Individualized threshold of the involuntary weight loss in prognostic assessment of cancer.

BACKGROUND: Involuntary weight loss (WL) is a common symptom in cancer patients and is associated with poor outcomes. However, there is no standardized definition of WL, and it is unclear what magnitude of weight loss should be considered significant for prognostic purposes. This study aimed to determine an individualized threshold for WL that can be used for prognostic assessment in cancer patients. METHODS: Univariate and multivariate analyses of overall survival (OS) were performed using Cox proportional hazard models. The Kaplan-Meier method was performed to estimate the survival distribution of different WL levels. Logistic regression analysis was used to determine the relationship between WL and 90-day outcomes. Restricted cubic splines with three knots were used to examine the effects of WL on survival under different body mass index (BMI) conditions. RESULTS: Among the 8806 enrolled patients with cancer, median survival time declined as WL increased, from 25.1 to 20.1, 17.8 and 16.4 months at <2%, 2-5%, 5-10% and ≥10% WL, respectively (P < 0.001). Multivariate adjusted Cox regression analysis showed that the risk of adverse prognosis increased by 18.1% based on the SD of WL (5.45 U) (HR: 1.181, 95% CI: 1.144-1.219, P < 0.001). Similarly, categorical WL was independently associated with OS in patients with cancer. With the worsening of WL, the risk of a poor prognosis in patients increases stepwise. Compared with <2% WL, all-cause mortalities were 15.1%, 37% and 64.2% higher in 2-5%, 5-10%, and ≥10% WL, respectively. WL can effectively stratify the prognosis of both overall and site-specific cancers. The clinical prognostic thresholds for WL based on different BMI levels were 4.21% (underweight), 5.03% (normal), 6.33% (overweight), and 7.60% (obese). Multivariate logistic regression analysis showed that WL was independently associated with 90-day outcomes in patients with cancer. Compared with patients with <2% WL, those with ≥10% WL had more than twice the risk of 90-day outcomes (OR: 3.277, 95% CI: 2.287-4.694, P < 0.001). Systemic inflammation was a cause of WL deterioration. WL mediates 6.3-10.3% of the overall association between systemic inflammation and poor prognoses in patients with cancer. CONCLUSIONS: An individualized threshold for WL based on baseline BMI can be used for prognostic assessment in cancer patients. WL and BMI should be evaluated simultaneously in treatment decision-making, nutritional intervention, and prognosis discussions of patients with cancer.

Improving the assessment of malnutrition in cancer: Using systemic inflammation markers as a supplement to the inflammation items of the GLIM criteria.

BACKGROUND & AIMS: Systemic inflammation is a key pathogenic criterion for diagnosing malnutrition using the Global Leadership Initiative on Malnutrition (GLIM) criteria. Although cancer is commonly considered as a chronic inflammation-related disease, the inflammatory burden may vary depending on the type and stage of cancer. Therefore, a more precise definition of inflammation criteria could facilitate the identification of malnutrition in cancer. METHODS: This prospective multicenter study included 1683 cancer patients screened via NRS2002 for malnutrition risk. The inflammatory burden index (IBI), C-reactive protein (CRP) level, neutrophil-to-lymphocyte ratio (NLR), and albumin (ALB) level were used to assess the inflammatory burden. Kaplan-Meier and Cox regression analyses were used to determine the relationship between the GLIM criteria and overall survival. Harrell's concordance index (C-index) was used to compare the discriminative performance of the original, IBI-based, CRP-based, NLR-based, and ALB-based GLIM criteria for survival. Logistic regression models were used to assess the association between GLIM criteria and short-term outcomes, length of hospital stay, and hospitalization costs. RESULTS: Compared to the original GLIM criteria, the IBI/CRP/NLR/ALB-based GLIM criteria better predicted the long-term outcomes of patients with cancer (chi-square: 1.316 vs. 78.321 vs. 74.740 vs. 88.719 vs. 100.921). The C-index revealed that the inflammation marker-based GLIM criteria showed significantly better prognostic accuracy than the original GLIM criteria. The ALB-based GLIM criteria exhibited the best prognostic accuracy. The inflammation marker-based GLIM criteria were independent predictive factors for the long-term prognosis of cancer. Patients with malnutrition had a 45% higher risk of adverse long-term prognoses than those without malnutrition. The inflammation marker-based GLIM criteria had good prognostic ability to predict outcomes at 3, 6, and 12 months. The stepwise effect of the grading of severity via the IBI-based GLIM criteria and CRP-based GLIM criteria was notable. The inflammation marker-based GLIM criteria are useful for predicting short-term outcomes, length of hospitalization, and hospitalization costs. CONCLUSION: The inflammation marker-based GLIM criteria have a stronger predictive value than the original GLIM criteria in evaluating both the short- and long-term prognoses of cancer patients. It is recommended to use the inflammation marker-based GLIM criteria for nutritional evaluation of cancer patients.

A Soft-Reference Breast Ultrasound Image Quality Assessment Method That Considers the Local Lesion Area.

The quality of breast ultrasound images has a significant impact on the accuracy of disease diagnosis. Existing image quality assessment (IQA) methods usually use pixel-level feature statistical methods or end-to-end deep learning methods, which focus on the global image quality but ignore the image quality of the lesion region. However, in clinical practice, doctors' evaluation of ultrasound image quality relies more on the local area of the lesion, which determines the diagnostic value of ultrasound images. In this study, a global-local integrated IQA framework for breast ultrasound images was proposed to learn doctors' clinical evaluation standards. In this study, 1285 breast ultrasound images were collected and scored by experienced doctors. After being classified as either images with lesions or images without lesions, they were evaluated using soft-reference IQA or bilinear CNN IQA, respectively. Experiments showed that for ultrasound images with lesions, our proposed soft-reference IQA achieved PLCC 0.8418 with doctors' annotation, while the existing end-to-end deep learning method that did not consider the local lesion features only achieved PLCC 0.6606. Due to the accuracy improvement for the images with lesions, our proposed global-local integrated IQA framework had better performance in the IQA task than the existing end-to-end deep learning method, with PLCC improving from 0.8306 to 0.8851.

A PREDICTION MODEL FOR SEPSIS IN INFECTED PATIENTS: EARLY ASSESSMENT OF SEPSIS ENGAGEMENT.

ABSTRACT: Purpose: To evaluate significant risk variables for sepsis incidence and develop a predictive model for rapid screening and diagnosis of sepsis in patients from the emergency department (ED). Methods: Sepsis-related risk variables were screened based on the PIRO (Predisposition, Insult, Response, Organ dysfunction) system. Training (n = 1,272) and external validation (n = 568) datasets were collected from Peking Union Medical College Hospital (PUMCH) and Beijing Tsinghua Changgung Hospital (BTCH), respectively. Variables were collected at the time of admission. Sepsis incidences were determined within 72 h after ED admissions. A predictive model, Early Assessment of Sepsis Engagement (EASE), was developed, and an EASE-based nomogram was generated for clinical applications. The predictive ability of EASE was evaluated and compared with the National Early Warning Score (NEWS) scoring system. In addition, internal and external validations were performed. Results: A total of 48 characteristics were identified. The EASE model, which consists of alcohol consumption, lung infection, temperature, respiration rate, heart rate, serum urea nitrogen, and white blood cell count, had an excellent predictive performance. The EASE-based nomogram showed a significantly higher area under curve (AUC) value of 86.5% (95% CI, 84.2%-88.8%) compared with the AUC value of 78.2% for the NEWS scoring system. The AUC of EASE in the external validation dataset was 72.2% (95% CI, 66.6%-77.7%). Both calibration curves of EASE in training and external validation datasets were close to the ideal model and were well-calibrated. Conclusions: The EASE model can predict and screen ED-admitted patients with sepsis. It demonstrated superior diagnostic performance and clinical application promise by external validation and in-parallel comparison with the NEWS scoring system.

A scoping review of the clinical application of machine learning in data-driven population segmentation analysis.

OBJECTIVE: Data-driven population segmentation is commonly used in clinical settings to separate the heterogeneous population into multiple relatively homogenous groups with similar healthcare features. In recent years, machine learning (ML) based segmentation algorithms have garnered interest for their potential to speed up and improve algorithm development across many phenotypes and healthcare situations. This study evaluates ML-based segmentation with respect to (1) the populations applied, (2) the segmentation details, and (3) the outcome evaluations. MATERIALS AND METHODS: MEDLINE, Embase, Web of Science, and Scopus were used following the PRISMA-ScR criteria. Peer-reviewed studies in the English language that used data-driven population segmentation analysis on structured data from January 2000 to October 2022 were included. RESULTS: We identified 6077 articles and included 79 for the final analysis. Data-driven population segmentation analysis was employed in various clinical settings. K-means clustering is the most prevalent unsupervised ML paradigm. The most common settings were healthcare institutions. The most common targeted population was the general population. DISCUSSION: Although all the studies did internal validation, only 11 papers (13.9%) did external validation, and 23 papers (29.1%) conducted methods comparison. The existing papers discussed little validating the robustness of ML modeling. CONCLUSION: Existing ML applications on population segmentation need more evaluations regarding giving tailored, efficient integrated healthcare solutions compared to traditional segmentation analysis. Future ML applications in the field should emphasize methods' comparisons and external validation and investigate approaches to evaluate individual consistency using different methods.