracellular communication. A deeper understanding of signaling pathways rewiring occurring in resistant cells is crucial to propose alternative effective strategies for cancer patients.

METHODS: To achieve this goal, we developed a novel multi-step strategy, which integrates high sensitive mass spectrometry-based phosphoproteomics with network-based analysis. This strategy builds context-specific networks recapitulating the signaling rewiring upon drug treatment in therapy-resistant and sensitive cells.

RESULTS: We applied this strategy to elucidate the BCR::ABL1-independent mechanisms that drive relapse upon therapy discontinuation in chronic myeloid leukemia (CML) patients. We built a signaling map, detailing - from receptor to key phenotypes - the molecular mechanisms implicated in the control of proliferation, DNA damage response and inflammation of therapy-resistant cells. In-depth analysis of this map uncovered novel therapeutic vulnerabilities. Functional validation in patient-derived leukemic stem cells revealed a crucial role of acquired FLT3-dependency and its underlying molecular mechanism.

CONCLUSIONS: In conclusion, our study presents a novel generally applicable strategy and the reposition of FLT3, one of the most frequently mutated drivers of acute leukemia, as a potential therapeutic target for CML relapsed patients.

PMID:40211380 | DOI:10.1186/s12964-025-02185-0

Front Psychiatry. 2025 Mar 27;16:1542000. doi: 10.3389/fpsyt.2025.1542000. eCollection 2025.

ABSTRACT

BACKGROUND: Although several guidelines provide dosing recommendations for antidepressants based on patients' genetic information, pharmacogenetic testing for antidepressant use is rarely routinely performed in Japan. To clarify the clinical impact of pharmacogenetic testing, this study estimated the potential drug-gene interactions for first-time antidepressant treatment in Japanese patients with major depressive disorder.

METHODS: This study retrospectively included Japanese patients who were registered for depressive episodes (F32, International Classification of Diseases, Tenth Revision) and initiated on antidepressants between July 2022 and March 2023. Antidepressant prescription rates were calculated using a nationwide hospital-based database (Medical Data Vision Co., Ltd). The incidence of actionable drug-gene interactions was estimated by multiplying the first-time prescription rate of each relevant antidepressant by the frequency of its corresponding actionable phenotype.

RESULTS: A total of 3,197 patients were included in the analysis. Escitalopram was the most frequently prescribed antidepressant (18.7%, n = 597), followed by mirtazapine (17.5%, n = 561), and sertraline (15.4%, n = 493). Of the patients receiving their first treatment of major depressive disorder, 56.5% (n = 1,807) were prescribed a drug with actionable pharmacogenetic implications, and 26.4% (n = 844) were estimated to have required actionable therapeutic recommendations. The highest incidence of actionable drug-gene interactions was observed in escitalopram and CYP2C19 pairs (12.4%, n = 398). For sertraline and CYP2C19 or CYP2B6 pairs, the incidence was 11.0% (n = 352). Among all antidepressants, paroxetine had the highest incidence of actionable drug-gene interactions related to CYP2D6 at 1.8% (n = 56); this interaction was rarely observed with other antidepressants (<1%).

CONCLUSIONS: We estimated that one in four Japanese patients with major depressive disorder who were prescribed first-time antidepressants had actionable drug-gene interactions. These results suggest that pre-emptive pharmacogenetic testing in the treatment of major depressive disorder could have important clinical implications.

PMID:40212835 | PMC:PMC11983551 | DOI:10.3389/fpsyt.2025.1542000

Pharmacogenomics. 2025 Apr 11:1-13. doi: 10.1080/14622416.2025.2490461. Online ahead of print.

ABSTRACT

As an emerging health technology, pharmacogenetic (PGx) testing has the capacity to improve medication therapy. However, implementation in medically underserved populations (MUPs) remains limited, which has the potential to increase healthcare disparities. While there is no single accepted definition for MUPs, demographic, socioeconomic, cultural, and geographic factors can lead to reduced access to healthcare, which contributes to disparate health outcomes in these populations. In the case of PGx testing, as MUPs have an increased risk of adverse drug events, have lower numbers of healthcare encounters, and are prescribed more medications which can be guided by PGx testing, additional benefits from PGx testing may occur in MUPs. Study of the acceptability and perceptions of PGx testing in MUPs, as reported in literature, provides support for the development of successful PGx testing implementations. Additionally, a few limited pilot PGx testing implementations in MUPs have assessed feasibility. However, further studies establishing the feasibility and effectiveness of PGx testing implementations in MUPs will enable more widespread PGx testing in those who are medically underserved. Thus, this narrative review explores the impact of medical underservice on health, PGx testing's potential impact on MUPs, and the research and early clinical implementations of PGx in MUPs.

PMID:40211878 | DOI:10.1080/14622416.2025.2490461

Ann Med Surg (Lond). 2025 Mar 27;87(4):2398-2401. doi: 10.1097/MS9.0000000000003087. eCollection 2025 Apr.

ABSTRACT

INTRODUCTION: Allergic bronchopulmonary mycoses, primarily from Aspergillus fumigatus, complicate asthma and cystic fibrosis, presenting diagnostic challenges due to overlapping respiratory symptoms.

CASE PRESENTATION: A 48-year-old male with asthma and a history of Guillain-Barré syndrome presented with cough, chest pain, dyspnea, and weight loss. He was diagnosed with allergic bronchopulmonary aspergillosis after a series of investigations, including CT scans and bronchoscopy.

CLINICAL DISCUSSION: Allergic bronchopulmonary aspergillosis (ABPA) is a rare lung disease caused by an immune reaction to Aspergillus fungi in individuals with pre-existing respiratory conditions like asthma or cystic fibrosis. The primary treatment for ABPA involves systemic corticosteroids, often combined with antifungal agents, to reduce the need for long-term high-dose steroid therapy.

CONCLUSION: This case highlights the need for accurate and early diagnosis of ABPA, especially in patients with asthma or other respiratory diseases, which helps prevent potential complications. Additionally, the case provides valuable insights into how to manage patients with ABPA, contributing to the improvement of protocols and medical care in the future.

PMID:40212161 | PMC:PMC11981463 | DOI:10.1097/MS9.0000000000003087

Small Sci. 2023 Jun 11;3(8):2300031. doi: 10.1002/smsc.202300031. eCollection 2023 Aug.

ABSTRACT

Recent advancements in transmission electron microscopy (TEM) have enabled the study of atomic structures of materials at unprecedented scales as small as tens of picometers (pm). However, accurately detecting atomic positions from TEM images remains a challenging task. Traditional Gaussian fitting and peak-finding algorithms are effective under ideal conditions but perform poorly on images with strong background noise or contamination areas (shown as ultrabright or ultradark contrasts). Moreover, these traditional algorithms require parameter tuning for different magnifications. To overcome these challenges, AtomID-Net is presented, a deep neural network model for atomic detection from multiscale low-SNR experimental images of scanning TEM (scanning transmission electron microscopy (STEM)). The model is trained on real images, which allows the robust and efficient detection of atomic positions, even in the presence of background noise and contamination. The evaluation on a test set of 50 images with a resolution of 800 × 800 yields an average F1-Score of 0.964, which demonstrates significant improvements over existing peak-finding algorithms.

PMID:40213610 | PMC:PMC11935788 | DOI:10.1002/smsc.202300031

Small Sci. 2024 Mar 19;4(5):2300185. doi: 10.1002/smsc.202300185. eCollection 2024 May.

ABSTRACT

Herein, state-of-the-art multiscale modeling methods have been described. This research includes notable molecular, micro-, meso-, and macroscale models for hard (polymer, metal, yarn, fiber, fiber-reinforced polymer, and polymer matrix composites) and soft (biological tissues such as brain white matter [BWM]) composite materials. These numerical models vary from molecular dynamics simulations to finite-element (FE) analyses and machine learning/deep learning surrogate models. Constitutive material models are summarized, such as viscoelastic hyperelastic, and emerging models like fractional viscoelastic. Key challenges such as meshing, data variability and material nonlinearity-driven uncertainty, limitations in terms of computational resources availability, model fidelity, and repeatability are outlined with state-of-the-art models. Latest advancements in FE modeling involving meshless methods, hybrid ML and FE models, and nonlinear constitutive material (linear and nonlinear) models aim to provide readers with a clear outlook on futuristic trends in composite multiscale modeling research and development. The data-driven models presented here are of varying length and time scales, developed using advanced mathematical, numerical, and huge volumes of experimental results as data for digital models. An in-depth discussion on data-driven models would provide researchers with the necessary tools to build real-time composite structure monitoring and lifecycle prediction models.

PMID:40213577 | PMC:PMC11935080 | DOI:10.1002/smsc.202300185

Small Sci. 2024 Aug 10;4(11):2400252. doi: 10.1002/smsc.202400252. eCollection 2024 Nov.

ABSTRACT

Developing scalable and accurate predictive analytical methods for the classification of protein-DNA binding is critical for advancing our understanding of molecular biology, disease mechanisms, and a wide spectrum of biotechnological and medical applications. It is discovered that histone-DNA interactions can be stratified based on stain patterns created by the deposition of various nucleoprotein solutions onto a substrate. In this study, a deep-learning neural network is applied to categorize polarized light microscopy images of drying droplet deposits originating from different histone-DNA mixtures. These DNA stain patterns featured high reproducibility across different species and thus enabled comprehensive DNA categorization (100% accuracy) and accurate prediction of their respective binding affinities to histones. Eukaryotic DNA, which has a higher binding affinity to mammalian histones than prokaryotic DNA, is associated with a higher overall prediction accuracy. For a given species, the average prediction accuracy increased with DNA size. To demonstrate generalizability, a pre-trained CNN is challenged with unknown images that originated from DNA samples of species not included in the training set. The CNN classified these unknown histone-DNA samples as either strong or medium binders with 84.4% and 96.25% accuracy, respectively.

PMID:40213456 | PMC:PMC11935254 | DOI:10.1002/smsc.202400252

JAMIA Open. 2025 Apr 10;8(2):ooaf026. doi: 10.1093/jamiaopen/ooaf026. eCollection 2025 Apr.

ABSTRACT

OBJECTIVES: Recent advances in deep learning show significant potential in analyzing continuous monitoring electronic health records (EHR) data for clinical outcome prediction. We aim to develop a Transformer-based, Encounter-level Clinical Outcome (TECO) model to predict mortality in the intensive care unit (ICU) using inpatient EHR data.

MATERIALS AND METHODS: The TECO model was developed using multiple baseline and time-dependent clinical variables from 2579 hospitalized COVID-19 patients to predict ICU mortality and was validated externally in an acute respiratory distress syndrome cohort (n = 2799) and a sepsis cohort (n = 6622) from the Medical Information Mart for Intensive Care IV (MIMIC-IV). Model performance was evaluated based on the area under the receiver operating characteristic (AUC) and compared with Epic Deterioration Index (EDI), random forest (RF), and extreme gradient boosting (XGBoost).

RESULTS: In the COVID-19 development dataset, TECO achieved higher AUC (0.89-0.97) across various time intervals compared to EDI (0.86-0.95), RF (0.87-0.96), and XGBoost (0.88-0.96). In the 2 MIMIC testing datasets (EDI not available), TECO yielded higher AUC (0.65-0.77) than RF (0.59-0.75) and XGBoost (0.59-0.74). In addition, TECO was able to identify clinically interpretable features that were correlated with the outcome.

DISCUSSION: The TECO model outperformed proprietary metrics and conventional machine learning models in predicting ICU mortality among patients with COVID-19, widespread inflammation, respiratory illness, and other organ failures.

CONCLUSION: The TECO model demonstrates a strong capability for predicting ICU mortality using continuous monitoring data. While further validation is needed, TECO has the potential to serve as a powerful early warning tool across various diseases in inpatient settings.

PMID:40213364 | PMC:PMC11984207 | DOI:10.1093/jamiaopen/ooaf026