Recent medical MLLMs have made significant progress in generating step by step textual reasoning chains. However, they still struggle with complex clinical tasks that necessitate dynamic and iterative focusing on fine-grained visual regions. To close this gap, we introduce Ophiuchus, a versatile, tool-augmented framework that equips an MLLM to (i) decide when fine-grained visual evidence is needed, (ii) determine where to probe and ground within the medical image, and (iii) seamlessly weave the relevant sub-image content back into an interleaved, multimodal chain of thought for precise segmentation and diagnosis. Ophiuchus moves beyond mere tool-calling by tightly fusing the MLLM’s inherent grounding and reasoning capabilities with external tools, enabling more accurate and trustworthy decisions. The core of our method is a three-stage training strategy: cold-start SFT for basic tool selection; self-reflection fine-tuning to strengthen decision revision; and agentic tool reinforcement learning to elicit sophisticated, expert-like diagnostic behaviors. Extensive experiments show that Ophiuchus consistently outperforms both closed-source and open-source SOTA methods across diverse medical benchmarks, including VQA, detection, and reasoning-based segmentation.

OBJECTIVE:To determine the accuracy of injury severity score (ISS)in the assessment of patients with severe trauma by the consistency analysis of the patients'ISS with severe trauma scored by three clinicians, and to guide the allocation of medical resource.METHODS:Through retrospective analysis of 100 patients with serious or severe trauma admitted to Peking University People's Hospital since September 2020 to December 2021 (ISS≥16 points), we conducted a consistency analysis of ISS within different evaluators. The general information (gender, age), vital signs, physical examination, imaging, laboratory examination and other associated data of the patients after admission were retrospectively diagnosed by 3 clinicians specializing in trauma surgery and ISS was determined. SPSS 22.0 software was used for statistical analysis, descriptive reports were made on the observed values of each set of data, and Fleiss kappa test was used for consistency analysis of the credibility of the ISS within three clinicians.RESULTS:Through the consistency analysis of the ISS in 100 patients with severe trauma scored by 3 eva-luators, the total Fleiss kappa value was 0.581, and the overall consistency was medium. Consistency analysis of the different scores was conducted according to the calculation rules of ISS. Among the patients with single-site severe trauma, abbreviated injury scale (AIS) was 4 or 5 points, ISS was 16 or 25 points, and Fleiss kappa value was 0.756 and 0.712 within the three evaluators, showing a relatively high consistency. AIS of each part was more than 4 points, and total ISS was more than 41 points in the severe trauma patients, Fleiss kappa values are higher than 0.8 within the 3 evaluators, showing a high consistency.CONCLUSION:According to the consistency analysis of severe trauma patients ISS within the three evaluators, when the severe trauma patients with ISS≥16 points are treated or transported, there is a certain accuracy error when the score is used for inter-department communication or inter-hospital transportation, and the consistency of different evaluators for the same injury is moderate. It may lead to misjudgment of the severity of trauma and misallocation of medical resources. However, for trauma patients with single or multiple site AIS≥4 points, ISS is highly consistent among different evaluators, which can accurately indicate the severity of the patient's condition.

Hemodialysis is the primary treatment for end-stage renal disease patients, but its mortality rate is still unacceptably high. Based on multi-modality examination data of 63,499 patients from 333medical centers, we developed a Hemodialysis Early Warning and Intervention Copilot (HEWIC) system. This system assists healthcare professionals in identifying hemodialysis patients at high risk of mortality and provides risk factors to makeintervention decisions jointly with healthcare professionals. On the retrospective cohort, HEWICachieved ROC-AUC scores of 0.82and 0.79 on one-month and three-month mortality probability prediction, respectively. We then conducted a pragmatic clinical trial (RCT, ChiCTR2100052662) to evaluate whether HEWIC could assist healthcare professionals in intervention to reduce the mortality rate of hemodialysis patients in the real world. Involving 9,965 hemodialysis patients (5,216 intervention and 4,749 control) from 58 dialysis centers, the trial indicates that HEWIC’s high-risk patient identification and treatment recommendation can help reduce the three-month mortality rate of hemodialysis patients by 38.3%, with a more pronounced effect in primary hospitals. Patients managed by the intervention group (where professionals assisted by HEWIC) received more types of drug treatment and showed varying degrees of improvement in anemia, blood pressure, blood lipids, electrolytes, and inflammatory conditions, thanthe control group. Furthermore, HEWICdoes not require additional time investment from healthcare professionals, nor does it interfere with their clinical work. This study proves that the AI-copilot system not only can benefit hemodialysis treatment but also enhance the standardization of medical care across different regions. Additionally, it also suggests that the human-AIcollaborationframework has the potential to revolutionize clinical diagnosis and treatment practice for other diseases.

In this article, we propose a new method for calculating the mixed correlation coefficient (Pearson, polyserial and polychoric) matrix and its covariance matrix based on the GMM framework. We build moment equations for each coefficient and align them together, then solve the system with Two-Step IGMM algorithm. Theory and simulation show that this estimation has consistency and asymptotic normality, and its efficiency is asymptotically equivalent to MLE. Moreover, it is much faster and the model setting is more flexible (the equations for each coefficient are blocked designed, you can only include the coefficients of interest instead of the entire correlation matrix), which can be a better initial estimation for structural equation model.

Load forecasting, as a classical problem, exhibits decreased accuracy in existing algorithms when the prediction window lengthens. This paper proposes a novel approach based on a downscaling-upscaling paradigm. In the downscaled process, the daily minimum load and the daily total load sequence are extracted from the load sequence. These two features are predicted using piecewise linear regression. The upscaling process involves clustering the daily load curves into multiple clusters using the K-means clustering method. The centroids of each cluster form a standard load curve library. During prediction, a standard load curve is selected from the library based on the daily load curve of the current day, and the final daily load curve prediction is obtained by applying translation and scaling operations. The translation and scaling parameters are determined by the predicted daily minimum load, predicted daily total load, and the daily load curve of the current day. Experimental results demonstrate that the proposed method achieves superior accuracy compared to existing methods for long prediction window widths.

An iteratively reweighted least squares (IRLS) method is proposed for estimating polyserial and polychoric correlation coefficients in this paper. It iteratively calculates the slopes in a series of weighted linear regression models fitting on conditional expected values. For polyserial correlation coefficient, conditional expectations of the latent predictor is derived from the observed ordinal categorical variable, and the regression coefficient is obtained using weighted least squares method. In estimating polychoric correlation coefficient, conditional expectations of the response variable and the predictor are updated in turns. Standard errors of the estimators are obtained using the delta method based on data summaries instead of the whole data. Conditional univariate normal distribution is exploited and a single integral is numerically evaluated in the proposed algorithm, comparing to the double integral computed numerically based on the bivariate normal distribution in the traditional maximum likelihood (ML) approaches. This renders the new algorithm very fast in estimating both polyserial and polychoric correlation coefficients. Thorough simulation studies are conducted to compare the performances of the proposed method with the classical ML methods. Real data analyses illustrate the advantage of the new method in computation speed.

Odontoid fractures are increasingly prevalent in older adults and associated with high morbidity and mortality. Optimal management remains controversial. Our study aims to investigate the association between surgical management of odontoid fractures and in-hospital mortality in a multi-center geriatric cohort. We identified patients 65 years or older with C2 odontoid fractures from the Trauma Quality Improvement Program database. The primary study outcome was in-hospital mortality. Secondary outcomes were in-hospital complications and hospital length of stay. Generalized estimating equation models were used to compare outcomes between operative and non-operative cohorts. Among the 13,218 eligible patients, 1100 (8.3%) were treated surgically. The risk of in-hospital mortality did not differ between surgical and non-surgical groups, after patient and hospital-level adjustment (OR: 0.94, 95%CI: 0.55–1.60). The risks of major complications and immobility-related complications were higher in the operative cohort (adjusted OR: 2.12, 95%CI: 1.53–2.94; and OR: 2.24, 95%CI: 1.38–3.63, respectively). Patients undergoing surgery had extended in-hospital length of stay compared to the non-operative group (9 days, IQR: 6–12 days vs. 4 days, IQR: 3–7 days). These findings were supported by secondary analyses that considered between-center differences in rates of surgery. Among geriatric patients with odontoid fractures surgical management was associated with similar in-hospital mortality, but higher in-hospital complication rates compared to non-operative management. Surgical management of geriatric patients with odontoid fractures requires careful patient selection and consideration of pre-existing comorbidities.

Daily electricity consumption forecasting is a classical problem. Existing forecasting algorithms tend to have decreased accuracy on special dates like holidays. This study decomposes the daily electricity consumption series into three components: trend, seasonal, and residual, and constructs a two-stage prediction method using piecewise linear regression as a filter and Dilated Causal CNN as a predictor. The specific steps involve setting breakpoints on the time axis and fitting the piecewise linear regression model with one-hot encoded information such as month, weekday, and holidays. For the challenging prediction of the Spring Festival, distance is introduced as a variable using a third-degree polynomial form in the model. The residual sequence obtained in the previous step is modeled using Dilated Causal CNN, and the final prediction of daily electricity consumption is the sum of the two-stage predictions. Experimental results demonstrate that this method achieves higher accuracy compared to existing approaches.

Frailty, as measured by the modified frailty index-5 (mFI-5), and older age are associated with increased mortality in the setting of spinal cord injury (SCI). However, there is limited evidence demonstrating an incremental prognostic value derived from patient mFI-5. We conducted a retrospective cohort study to evaluate in-hospital mortality among adult complete cervical SCI patients at participating centers of the Trauma Quality Improvement Program from 2010 to 2018. Logistic regression was used to model in-hospital mortality, and the area under the receiver operating characteristic curve (AUROC) of regression models with age, mFI-5, or age with mFI-5 was used to compare the prognostic value of each model. 4733 patients were eligible. We found that both age (80 y versus 60 y: OR 3.59 95% CI [2.82 4.56], P < 0.001) and mFI-5 (score ≥ 2 versus < 2: OR 1.53 95% CI [1.19 1.97], P < 0.001) had statistically significant associations with in-hospital mortality. There was no significant difference in the AUROC of a model including age and mFI-5 when compared to a model including age without mFI-5 (95% CI Δ AUROC [− 8.72 × 10 –4 0.82], P = 0.199). Both models were superior to a model including mFI-5 without age (95% CI Δ AUROC [0.06 0.09], P < 0.001). Our findings suggest that mFI-5 provides minimal incremental prognostic value over age with respect to in-hospital mortality for patients complete cervical SCI.

Abstract Frailty, as measured by the modified frailty index-5 (mFI-5), and older age are associated with increased mortality in the setting of spinal cord injury (SCI). However, a comparison of the predictive power of each measure has not been completed. We conducted a retrospective cohort study to evaluate in-hospital mortality among adult complete cervical SCI patients at participating centers of the Trauma Quality Improvement Program from 2010 to 2018. Logistic regression was used to predict in-hospital mortality, and the area under the Receiver Operating Characteristic curve (AUROC) of regression models with age, mFI-5, or age with mFI-5 was used to compare predictive power. 4,733 patients were eligible. We found significant effect of age > 75 years (OR 9.77 95% CI [7.21 13.29]) and mFI-5 ≥ 2 (OR 3.09 95% CI [1.85 4.99]) on in-hospital mortality. The AUROC of a model including age and mFI-5 (0.81 95%CI [0.79 0.84] AUROC) was comparable to a model with age alone (0.81 95%CI [0.79 0.83] AUROC). Both models were superior to a model with mFI-5 alone (0.75 95% CI [0.72 0.77] AUROC)). Our findings suggest that age provides more predictive power than mFI-5 in the prediction of in-hospital mortality for complete cervical SCI.

BACKGROUND:Long-term exposure to particulate air pollutants can lead to an increase in mortality of hemodialysis patients, but evidence of mortality risk with short-term exposure to ambient particulate matter is lacking. This study aimed to estimate the association of short-term exposure to ambient particulate matter across a wide range of concentrations with hemodialysis patients mortality. METHODS:We performed a time-stratified case-crossover study to estimate the association between short-term exposures to PM2.5 and PM10 and mortality of hemodialysis patients. The study included 18,114 hemodialysis death case from 279 hospitals in 41 cities since 2013. Daily particulate matter exposures were calculated by the inverse distance-weighted model based on each case's dialysis center address. Conditional logistic regression were implemented to quantify exposure-response associations. The sensitivity analysis mainly explored the lag effect of particulate matter. RESULTS:During the study period, there were 18,114 case days and 61,726 control days. Of all case and control days, average PM2.5 and PM10 levels were 43.98 μg/m3 and 70.86 μg/m3, respectively. Each short-term increase of 10 μg/m3 in PM2.5 and PM10 were statistically significantly associated with a relative increase of 1.07 % (95 % confidence interval [CI]: 0.99 % - 1.15 %) and 0.89 % (95 % CI: 0.84 % - 0.94 %) in daily mortality rate of hemodialysis patients, respectively. There was no evidence of a threshold in the exposure-response relationship. The mean of daily exposure on the same day of death and one-day prior (Lag 01 Day) was the most plausible exposure time window. CONCLUSIONS:This study confirms that short-term exposure to particulate matter leads to increased mortality in hemodialysis patients. Policy makers and public health practices have a clear and urgent opportunity to pass air quality control policies that care for hemodialysis populations and incorporate air quality into the daily medical management of hemodialysis patients.

Abstract Odontoid fractures are increasingly prevalent in older adults and associated with high morbidity and mortality. Optimal management remains controversial. Our study aims to investigate the association between surgical management of odontoid fractures and in-hospital mortality in a multi-center geriatric cohort. We identified patients 65 years or older with C2 odontoid fractures from the Trauma Quality Improvement Program database. The primary study outcome was in-hospital mortality. Secondary outcomes were in-hospital complications and hospital length of stay. Generalized estimating equation models were used to compare outcomes between operative and non-operative cohorts. Among the 13218 eligible patients, 1100 (8.3%) were treated surgically. The risk of in-hospital mortality did not differ between surgical and non-surgical groups, after patient and hospital-level adjustment (OR: 0.94, 95%CI: 0.55–1.60). The risks of major complications and immobility-related complications were higher in the operative cohort (adjusted OR: 2.12, 95%CI: 1.53–2.94; and OR: 2.24, 95%CI: 1.38–3.63, respectively). Patients undergoing surgery had extended in-hospital length of stay compared to the non-operative group (9 days, IQR: 6–12days vs. 4 days, IQR: 3-7days). These findings were supported by secondary analyses that considered between-center differences in rates of surgery. Among geriatric patients with odontoid fractures surgical management was associated with similar in-hospital mortality, but higher in-hospital complication rates compared to non-operative management. Surgical management of geriatric patients with odontoid fractures requires careful patient selection and consideration of pre-existing comorbidities.

Fatal car crashes are the leading cause of death among teenagers in the USA. The Graduated Driver Licensing (GDL) programme is one effective policy for reducing the number of teen fatal car crashes. Our study focuses on the number of fatal car crashes in Michigan during 1990–2004 excluding 1997, when the GDL started. We use Poisson regression with spatially dependent random effects to model the county level teen car crash counts. We develop a measurement error model to account for the fact that the total teenage population in the county level is used as a proxy for the teenage driver population. To the best of our knowledge, there is no existing literature that considers adjustment for measurement error in an offset variable. Furthermore, limited work has addressed the measurement errors in the context of spatial data. In our modelling, a Berkson measurement error model with spatial random effects is applied to adjust for the error-prone offset variable in a Bayesian paradigm. The Bayesian Markov chain Monte Carlo (MCMC) sampling is implemented in rstan. To assess the consequence of adjusting for measurement error, we compared two models with and without adjustment for measurement error. We found the effect of a time indicator becomes less significant with the measurement-error adjustment. It leads to our conclusion that the reduced number of teen drivers can help explain, to some extent, the effectiveness of GDL.

Background: An epidemic of Coronavirus Disease 2019 (COVID-19) began in December 2019 and triggered a Public Health Emergency of International Concern (PHEIC). We aimed to find risk factors for the progression of COVID-19 to help reducing the risk of critical illness and death for clinical help. Methods: The data of COVID-19 patients until March 20, 2020 were retrieved from four databases. We statistically analyzed the risk factors of critical/mortal and non-critical COVID-19 patients with meta-analysis. Results: Thirteen studies were included in Meta-analysis, including a total number of 3027 patients with SARS-CoV-2 infection. Male, older than 65, and smoking were risk factors for disease progression in patients with COVID-19 (male: OR= 1.76, 95% CI (1.41, 2.18), P < 0.00001; age over 65 years old: OR =6.06, 95% CI(3.98, 9.22), P < 0.00001; current smoking: OR =2.51, 95% CI(1.39, 3.32), P =0.0006). The proportion of underlying diseases such as hypertension, diabetes, cardiovascular disease, and respiratory disease were statistically significant higher in critical/mortal patients compared to the non-critical patients (diabetes: OR=3.68, 95% CI (2.68, 5.03), P < 0.00001; hypertension: OR= 2.72, 95% CI (1.60,4.64), P=0.0002; cardiovascular disease: OR =5.19, 95% CI(3.25, 8.29), P < 0.00001; respiratory disease: OR 5.15, 95% CI(2.51, 10.57), P < 0.00001). Clinical manifestations such as fever, shortness of breath or dyspnea were associated with the progression of disease [fever: OR =0.56, 95% CI (0.38, 0.82), P =0.003;shortness of breath or dyspnea: OR=4.16, 95% CI (3.13, 5.53), P < 0.00001]. Laboratory examination such as aspartate amino transferase(AST) > 40U/L, creatinine(Cr) >= 133mol/L, hypersensitive cardiac troponin I(hs-cTnI) > 28pg/mL, procalcitonin(PCT) > 0.5ng/mL, lactatede hydrogenase(LDH) > 245U/L, and D-dimer > 0.5mg/L predicted the deterioration of disease while white blood cells(WBC)<4 x 10(9)/L meant a better clinical status[AST > 40U/L:OR=4.00, 95% CI (2.46, 6.52), P < 0.00001; Cr > 133 mu mol/L: OR= 5.30, 95% CI (2.19, 12.83), P=0.0002; hs-cTnI > 28 pg/mL: OR= 43.24, 95% CI (9.92, 188.49), P < 0.00001; PCT > 0.5 ng/mL: OR =43.24, 95% CI (9.92, 188.49), P < 0.00001;LDH > 245U/L: OR= 43.24, 95% CI (9.92, 188.49), P < 0.00001; D-dimer > 0.5mg/L: OR= 43.24, 95% CI (9.92, 188.49), P < 0.00001; WBC < 4 x 10(9)/L: OR= 0.30, 95% CI (0.17, 0.51), P < 0.00001]. Conclusion: Male, aged over 65, smoking patients might face a greater risk of developing into the critical or mortal condition and the comorbidities such as hypertension, diabetes, cardiovascular disease, and respiratory diseases could also greatly affect the prognosis of the COVID-19. Clinical manifestation such as fever, shortness of breath or dyspnea and laboratory examination such as WBC, AST, Cr, PCT, LDH, hs-cTnI and D-dimer could imply the progression of COVID-19. (C) 2020 The British Infection Association. Published by Elsevier Ltd. All rights reserved.

The integration of industry and education promotes the cooperation between colleges and enterprises, and contributes to regional development. This paper aims to evaluate the performance of industry-education integration in higher vocational colleges of the Yangtze River Delta. Firstly, 138 higher vocational colleges were selected from the Yangtze River Delta as the research samples. Then, a comprehensive evaluation system was designed, including 11 evaluation indices. On this basis, the 2018 data of the samples were evaluated from the perspectives of faculty, teaching, scientific research, and service. The results show that, in the Yangtze River Delta, the higher vocational colleges generally do well in industry-education integration, but the integration level varied greatly from college to college; the performance of some indices should be further optimized.

Background: Based on standard computed tomography (CT) and micro-CT scan axis images, our study aims to analyse the incidence of variation of non-fusion ossification centre in the base of the odontoid and its anatomical structure characteristics, to compare ossification centre images and analyse the possible features of the ossification centre that can influence adult odontoid fractures. Materials and methods: Fifty cases were selected for standard cervical CT of the normal axis bone (second cervical) anatomy to calculate the incidence of variation of the non-fusion ossification centre in the base of the odontoid and the indexes of associated anatomical structure. In addition, five dry bone samples with the odontoid were chosen for micro-CT to analyse the clear anatomic structure of the trabecular bone in the ossification centre. Results: Incidence of variation of non-fusion ossification centre in the base of the odontoid was 28%. In the non-ossification group, the mean sagittal diameter of the base of odontoid (SDBO, mm) was 7.64 ± 1.29 mm, the mean transverse diameter of the base of odontoid (TDBO, mm) was 7.14 ± 1.55 mm, and the SDBO:TDBO ratio was 1.1 ± 0.22. In the ossification group, the mean SDBO was 7.7 ± 1.15 mm, the mean TDBO was 7.38 ± 1.32 mm, and the SDBO:TDBO ratio was 1.07 ± 0.21. There was no significant difference in the associated indexes between the ossification and non-ossification groups (p > 0.05). Micro-CT revealed the micro-structure of trabecular bone in the ossification centre and the close relationship between the trabecular bone and the odontoid. One existing non-ossification centre in the base of the odontoid was found in the five odontoid images. The trabecular bone indexes chosen in the target area of the ossification centre were weaker than those in other areas. Conclusions: The variation rate of the non-fusion ossification centre in the base of the odontoid is relatively high and may be an important factor in the aetiology of type II and III odontoid fractures.

Sparsity in features presents a big technical challenge to existing clustering methods for categorical data. Hierarchical Bayesian Bernoulli mixture model (HBBMM) incorporates constrained empirical Bayes priors for model parameters, so the resulting Expectation Maximization (EM) algorithm of estimator searching is confined in a proper region. The EM algorithm enables to obtain the maximum a posterior (MAP) estimation, in which cluster labels are simultaneously assigned. Three criteria are proposed to identify defining features of individual clusters, leading to understanding of the underlying data structures. Information based model selection criterion is applied to determine the number of clusters. Estimation consistency and performance of model selection criteria are investigated. Two real-world sparse categorical datasets are analyzed with the proposed method.

Cucurbit[7]uril (CB[7]), a representative member of the host family cucurbit[n]uril, can host-guest interact with many guest molecules such as adamantane, viologen and naphthalene derivatives. This host-guest interaction provides an easy strategy in gene vector assembling. Furthermore, CB[7] can self-assemble on gold nanospheres (AuNSs). Herein, the combination of CB[7] and AuNSs provides both advantages of host-guest interaction and photo-thermal effect of AuNSs. In this study, polyethyleneimine (PEI) and polyethylene glycol (PEG) were separately interacted with CB[7] via host-guest interaction. Then by assembling on AuNSs, PEI and PEG were combined together to condense DNA into polyplexes as well as enhance circulation stability of the polyplexes. These gene vectors were found to have high cellular uptake efficiency and low cytotoxicity. Furthermore, the well distributed AuNSs in the polyplexes could transform light into heat under light exposure because of the photothermal effect. This was found to effectively promote the entry of gene into cytoplasm and highly enhanced gene transfection efficiency.