The EN-TEx resource of multi-tissue personal epigenomes & variant-impact models.

Understanding how genetic variants impact molecular phenotypes is a key goal of functional genomics, currently hindered by reliance on a single haploid reference genome. Here, we present the EN-TEx resource of 1,635 open-access datasets from four donors (∼30 tissues × âˆ¼15 assays). The datasets are mapped to matched, diploid genomes with long-read phasing and structural variants, instantiating a catalog of >1 million allele-specific loci. These loci exhibit coordinated activity along haplotypes and are less conserved than corresponding, non-allele-specific ones. Surprisingly, a deep-learning transformer model can predict the allele-specific activity based only on local nucleotide-sequence context, highlighting the importance of transcription-factor-binding motifs particularly sensitive to variants. Furthermore, combining EN-TEx with existing genome annotations reveals strong associations between allele-specific and GWAS loci. It also enables models for transferring known eQTLs to difficult-to-profile tissues (e.g., from skin to heart). Overall, EN-TEx provides rich data and generalizable models for more accurate personal functional genomics.

Predicting the attention of others.

As social animals, people are highly sensitive to the attention of others. Seeing someone else gaze at an object automatically draws one's own attention to that object. Monitoring the attention of others aids in reconstructing their emotions, beliefs, and intentions and may play a crucial role in social alignment. Recently, however, it has been suggested that the human brain constructs a predictive model of other people's attention that is far more involved than a moment-by-moment monitoring of gaze direction. The hypothesized model learns the statistical patterns in other people's attention and extrapolates how attention is likely to move. Here, we tested the hypothesis of a predictive model of attention. Subjects saw movies of attention displayed as a bright spot shifting around a scene. Subjects were able to correctly distinguish natural attention sequences (based on eye tracking of prior participants) from altered sequences (e.g., played backward or in a scrambled order). Even when the attention spot moved around a blank background, subjects could distinguish natural from scrambled sequences, suggesting a sensitivity to the spatial-temporal statistics of attention. Subjects also showed an ability to recognize the attention patterns of different individuals. These results suggest that people possess a sophisticated model of the normal statistics of attention and can identify deviations from the model. Monitoring attention is therefore more than simply registering where someone else's eyes are pointing. It involves predictive modeling, which may contribute to our remarkable social ability to predict the mind states and behavior of others.

A CAR-T response prediction model for r/r B-NHL patients based on a T cell subset nomogram.

BACKGROUND: Chimeric antigen receptor (CAR) T cells for refractory or relapsed (r/r) B cell no-Hodgkin lymphoma (NHL) patients have shown promising clinical effectiveness. However, the factors impacting the clinical response of CAR-T therapy have not been fully elucidated. We here investigate the independent influencing factors of the efficacy of CD19 CAR-T cell infusion in the treatment of r/r B-NHL and to establish an early prediction model. METHODS: A total of 43 r/r B-NHL patients were enrolled in this retrospective study. The patients' general data were recorded, and the primary endpoint is the patients' treatment response. The independent factors of complete remission (CR) and partial remission (PR) were investigated by univariate and binary logistic regression analysis, and the prediction model of the probability of CR was constructed according to the determined independent factors. Receiver operating characteristic (ROC) and calibration plot were used to assess the discrimination and calibration of the established model. Furthermore, we collected 15 participators to validate the model. RESULTS: Univariate analysis and binary logistic regression analysis of 43 patients showed that the ratio of central memory T cell (Tcm) and naïve T cell (Tn) in cytotoxic T cells (Tc) was an independent risk factor for response to CD19 CAR-T cell therapy in r/r B-NHL. On this basis, the area under the curve (AUC) of Tcm in the Tc and Tn in the Tc nomogram model was 0.914 (95%CI 0.832-0.996), the sensitivity was 83%, and the specificity was 74.2%, which had excellent predictive value. We did not found the difference of the progression-free survival (PFS). CONCLUSIONS: The ratio of Tcm and Tn in Tc was found to be able to predict the treatment response of CD19 CAR-T cells in r/r B-NHL. We have established a nomogram model for the assessment of the CD19 CAR-T therapy response presented high specificity and sensitivity.

Validation of the Melanoma Institute of Australia's Sentinel Lymph Node Biopsy Risk Prediction Tool for Cutaneous Melanoma.

BACKGROUND: For patients with cutaneous melanoma, sentinel lymph node biopsy (SLNB) is used to stage regional lymph nodes pathologically and inform prognosis, treatment, and surveillance. To reduce unnecessary surgeries, predictive tools aim to identify those at lowest risk for node-positive disease. The Melanoma Institute of Australia (MIA)'s Prediction Tool for Sentinel Node Metastasis Risk estimates risk of a positive SLNB using patient age and primary melanoma Breslow depth, histologic subtype, ulceration, mitotic rate, and lymphovascular invasion. METHODS: A single-institution validation was performed of the MIA Calculator with 982 cutaneous melanoma patients that included all relevant clinicopathologic factors and SLNB pathology outcomes. The study evaluated discrimination via receiver operating characteristic (ROC) curves, calibration via calibration plots, and clinical utility via decision curve analysis of the MIA model in various subgroups. The data were fit to MIA model parameters via a generalized linear model to assess the odds ratio of parameters in our dataset. RESULTS: The Calculator demonstrated limited discrimination based on ROC curves (C-statistic, 0.709) and consistently underestimated risk of SLN positivity. It did not provide a net benefit over SLNB performed on all patients or reduce unnecessary procedures in the risk domain of 0% to 16%. Compared with the original development and validation cohorts, the current study cohort had thinner tumors and a larger proportion of acral melanomas. CONCLUSIONS: The Calculator generally underestimated SLN positivity risk, including assessment in patients who would be counseled to forego SLNB based on a predicted risk lower than 5%. Recognition of the tool's current limitations emphasizes the need to refine it further for use in medical decision-making.

Derivation and External Validation of the Ottawa Bloodstream Infection Model for Acutely Ill Adults.

BACKGROUND: Knowing the probability that patients have a bloodstream infection (BSI) could influence the ordering of blood cultures and interpretation of their preliminary results. Many previous BSI probability models have limited applicability and accuracy. This study used currently recommended modeling techniques and a large sample to derive and validate the Ottawa BSI Model. METHODS: At a tertiary care teaching hospital, we retrieved a random sample of 4180 adults having blood cultures in our emergency department or during the initial 48 h of the encounter. Variable selection was based on clinical experience and a systematic review of previous model performance. Model performance was measured in a temporal external validation group of 4680 patients. RESULTS: A total of 327 derivation patients had a BSI (8.0%). BSI risk increased with increased number of culture sets (2 sets: adjusted odds ratio [aOR] 1.52 [1.10-2.11]; 3 sets: 1.99 [0.86-4.58]); with indwelling catheter (aOR 2.07 [1.34-3.20); with increasing temperature, heart rate, and neutrophil-lymphocyte ratio; and with decreasing systolic blood pressure, platelet count, urea-creatinine ratio, and estimated glomerular filtration rate. In the temporal external validation group, model discrimination was good (c-statistic 0.71 [0.69-0.74]) and calibration was very good (integrated calibration index .016 [.010-.024]). Exclusion of validation patients with acute SARS-CoV-2 infection improved discrimination slightly (c-statistic 0.73 [0.69-0.76]). CONCLUSIONS: The Ottawa BSI Model uses commonly available data to return an expected BSI probability for acutely ill patients. However, it cannot exclude BSI and its complexity requires computational assistance to use.

The role of adenocarcinoma subtypes and immunohistochemistry in predicting lymph node metastasis in early invasive lung adenocarcinoma.

BACKGROUND: Identifying lymph node metastasis areas during surgery for early invasive lung adenocarcinoma remains challenging. The aim of this study was to develop a nomogram mathematical model before the end of surgery for predicting lymph node metastasis in patients with early invasive lung adenocarcinoma. METHODS: In this study, we included patients with invasive lung adenocarcinoma measuring ≤ 2 cm who underwent pulmonary resection with definite pathology at Qilu Hospital of Shandong University from January 2020 to January 2022. Preoperative biomarker results, clinical features, and computed tomography characteristics were collected. The enrolled patients were randomized into a training cohort and a validation cohort in a 7:3 ratio. The training cohort was used to construct the predictive model, while the validation cohort was used to test the model independently. Univariate and multivariate logistic regression analyses were performed to identify independent risk factors. The prediction model and nomogram were established based on the independent risk factors. Recipient operating characteristic (ROC) curves were used to assess the discrimination ability of the model. Calibration capability was assessed using the Hosmer-Lemeshow test and calibration curves. The clinical utility of the nomogram was assessed using decision curve analysis (DCA). RESULTS: The overall incidence of lymph node metastasis was 13.23% (61/461). Six indicators were finally determined to be independently associated with lymph node metastasis. These six indicators were: age (P < 0.001), serum amyloid (SA) (P = 0.008); carcinoma antigen 125 (CA125) (P = 0. 042); mucus composition (P = 0.003); novel aspartic proteinase of the pepsin family A (Napsin A) (P = 0.007); and cytokeratin 5/6 (CK5/6) (P = 0.042). The area under the ROC curve (AUC) was 0.843 (95% CI: 0.779-0.908) in the training cohort and 0.838 (95% CI: 0.748-0.927) in the validation cohort. the P-value of the Hosmer-Lemeshow test was 0.0613 in the training cohort and 0.8628 in the validation cohort. the bias of the training cohort corrected C-index was 0.8444 and the bias-corrected C-index for the validation cohort was 0.8375. demonstrating that the prediction model has good discriminative power and good calibration. CONCLUSIONS: The column line graphs created showed excellent discrimination and calibration to predict lymph node status in patients with ≤ 2 cm invasive lung adenocarcinoma. In addition, the predictive model has predictive potential before the end of surgery and can inform clinical decision making.

Ferroptosis-related lncRNA NRAV affects the prognosis of hepatocellular carcinoma via the miR-375-3P/SLC7A11 axis.

Ferroptosis has important value in cancer treatment. It is significant to explore the new ferroptosis-related lncRNAs prediction model in Hepatocellular carcinoma (HCC) and the potential molecular mechanism of ferroptosis-related lncRNAs. We constructed a prognostic multi-lncRNA signature based on ferroptosis-related differentially expressed lncRNAs in HCC. qRT-PCR was applied to determine the expression of lncRNA in HCC cells. The biological roles of NRAV in vitro and in vivo were determined by performing a series of functional experiments. Furthermore, dual-luciferase reporter and RNA immunoprecipitation (RIP) assays were used to confirm the interaction of NRAV with miR-375-3P. We identified 6 differently expressed lncRNAs associated with the prognosis of HCC. Kaplan-Meier analyses revealed the high-risk lncRNAs signature associated with poor prognosis of HCC. Moreover, the AUC of the lncRNAs signature showed utility in predicting HCC prognosis. Further functional experiments show that the high expression of NRAV can strengthen the viciousness of HCC. Interestingly, we found that NRAV can enhance iron export and ferroptosis resistance. Further study showed that NRAV competitively binds to miR-375-3P and attenuates the inhibitory effect of miR-375-3P on SLC7A11, affecting the prognosis of patients with HCC. In conclusion, We developed a novel ferroptosis-related lncRNAs prognostic model with important predictive value for the prognosis of HCC. NRAV is important in ferroptosis induction through the miR-375-3P/SLC7A11 axis.

Data processing for predicting DNA damaging properties of complex UV sources.

A growing number of experimental evidence emphasizes that photobiological phenomena are not always the sum of the effect of individual wavelengths present in the emission spectrum of light sources. Unfortunately, tools are missing to identify such non-additive effects and predict effects of various exposure conditions. In the present work, we addressed these points for the formation of pyrimidine dimers in DNA upon co-exposure to UVC, UVB and UVA radiation. We first applied a combination index approach to determine whether mixtures of theses UV ranges exhibited additive, inhibitory or synergistic effects on the formation of cyclobutane pyrimidine dimers, (6-4) photoproducts and Dewar isomers. A predictive approach based on an experiments plan strategy was then used to quantify the contribution of each wavelength range to the formation of DNA photoproducts. The obtained models allowed us to accurately predict the level of pyrimidine dimers in DNA irradiated under different conditions. The data were found to be more accurate than those obtained with the simple additive approach underlying the use of action spectra. Experiment plans thus appear as an attractive concept that could be widely applied in photobiology even for cellular experiments.

A nomogram model for predicting the efficacy of cyclosporine in patients with pure red cell aplasia.

Pure red cell aplasia (PRCA) is a rare bone marrow disorder characterized by a severe reduction or absence of erythroid precursor cells, without affecting granulocytes and megakaryocytes. Immunosuppressive therapies, particularly cyclosporine, have demonstrated efficacy as a primary treatment. This study aims to develop a predictive model for assessing the efficacy of cyclosporine in acquired PRCA (aPRCA). This retrospective study encompasses newly treated aPRCA patients at the General Hospital of Tianjin Medical University. Diagnosis criteria include severe anemia, and absolute reticulocyte count below 10 × 109/L, with normal white blood cell and platelet counts, and a severe reduction in bone marrow erythroblasts. Cyclosporine therapy was administered, with dose adjustments based on blood concentration. Response to cyclosporine was evaluated according to established criteria. Statistical analysis involved logistic multi-factor regression, generating a predictive model. The study included 112 aPRCA patients with a median age of 63.5 years. Patients presented with severe anemia (median Hb, 56 g/L) and reduced reticulocyte levels. Eighty-six patients had no bone marrow nucleated erythroblasts. Primary PRCA accounted for 62 cases (55.4%), and secondary PRCA accounted for 50 cases (44.6%). Univariate analysis revealed that ferritin, platelet to lymphocyte ratio (PLR), and CD4/CD8 ratio influenced treatment response. Multivariate analysis further supported the predictive value of these factors. A prediction model was constructed using ferritin, PLR, and CD4/CD8 ratio, demonstrating high sensitivity and specificity. The ferritin, PLR, and CD4/CD8-based nomogram showed good predictive ability for aPRCA response to cyclosporine. This model has potential clinical value for individualized diagnosis and treatment of aPRCA patients.

Pattern-centric transformation of omics data grounded on discriminative gene associations aids predictive tasks in TCGA while ensuring interpretability.

The increasing prevalence of omics data sources is pushing the study of regulatory mechanisms underlying complex diseases such as cancer. However, the vast quantities of molecular features produced and the inherent interplay between them lead to a level of complexity that hampers both descriptive and predictive tasks, requiring custom-built algorithms that can extract relevant information from these sources of data. We propose a transformation that moves data centered on molecules (e.g., transcripts and proteins) to a new data space focused on putative regulatory modules given by statistically relevant co-expression patterns. To this end, the proposed transformation extracts patterns from the data through biclustering and uses them to create new variables with guarantees of interpretability and discriminative power. The transformation is shown to achieve dimensionality reductions of up to 99% and increase predictive performance of various classifiers across multiple omics layers. Results suggest that omics data transformations from gene-centric to pattern-centric data supports both prediction tasks and human interpretation, notably contributing to precision medicine applications.

AI-powered prediction of HCC recurrence after surgical resection: Personalised intervention opportunities using patient-specific risk factors.

BACKGROUND: Hepatocellular carcinoma (HCC) recurrence following surgical resection remains a significant clinical challenge, necessitating reliable predictive models to guide personalised interventions. In this study, we sought to harness the power of artificial intelligence (AI) to develop a robust predictive model for HCC recurrence using comprehensive clinical datasets. METHODS: Leveraging data from 958 patients across multiple centres in Australia and Hong Kong, we employed a multilayer perceptron (MLP) as the optimal classifier for model generation. RESULTS: Through rigorous internal cross-validation, including a cohort from the Chinese University of Hong Kong (CUHK), our AI model successfully identified specific pre-surgical risk factors associated with HCC recurrence. These factors encompassed hepatic synthetic function, liver disease aetiology, ethnicity and modifiable metabolic risk factors, collectively contributing to the predictive synergy of our model. Notably, our model exhibited high accuracy during cross-validation (.857 ± .023) and testing on the CUHK cohort (.835), with a notable degree of confidence in predicting HCC recurrence within accurately classified patient cohorts. To facilitate clinical application, we developed an online AI digital tool capable of real-time prediction of HCC recurrence risk, demonstrating acceptable accuracy at the individual patient level. CONCLUSION: Our findings underscore the potential of AI-driven predictive models in facilitating personalised risk stratification and targeted interventions to mitigate HCC recurrence by identifying modifiable risk factors unique to each patient. This model aims to aid clinicians in devising strategies to disrupt the underlying carcinogenic network driving recurrence.

Defining the role of multiparametric MRI in predicting prostate cancer extracapsular extension.

OBJECTIVES: To identify the predictive factors of prostate cancer extracapsular extension (ECE) in an institutional cohort of patients who underwent multiparametric MRI of the prostate prior to radical prostatectomy (RP). PATIENTS AND METHODS: Overall, 126 patients met the selection criteria, and their medical records were retrospectively collected and analysed; 2 experienced radiologists reviewed the imaging studies. Logistic regression analysis was conducted to identify the variables associated to ECE at whole-mount histology of RP specimens; according to the statistically significant variables associated, a predictive model was developed and calibrated with the Hosmer-Lomeshow test. RESULTS: The predictive ability to detect ECE with the generated model was 81.4% by including the length of capsular involvement (LCI) and intraprostatic perineural invasion (IPNI). The predictive accuracy of the model at the ROC curve analysis showed an area under the curve (AUC) of 0.83 [95% CI (0.76-0.90)], p < 0.001. Concordance between radiologists was substantial in all parameters examined (p < 0.001). Limitations include the retrospective design, limited number of cases, and MRI images reassessment according to PI-RADS v2.0. CONCLUSION: The LCI is the most robust MRI factor associated to ECE; in our series, we found a strong predictive accuracy when combined in a model with the IPNI presence. This outcome may prompt a change in the definition of PI-RADS score 5.

A meta-analysis of diabetes risk prediction models applied to prediabetes screening.

AIM: To provide a systematic overview of diabetes risk prediction models used for prediabetes screening to promote primary prevention of diabetes. METHODS: The Cochrane, PubMed, Embase, Web of Science and China National Knowledge Infrastructure (CNKI) databases were searched for a comprehensive search period of 30 August 30, 2023, and studies involving diabetes prediction models for screening prediabetes risk were included in the search. The Quality Assessment Checklist for Diagnostic Studies (QUADAS-2) tool was used for risk of bias assessment and Stata and R software were used to pool model effect sizes. RESULTS: A total of 29 375 articles were screened, and finally 20 models from 24 studies were included in the systematic review. The most common predictors were age, body mass index, family history of diabetes, history of hypertension, and physical activity. Regarding the indicators of model prediction performance, discrimination and calibration were only reported in 79.2% and 4.2% of studies, respectively, resulting in significant heterogeneity in model prediction results, which may be related to differences between model predictor combinations and lack of important methodological information. CONCLUSIONS: Numerous models are used to predict diabetes, and as there is an association between prediabetes and diabetes, researchers have also used such models for screening the prediabetic population. Although it is a new clinical practice to explore, differences in glycaemic metabolic profiles, potential complications, and methods of intervention between the two populations cannot be ignored, and such differences have led to poor validity and accuracy of the models. Therefore, there is no recommended optimal model, and it is not recommended to use existing models for risk identification in alternative populations; future studies should focus on improving the clinical relevance and predictive performance of existing models.

A Predictive Model for Prolonged Mechanical Ventilation After Triple-Branched Stent Graft for Acute Type A Aortic Dissection.

INTRODUCTION: The aim of this study is to develop a model for predicting the risk of prolonged mechanical ventilation (PMV) following surgical repair of acute type A aortic dissection (AAAD). METHODS: We retrospectively collected clinical data from 381 patients with AAAD who underwent emergency surgery. Clinical features variables for predicting postoperative PMV were selected through univariate analysis, least absolute shrinkage and selection operator regression analysis, and multivariate logistic regression analysis. A risk prediction model was established using a nomogram. The model's accuracy and reliability were evaluated using the area under the curve of the receiver operating characteristic curve and the calibration curve. Internal validation of the model was performed using bootstrap resampling. The clinical applicability of the model was assessed using decision curve analysis and clinical impact curve. RESULTS: Among the 381 patients, 199 patients (52.2%) experienced postoperative PMV. The predictive model exhibited good discriminative ability (area under the curve = 0.827, 95% confidence interval: 0.786-0.868, P < 0.05). The calibration curve confirmed that the predicted outcomes of the model closely approximated the ideal curve, indicating agreement between the predicted and actual results (with an average absolute error of 0.01 based on 1000 bootstrap resampling). The decision curve analysis curve demonstrated that the model has significant clinical value. CONCLUSIONS: The nomogram model established in this study can be used to predict the risk of postoperative PMV in patients with AAAD. It serves as a practical tool to assist clinicians in adjusting treatment strategies promptly and implementing targeted therapeutic measures.

Importance of Transplant Era on Post-Heart Transplant Predictive Models: A UNOS Cohort Analysis.

BACKGROUND: The application of posttransplant predictive models is limited by their poor statistical performance. Neglecting the dynamic evolution of demographics and medical practice over time may be a key issue. OBJECTIVES: Our objective was to develop and validate era-specific predictive models to assess whether these models could improve risk stratification compared to non-era-specific models. METHODS: We analyzed the United Network for Organ Sharing (UNOS) database including first noncombined heart transplantations (2001-2018, divided into four transplant eras: 2001-2005, 2006-2010, 2011-2015, 2016-2018). The endpoint was death or retransplantation during the 1st-year posttransplant. We analyzed the dynamic evolution of major predictive variables over time and developed era-specific models using logistic regression. We then performed a multiparametric evaluation of the statistical performance of era-specific models and compared them to non-era-specific models in 1000 bootstrap samples (derivation set, 2/3; test set, 1/3). RESULTS: A total of 34 738 patients were included, 3670 patients (10.5%) met the composite endpoint. We found a significant impact of transplant era on baseline characteristics of donors and recipients, medical practice, and posttransplant predictive models, including significant interaction between transplant year and major predictive variables (total serum bilirubin, recipient age, recipient diabetes, previous cardiac surgery). Although the discrimination of all models remained low, era-specific models significantly outperformed the statistical performance of non-era-specific models in most samples, particularly concerning discrimination and calibration. CONCLUSIONS: Era-specific models achieved better statistical performance than non-era-specific models. A regular update of predictive models may be considered if they were to be applied for clinical decision-making and allograft allocation.

Artificial intelligence in head and neck surgery: Potential applications and future perspectives.

Artificial intelligence (AI) has the potential to improve the surgical treatment of patients with head and neck cancer. AI algorithms can analyse a wide range of data, including images, voice, molecular expression and raw clinical data. In the field of oncology, there are numerous AI practical applications, including diagnostics and treatment. AI can also develop predictive models to assess prognosis, overall survival, the likelihood of occult metastases, risk of complications and hospital length of stay.

Advances in the management of gallbladder polyps: establishment of predictive models and the rise of gallbladder-preserving polypectomy procedures.

Gallbladder polyps are a common biliary tract disease whose treatment options have yet to be fully established. The indication of "polyps ≥ 10 mm in diameter" for cholecystectomy increases the possibility of gallbladder excision due to benign polyps. Compared to enumeration of risk factors in clinical guidelines, predictive models based on statistical methods and artificial intelligence provide a more intuitive representation of the malignancy degree of gallbladder polyps. Minimally invasive gallbladder-preserving polypectomy procedures, as a combination of checking and therapeutic approaches that allow for eradication of lesions and preservation of a functional gallbladder at the same time, have been shown to maximize the benefits to patients with benign polyps. Despite the reported good outcomes of predictive models and gallbladder-preserving polypectomy procedures, the studies were associated with various limitations, including small sample sizes, insufficient data types, and unknown long-term efficacy, thereby enhancing the need for multicenter and large-scale clinical studies. In conclusion, the emergence of predictive models and minimally invasive gallbladder-preserving polypectomy procedures has signaled an ever increasing attention to the role of the gallbladder and clinical management of gallbladder polyps.

Using new indices to predict metabolism dysfunction-associated fatty liver disease (MAFLD): analysis of the national health and nutrition examination survey database.

BACKGROUND: Metabolism dysfunction-associated fatty liver disease (MAFLD), is the most common chronic liver disease. Few MAFLD predictions are simple and accurate. We examined the predictive performance of the albumin-to-glutamyl transpeptidase ratio (AGTR), plasma atherogenicity index (AIP), and serum uric acid to high-density lipoprotein cholesterol ratio (UHR) for MAFLD to design practical, inexpensive, and reliable models. METHODS: The National Health and Nutrition Examination Survey (NHANES) 2007-2016 cycle dataset, which contained 12,654 participants, was filtered and randomly separated into internal validation and training sets. This study examined the relationships of the AGTR and AIP with MAFLD using binary multifactor logistic regression. We then created a MAFLD predictive model using the training dataset and validated the predictive model performance with the 2017-2018 NHANES and internal datasets. RESULTS: In the total population, the predictive ability (AUC) of the AIP, AGTR, UHR, and the combination of all three for MAFLD showed in the following order: 0.749, 0.773, 0.728 and 0.824. Further subgroup analysis showed that the AGTR (AUC1 = 0.796; AUC2 = 0.690) and the combination of the three measures (AUC1 = 0.863; AUC2 = 0.766) better predicted MAFLD in nondiabetic patients. Joint prediction outperformed the individual measures in predicting MAFLD in the subgroups. Additionally, the model better predicted female MAFLD. Adding waist circumference and or BMI to this model improves predictive performance. CONCLUSION: Our study showed that the AGTR, AIP, and UHR had strong MAFLD predictive value, and their combination can increase MAFLD predictive performance. They also performed better in females.

Assessing racial bias in healthcare predictive models: Practical lessons from an empirical evaluation of 30-day hospital readmission models.

OBJECTIVE: Despite increased availability of methodologies to identify algorithmic bias, the operationalization of bias evaluation for healthcare predictive models is still limited. Therefore, this study proposes a process for bias evaluation through an empirical assessment of common hospital readmission models. The process includes selecting bias measures, interpretation, determining disparity impact and potential mitigations. METHODS: This retrospective analysis evaluated racial bias of four common models predicting 30-day unplanned readmission (i.e., LACE Index, HOSPITAL Score, and the CMS readmission measure applied as is and retrained). The models were assessed using 2.4 million adult inpatient discharges in Maryland from 2016 to 2019. Fairness metrics that are model-agnostic, easy to compute, and interpretable were implemented and apprised to select the most appropriate bias measures. The impact of changing model's risk thresholds on these measures was further assessed to guide the selection of optimal thresholds to control and mitigate bias. RESULTS: Four bias measures were selected for the predictive task: zero-one-loss difference, false negative rate (FNR) parity, false positive rate (FPR) parity, and generalized entropy index. Based on these measures, the HOSPITAL score and the retrained CMS measure demonstrated the lowest racial bias. White patients showed a higher FNR while Black patients resulted in a higher FPR and zero-one-loss. As the models' risk threshold changed, trade-offs between models' fairness and overall performance were observed, and the assessment showed all models' default thresholds were reasonable for balancing accuracy and bias. CONCLUSIONS: This study proposes an Applied Framework to Assess Fairness of Predictive Models (AFAFPM) and demonstrates the process using 30-day hospital readmission model as the example. It suggests the feasibility of applying algorithmic bias assessment to determine optimized risk thresholds so that predictive models can be used more equitably and accurately. It is evident that a combination of qualitative and quantitative methods and a multidisciplinary team are necessary to identify, understand and respond to algorithm bias in real-world healthcare settings. Users should also apply multiple bias measures to ensure a more comprehensive, tailored, and balanced view. The results of bias measures, however, must be interpreted with caution and consider the larger operational, clinical, and policy context.

Predictive risk model of recurrence of acute diverticulitis after the first episode.

INTRODUCTION: The recurrence of acute diverticulitis (AD) of the colon is frequent and leads to hospital readmissions and the need for elective surgery in selected cases. It is important to individualize risk factors and develop predictive tools for their identification. MATERIALS AND METHODS: This prospective observational study included 368 patients who were diagnosed with AD between 2016 and 2021 in a tertiary general university hospital during their first episode and who had a good response to antibiotic, percutaneous, or peritoneal lavage treatment. Univariate and multivariate Cox regression analyses of the variables associated with recurrence were performed. Subsequently, a predictive risk score was developed and validated through survival studies. RESULTS: After a median follow-up of 50 months, there were 71 (19.3%) cases of recurrence out of a total of 368 patients. The mean time of recurrence was 15 months, and 73.3% of cases of recurrence occurred before 2 years of follow-up. Recurrence was independently associated with presentation with colonic perforation in the antimesenteric location (HR 3.67 95% CI [1.59-8.4]) and a CRP level greater than 100 mg/dl (HR 1.69 95% CI [1.04-2.77). A score with 5 variables was created that differentiated two risk groups: intermediate risk (0-3 points), with 19% recurrence and high risk (more than 3 points), with 42% recurrence. CONCLUSIONS: The risk of recurrence after the first episode of diverticulitis can be estimated using predictive scores. The detection of high-risk patients facilitates the individualization of follow-up and treatment.