Conventional Scores Research Articles

Fully automatic software for detecting radiographic joint space narrowing progression in rheumatoid arthritis: phantom study and comparison with visual assessment.

We have developed an in-house software equipped with partial image phase-only correlation (PIPOC) which can automatically quantify radiographic joint space narrowing (JSN) progression. The purpose of this study was to evaluate the software in phantom and clinical assessments. In the phantom assessment, the software's performance on radiographic images was compared to the joint space width (JSW) difference using a micrometer as ground truth. A phantom simulating a finger joint was scanned underwater. In the clinical assessment, 15 RA patients were included. The software measured the radiological progression of the finger joints between baseline and the 52nd week. The cases were also evaluated with the Genant-modified Sharp score (GSS), a conventional visual scoring method. We also quantitatively assessed these joints' synovial vascularity (SV) on power Doppler ultrasonography (0, 8, 20 and 52weeks). In the phantom assessment, the PIPOC software could detect changes in JSN with a smallest detectable difference of 0.044mm at 0.1mm intervals. In the clinical assessment, the JSW change of the joints with GSS progression detected by the software was significantly greater than those without GSS progression (p = 0.004). The JSW change of joints with positive SV at baseline was significantly higher than those with negative SV (p = 0.024). Our in-house software equipped with PIPOC can automatically and quantitatively detect slight radiographic changes of JSW in clinically inactive RA patients.

FROM CREDIT SCORING TO REGULATORY SCORING: COMPARING CREDIT SCORING MODELS FROM A REGULATORY PERSPECTIVE

Conventional credit scoring models evaluated by predictive accuracy or profitability typically serve the financial institutions and can hardly reflect their contribution on financial stability. To remedy this, we develop a novel regulatory scoring framework to quantify and compare the corresponding regulatory capital charge errors of credit scoring models. As an application of RegTech, the proposed framework considers the characteristic of example-dependence and costsensitivity in credit scoring, which is expected to enhance the ability of risk absorption of financial institutions and thus benefit the regulators. Validated on two real-world credit datasets, empirical results reveal that credit scoring models with good predictive accuracy or profitability do not necessarily provide low capital charge requirement error, which further highlights the importance of regulatory scoring framework. The family of gradient boosting decision tree (GBDT) provides significantly better average performance than industry benchmarks and deep multilayer perceptron network, especially when financial stability is the primary focus. To further examine the robustness of the proposed regulatory scoring, sampling techniques, cut-off value modification, and probability calibration are employed within the framework and the main conclusions hold in most cases. Furthermore, the analysis on the interpretability via TreeSHAP algorithm alleviates the concerns on transparency of GBDT-based models, and confirms the important roles of loan characteristics, borrowers’ solvency and creditworthiness as powerful predictors in credit scoring. Finally, the managerial implications for both financial institutions and regulators are discussed.

The Value of Open Banking Data for Application Credit Scoring: Case Study of a Norwegian Bank

Banks generally use credit scoring models to assess the creditworthiness of customers when they apply for loans or credit. These models perform significantly worse when used on potential new customers than existing customers, due to the lack of financial behavioral data for new bank customers. Access to such data could therefore increase banks’ profitability when recruiting new customers. If allowed by the customer, Open Banking APIs can provide access to balances and transactions from the past 90 days before the score date. In this study, we compare the performance of conventional application credit scoring models currently in use by a Norwegian bank with a deep learning model trained solely on transaction data available through Open Banking APIs. We evaluate the performance in terms of the AUC and Brier score and find that the models based on Open Banking data alone are surprisingly effective in predicting default compared to the conventional credit scoring models. Furthermore, an ensemble model trained on both traditional credit scoring data and features extracted from the deep learning model further outperforms the conventional application credit scoring model for new customers and narrows the performance gap between application credit scoring models for existing and new customers. Therefore, we argue that banks can increase their profitability by utilizing data available through Open Banking APIs when recruiting new customers.

In-hospital mortality risk stratification of Asian ACS patients with artificial intelligence algorithm.

Conventional risk score for predicting in-hospital mortality following Acute Coronary Syndrome (ACS) is not catered for Asian patients and requires different types of scoring algorithms for STEMI and NSTEMI patients. To derive a single algorithm using deep learning and machine learning for the prediction and identification of factors associated with in-hospital mortality in Asian patients with ACS and to compare performance to a conventional risk score. The Malaysian National Cardiovascular Disease Database (NCVD) registry, is a multi-ethnic, heterogeneous database spanning from 2006-2017. It was used for in-hospital mortality model development with 54 variables considered for patients with STEMI and Non-STEMI (NSTEMI). Mortality prediction was analyzed using feature selection methods with machine learning algorithms. Deep learning algorithm using features selected from machine learning was compared to Thrombolysis in Myocardial Infarction (TIMI) score. A total of 68528 patients were included in the analysis. Deep learning models constructed using all features and selected features from machine learning resulted in higher performance than machine learning and TIMI risk score (p < 0.0001 for all). The best model in this study is the combination of features selected from the SVM algorithm with a deep learning classifier. The DL (SVM selected var) algorithm demonstrated the highest predictive performance with the least number of predictors (14 predictors) for in-hospital prediction of STEMI patients (AUC = 0.96, 95% CI: 0.95-0.96). In NSTEMI in-hospital prediction, DL (RF selected var) (AUC = 0.96, 95% CI: 0.95-0.96, reported slightly higher AUC compared to DL (SVM selected var) (AUC = 0.95, 95% CI: 0.94-0.95). There was no significant difference between DL (SVM selected var) algorithm and DL (RF selected var) algorithm (p = 0.5). When compared to the DL (SVM selected var) model, the TIMI score underestimates patients' risk of mortality. TIMI risk score correctly identified 13.08% of the high-risk patient's non-survival vs 24.7% for the DL model and 4.65% vs 19.7% of the high-risk patient's non-survival for NSTEMI. Age, heart rate, Killip class, cardiac catheterization, oral hypoglycemia use and antiarrhythmic agent were found to be common predictors of in-hospital mortality across all ML feature selection models in this study. The final algorithm was converted into an online tool with a database for continuous data archiving for prospective validation. ACS patients were better classified using a combination of machine learning and deep learning in a multi-ethnic Asian population when compared to TIMI scoring. Machine learning enables the identification of distinct factors in individual Asian populations to improve mortality prediction. Continuous testing and validation will allow for better risk stratification in the future, potentially altering management and outcomes.

Quality assessment and refinement of chromatin accessibility data using a sequence-based predictive model

Chromatin accessibility assays are central to the genome-wide identification of gene regulatory elements associated with transcriptional regulation. However, the data have highly variable quality arising from several biological and technical factors. To surmount this problem, we developed a sequence-based machine learning method to evaluate and refine chromatin accessibility data. Our framework, gapped k-mer SVM quality check (gkmQC), provides the quality metrics for a sample based on the prediction accuracy of the trained models. We tested 886 DNase-seq samples from the ENCODE/Roadmap projects to demonstrate that gkmQC can effectively identify "high-quality" (HQ) samples with low conventional quality scores owing to marginal read depths. Peaks identified in HQ samples are more accurately aligned at functional regulatory elements, show greater enrichment of regulatory elements harboring functional variants, and explain greater heritability of phenotypes from their relevant tissues. Moreover, gkmQC can optimize the peak-calling threshold to identify additional peaks, especially for rare cell types in single-cell chromatin accessibility data.

Condition Assessment of Gas Insulated Switchgear Using Health Index and Conditional Factor Method

This paper proposes a comprehensive procedure to assess the condition of gas insulated switchgear (GIS) equipment by using the conventional weight and score method and introducing a conditional factor to improve the accuracy of the health index evaluation. Generally, the inspection and testing of GIS components are conducted according to manufacturer recommendations and guidelines in the international standards. However, this raw data has not been simplified and systematically processed for condition assessment. The score and weight technique are applied to transform the physical condition according to visible and measurable aging to numerical values in terms of component and bay health index values. The accuracy of the obtained health index has been improved by a conditional factor, which considers invisible aging factors, such as age, number of switching operations, degree of satisfactory operation, obsolescence, and adequacy of the interrupting rating. Here, a condition evaluation procedure has been developed and compared with the fuzzy logic method and the health index dominant score technique with satisfactory results. Subsequently, the proposed procedure has been developed as web application software to evaluate 175 bays of GIS in both 115 and 230 kV networks of an independent power producer supplying electricity of 3094 MW to a large industrial estate in Thailand. Eight GIS bays showed moderate or poor condition and the proper actions were assigned to prevent their failure. The software is in use in practice as a decision support tool to effectively manage the maintenance tasks and to improve supply reliability.

Molecular prognostication in Ph-negative MPNs in 2022.

The application of genomic techniques, including cytogenetics and DNA sequencing, to decipher the molecular landscape of patients with myeloproliferative neoplasms (MPNs) has radically modified diagnostic approach and management through improved risk stratification. Three driver mutated genes (JAK2, MPL, CALR) are variably harbored by >80% of patients and associated with clinical characteristics, as well as major disease-related complications and different survival outcomes. Therefore, JAK2 V617F mutation is included in the revised International Prognosis Score of Thrombosis for Essential Thrombocythemia score for prediction of thrombosis in patients with essential thrombocythemia and prefibrotic primary myelofibrosis, while a CALR type 1 mutated genotype constitutes a favorable variable for survival in patients with myelofibrosis (MF). Novel, integrated clinical and cytogenetic/mutation scores (Mutation-Enhanced International Prognostic Score System for Transplantation-Age Patients with Primary Myelofibrosis [MIPSS70/v2], genetically inspired prognostic scoring system [GIPSS], Myelofibrosis Secondary to PV and ET- Prognostic Model [MYSEC-PM]) have been devised that guide selection of stem cell transplantation candidates with MF or help predict the risk associated with the transplant procedure (Myelofibrosis Transplant Scoring System), with greater performance compared with conventional scores based on hematologic and clinical variables only. On the other hand, several clinical needs remain unmet despite the great amount of molecular information available nowadays. These include the prediction of evolution to acute leukemia in a clinically actionable time frame, the identification of patients most likely to derive durable benefits from target agents, in primis JAK inhibitors, and, conversely, the significance of molecular responses that develop in patients receiving interferon or some novel agents. Here, we discuss briefly the significance and the role of genomic analysis for prognostication in patients with MPNs from a clinician's point of view, with the intent to provide how-to-use hints.

Prediction model for patient prognosis in idiopathic pulmonary fibrosis using hybrid radiomics analysis.

To develop an imaging prognostic model for idiopathic pulmonary fibrosis (IPF) patients using hybrid auto-segmentation radiomics analysis, and compare the predictive ability between the radiomics analysis and conventional visual score methods. Data from 72 IPF patients who had undergone CT were analyzed. In the radiomics analysis, quantitative CT analysis was performed using the semi-auto-segmentation method. In the visual method, the extent of radiologic abnormalities was evaluated and the overall percentage of lung involvement was calculated by averaging values for six lung zones. Using a training cohort of 50 cases, we generated a radiomics model and a visual score model. Subsequently, we investigated the predictive ability of these models in a testing cohort of 22 cases. Three significant prognostic factors such as contrast, Idn, and cluster shade were selected by LASSO Cox regression analysis. In the visual method, multivariate Cox regression analysis revealed that honeycombing and reticulation were significant prognostic factors. Subsequently, a predictive nomogram for prognosis in IPF patients was established using these factors. In the testing cohort, the c-index of the visual and radiomics nomograms were 0.68 and 0.74, respectively. When dividing the cohort into high-risk and low-risk groups using the median nomogram score, significant differences in overall survival (OS) in the visual and radiomics models were observed (P=0.000 and P=0.0003, respectively). The prediction model with hybrid radiomics analysis had a better ability to predict OS in IPF patients than that of the visual method.

Artificial Intelligence for Detecting and Quantifying Fatty Liver in Ultrasound Images: A Systematic Review.

Non-alcoholic Fatty Liver Disease (NAFLD) is growing more prevalent worldwide. Although non-invasive diagnostic approaches such as conventional ultrasonography and clinical scoring systems have been proposed as alternatives to liver biopsy, their efficacy has been called into doubt. Artificial Intelligence (AI) is now combined with traditional diagnostic processes to improve the performance of non-invasive approaches. This study explores how well various AI methods function and perform on ultrasound (US) images to diagnose and quantify non-alcoholic fatty liver disease. A systematic review was conducted to achieve this objective. Five science bibliographic databases were searched, including PubMed, Association for Computing Machinery ACM Digital Library, Institute of Electrical and Electronics Engineers IEEE Xplore, Scopus, and Google Scholar. Only peer-reviewed English articles, conferences, theses, and book chapters were included. Data from studies were synthesized using narrative methodologies per Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) criteria. Forty-nine studies were included in the systematic review. According to the qualitative analysis, AI significantly enhanced the diagnosis of NAFLD, Non-Alcoholic Steatohepatitis (NASH), and liver fibrosis. In addition, modalities, image acquisition, feature extraction and selection, data management, and classifiers were assessed and compared in terms of performance measures (i.e., accuracy, sensitivity, and specificity). AI-supported systems show potential performance increases in detecting and quantifying steatosis, NASH, and liver fibrosis in NAFLD patients. Before real-world implementation, prospective studies with direct comparisons of AI-assisted modalities and conventional techniques are necessary.

Validity of a remote supervised preclinical Alzheimer’s disease cognitive composite (PACC) score in preclinical and prodromal Alzheimer’s disease.

AbstractBackgroundNeuropsychological tests administered to individuals in locations different from the assessor (i.e., remote assessment) can provide substantial reductions in patient burden, study cost and barriers to access. However, such tests must be optimized and validated for use in remote settings. The preclinical cognitive composite score (PACC) is derived from neuropsychological tests of episodic memory, complex attention, and general mental status. This study determined whether a set of neuropsychological tests of these domains, which can provide a PACC score, can be given in a remote supervised setting and remain sensitive to AD related cognitive impairment.MethodOlder adults classified as cognitively unimpaired (CU, n = 30), preclinical AD (n = 30) or prodromal AD (n = 25), with conventional PACC scores known, completed three computerized tests; the International Shopping list Test (ISLT), Continuous Paired Associate Learning Test (CPAL) and the International Digit Symbol Substitution Test, while supervised by a neuropsychologist in in‐clinic and remote settings. All individuals then completed the remote assessments two more times 7days (+/‐4 days) apartResultMain performance measure from each test on each assessment was standardized using the mean and SD of the CU group, assessed under supervision, at baseline. Standardized scores for each assessment were then averaged to yield a PACC score. There was no difference in rates of missing data or group mean score on individual test scores or PACC score, between in‐clinic and remote settings. Compared to the CU group, impairment in performance on the digital PACC in the prodromal AD was substantial (z = ‐1.56) but equivalent between in clinic (9z = 1.57) and remote (z = ‐1.56) settings. Performance on the digital PACC was slightly worse in the preclinical AD group (z = –0.42) but did not differ for settings. The reliability of the digital PACC score was very high (r = 0.87).ConclusionThe digital PACC derived from neuropsychological tests given in remote supervised contexts is acceptable, valid, reliable and sensitive to AD related cognitive impairment when given in‐clinic or remotely.

Relationship Between Symptom Distress, Functional Status and Quality of Life in Patients with Low Back Myofascial Pain Syndrome

Objective: To explore the relationship between symptomatic functional status and quality of life of patients with low back myofascial pain syndrome (MPS). Methods: From July 2021 to June 2022, 106 patients with low back myofascial pain syndrome in the Affiliated Hospital of Hebei University were selected as the research subjects. A total of 106 MPS patients were investigated with general information questionnaire, Memory Symptom Assessment Scale (MSAS), Oswestry Disability Index (ODI) and Short Form Questionnaire (SF-36). The relationship between quality of life and symptom distress and dysfunction was observed and analyzed based on symptom distress and dysfunction scores, SF-36 scores, and so on. Results: The total score of MSAS was 1.79 ± 0.91. The overall symptom distress of the patients was moderate. The ODI score was 18.46 ± 5.95. The functional disability of the patients was classified as moderately impaired. The MSAS-PHYS, MSAS-PSYCH, MSAS-GDI three scale scores were 2.14 ± 0.75, 1.69 ± 0.88, 1.55 ± 0.46, respectively, and the variability of the three scales is relatively large; the dimension scores were significantly lower than those of the conventional scoring models, and P &lt; 0.05, indicating a statistical difference; the scores of each dimension of the patient’s quality of life were compared with the scores of symptom distress and functional status. The higher the symptom distress score, the lower the quality of life, with P &lt; 0.05, indicating a statistical difference; the higher the score of each dimension of functional status, the better the quality of life, showing a positive correlation, and P &lt; 0.05, indicating a statistical difference. Conclusion: MPS patients face a number of physical and psychological symptoms, and their functional status is limited. Nursing staff should implement health education and intervention measures according to the actual situation of the patients, so as to improve the quality of their lives.

Validation of a Three-Dimensional Computed Tomography Reconstruction Tool for Aortic Valve Calcium Quantification

BackgroundAortic valve calcification correlates with the severity of aortic valve stenosis and a high calcium score is associated with conduction disturbances and paravalvular leakage after transcatheter aortic valve replacement. The 3mensio Structural Heart is a semiautomated software package to facilitate aortic root analysis by multislice computed tomography.The aim of the contemporary study is to validate a semiautomated calcium quantification scoring tool with a conventional manual calcium quantification tool. MethodsFifty randomly selected patients who underwent multislice computed tomography for preprocedural planning were retrospectively selected to compare the semiautomated aortic valve Agatston calcium score by 3mensio with the manually obtained score using IntelliSpace Portal as standard reference. ResultsPatients had a mean age of 76.7 ± 7.4 years and 60% were male. The median Agatston score was 3390 [interquartile range 1877-4509] with 3mensio and 3434 [interquartile range 1839-4620] with IntelliSpace.The mean difference was -0.18 [95% confidence interval (CI) −53.8 to 53.4]. The intraclass correlation coefficient between the Agatston scores using IntelliSpace and 3mensio showed an excellent correlation of 0.995 [95% CI 0.992-0.997], p ≤ 0.001. The interobserver and intraobserver variability was 0.993 ([95% CI 0.961-0.998], p ≤ 0.001) and 0.995([95%CI 0.981-0.999], p = <0.001), respectively. ConclusionsThe semiautomated calcium quantification module in 3mensio Structural Heart highly correlated with a conventional manual calcium scoring tool.

A Novel Approach to Assessing the Severity of Acute Stroke and Neurological Deficits in Patients with Acute Ischemic Stroke Using Myocardial Work Echocardiography.

We aimed to evaluate the feasibility and performance of myocardial work echocardiography in assessing the severity of acute stroke and neurological deficits in patients with acute ischemic stroke. A total of 176 patients were examined by echocardiography within 24-48 hours of symptom onset with the measurement of global and regional myocardial work. The National Institutes of Health Stroke Scale score of each patient was documented. With the increase of the National Institutes of Health Stroke Scale score, myocardial constructive work or positive work decreased (P 15 or not. The optimal cutoff value was 3.89, with a sensitivity of 100%, a specificity of 93.0%, a positive predictive value of 84.9%, a negative predictive value of 100%, and an accuracy of 95.7%. Noninvasive myocardial work is highly competent in assessing the severity of acute stroke and neurological deficits, which can be used as a powerful supplement to the conventional scoring system.

Quantitative scoring of progression in transplant glomerulopathy using digital pathology may be superior to Banff cg scoring

Antibody-mediated rejection (ABMR) is a major cause of kidney allograft failure. Biopsy-based surrogate endpoints reflecting ABMR progression on sequential biopsies that predict long-term outcome offer the potential to make treatment trials for ABMR feasible. However, the Banff transplant glomerulopathy (TG) scoring system (chronic glomerular injury score [cg]) relies on relatively crude and arbitrary ordinal grades and has low inter-observer concordance that currently limits its usefulness as a surrogate endpoint for ABMR progression in clinical drug trials. Here, we describe and validate a novel quantitative method for quantifying progression of TG in ABMR. Using digital pathology in sequential biopsies from 75 patients at various stages of ABMR, we scored all capillaries in the most affected glomeruli for basement membrane duplication that were correlated with allograft function, outcome, Banff lesion scores, and gene expression. Our digital scoring reflected TG progression better than the categorical Banff cg score and correlated with Banff ABMR and chronicity lesions, but not transcript changes. In multivariate analysis, the delta change between biopsies with serum creatinine and mean percent duplicated glomerular basement membranes was significantly associated with graft loss. Neither the delta in any Banff lesion scores (including cg) nor in gene expression was associated with outcome. Receiver operating characteristic curve analysis showed that the digital pathology approach was superior to the conventional score for predicting graft failure. Thus, our digital pathology-based approach for scoring TG accurately assessed progression in TG. However, further validation as a potential surrogate endpoint in clinical trials for the treatment of ABMR is warranted.

Revisiting Protein Quality Assessment to Include Alternative Proteins.

The high demand for novel and existing sustainable protein sources (e.g., legumes, insects, algae, and cultured meat) to replace the animal-based sources is becoming crucial. This change in protein consumption calls for the re-evaluation of the current methods to assess their quality and bioavailability. The two conventional scores for PDCAAS (protein digestibility-corrected AA score) and DIAAS (Digestible Indispensable AA Score) have their limitations and have not been re-evaluated and updated to address plant and novel proteins' quality. We suggest a sensitive physiological preclinical model that can rapidly and confidently address proteins from different sources. Our model is based on the postnatal growth, a major parameter for development and health in children, that influenced by environmental nutritional and lifestyle factors. Our results demonstrate that, with an appropriate amount of protein in the diet, almost all tested proteins performed as well as casein, the animal source. However, upon restriction (10% of calories), all alternative sources did not accomplish normal growth performance. Surprisingly, when compared to PDCAAS and DIAAS parameters obtained from the literature, no correlations were found between growth performance and these parameters, demonstrating their limitations.

Multi-Modal Analysis and Federated Learning Approach for Classification and Personalized Prognostic Assessment in Myeloid Neoplasms

Background Myeloid neoplasms (MN) present clinical and molecular heterogeneity and therefore a risk-adapted treatment strategy is mandatory. In MN, classification and prognostic tools based on clinical and morphologic criteria are being complemented by introducing genomic features. The clinical implementation of next-generation classifications and prognostic systems requires the availability of a robust methodological framework together with a solution to provide access to these technologies for clinicians. Aims Machine learning (ML) and Deep Learning (DL) approaches produce powerful predictive models and offer explainable solutions to assure full interpretability of a model when applied in clinical settings. Here we provided a comprehensive assessment of explainable ML/DL-based methods for classification and prognostic assessment of MN and we developed a solution to apply these methods across different clinical Centres through a Federated Learning (FL) approach. Methods We analysed two cohorts of patients from GenoMed4All consortium with myelodysplastic syndrome (MDS), n=2,043 and n=2,384, with available clinical and molecular features to train and validate the models. Methods were then applied to other MN, i.e. acute myeloid leukemia (AML, n=1154) and chronic myelomonocytic leukemia (CMML, n=1037). We stratified patients by two clustering approaches based on Hierarchical Dirichlet Process (HDP) and HDBSCAN combined with UMAP data reduction. We trained a Random Forest (RF) classifier to assign new patients to the existing clusters, considering Balanced Accuracy (BA) and Cohen's K (CK) as performance metrics. We then compared different survival prediction methods: CoxPH model (and its penalized version), Random Survival Forests, DeepCox, Gradient Boosting and XGboost survival methods. Models’ explainability was performed through SHapley Additive exPlanations approach (SHAP). C-index was used to evaluate the models performance. Finally, we developed a Federated Learning (FL) environment together with an imputation approach to handle missing values by a deep decoder model. Results In MDS training cohort, we identified 18 and 8 clusters by using HDBSCAN and HDP, respectively (Figure 1). We measured the average Silhouette Coefficient on the data space obtaining the following performance in terms of classification task: HDBSCAN (BA:92.7±1.3%, CK:92.1±1.4%) and HDP (BA:85.8±0.8%, CK:83.3±0.9%). Similar distributions were observed when focusing on the validation cohort. Model explainability analysis (SHAP) showed that in both populations similar features drive patients’ classification. Comparison of survival prediction for MDS is displayed in Figure 2, showing the models’ performance in the two cohorts considering demographics, clinical, cytogenetics and genomic features. Non-linear ML/DL-based methods outperformed classical CoxPH-based approaches without requiring huge data pre-processing. Moreover, all the models showed higher C-indices with respect to that of conventional IPSS-R score. SHAP analysis showed similar feature importance ranking for both training and validation cohorts. Models were then applied to AML and CMML cohorts, showing consistent results across different type of MN. Finally, we aimed to develop a federated learning (FL) solution (FedAvg, with a deep decoder model for missing data imputation) to favour a wide clinical implementation of the models. Data were collected to a single server and used to build and train a centralized model. Using global data training was expected to improve the model efficiency. This approach also ensured that the data in each node adhere to data privacy policies. We implemented CoxPH model in a setting of 3 nodes (Centers) respectively contributing to 60%, 30% and 10% of the training data. We observed that the poor node (i.e., node contributing to 10% of data) benefit from FedAvg with respect to working on an isolated setting (C-index 0.63 vs. 0.54). The centralized model trained on the whole dataset presented the highest efficiency (C-index 0.74). Conclusion Machine Learning/Deep Learning approach produces explainable and robust solutions to optimize classification and prognostic assessment in MN, as a basis for personalized medicine programs in these disorders. Federate learning algorithms allow a wide clinical implementation of the models by ensuring high performance and data protection. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal

Impact of the Alberta Stroke Program CT Score subregions on long-term functional outcomes in acute ischemic stroke: Results from two multicenter studies in China

Abstract Background and Objectives The Alberta Stroke Program CT Score (ASPECTS) is a widely used rating system for assessing infarct extent and location. We aimed to investigate the prognostic value of ASPECTS subregions’ involvement in the long-term functional outcomes of acute ischemic stroke (AIS). Materials and Methods Consecutive patients with AIS and anterior circulation large-vessel stenosis and occlusion between January 2019 and December 2020 were included. The ASPECTS score and subregion involvement for each patient was assessed using posttreatment magnetic resonance diffusion-weighted imaging. Univariate and multivariable regression analyses were conducted to identify subregions related to 3-month poor functional outcome (modified Rankin Scale scores, 3–6) in the reperfusion and medical therapy cohorts, respectively. In addition, prognostic efficiency between the region-based ASPECTS and ASPECTS score methods were compared using receiver operating characteristic curves and DeLong’s test. Results A total of 365 patients (median age, 64 years; 70% men) were included, of whom 169 had poor outcomes. In the reperfusion therapy cohort, multivariable regression analyses revealed that the involvement of the left M4 cortical region in left-hemisphere stroke (adjusted odds ratio [aOR] 5.39, 95% confidence interval [CI] 1.53–19.02) and the involvement of the right M3 cortical region in right-hemisphere stroke (aOR 4.21, 95% CI 1.05–16.78) were independently associated with poor functional outcomes. In the medical therapy cohort, left-hemisphere stroke with left M5 cortical region (aOR 2.87, 95% CI 1.08–7.59) and caudate nucleus (aOR 3.14, 95% CI 1.00–9.85) involved and right-hemisphere stroke with right M3 cortical region (aOR 4.15, 95% CI 1.29–8.18) and internal capsule (aOR 3.94, 95% CI 1.22–12.78) affected were related to the increased risks of poststroke disability. In addition, region-based ASPECTS significantly improved the prognostic efficiency compared with the conventional ASPECTS score method. Conclusion The involvement of specific ASPECTS subregions depending on the affected hemisphere was associated with worse functional outcomes 3 months after stroke, and the critical subregion distribution varied by clinical management. Therefore, region-based ASPECTS could provide additional value in guiding individual decision making and neurological recovery in patients with AIS.

IPSS-M Is More Powerful for MDS Patients Older Than 60 Years Old

Introduction: Myelodysplastic syndromes (MDS) are a group of heterogeneous hematological malignancies with complex molecular backgrounds mainly affecting the elderly. The incorporation of prognostic mutations into conventional risk scores, for example, the revised International Prognostic Scoring System (IPSS-R), can improve prognostic power. Recently, Bernard et al. developed a molecular IPSS (IPSS-M) on the basis of hematological variables and 31 putative genes. The aim of our study was to compare the prognostic value of IPSS-M with the IPSS-R. Methods: Our study consisted of 852 patients with de novo MDS from 8/2016-9/2021. IPSS-R and IPSS-M scores were calculated to define different prognostic subgroups in our cohort. A targeted gene sequencing of 141 genes was performed in 592 patients from 8/2016 to 3/2020 and a 267-gene panel for 260 patients from 4/2020-9/2021. The statistical predictive power of the prognostic models was assessed by the concordance index (C-index). A higher C-index indicates a better quality of discrimination. Results: 550 (64.6%) male and 302 (35.4%) female individuals were included. The median age was 56 years (inter-quartile ranges [IQR]: 44, 64 years) with 63.4% of subjects ≤ 60 years old. Compared with the IPSS-M group, patients in our group were younger (median age: 56 vs. 73 years, P<0.001), had higher IPSS-R scores (P<0.001) and worse IPSS-R-defined cytogenetics (P<0.001). Also, the hemoglobin levels, absolute neutrophil counts (ANC), and platelet counts were all significantly lower in Chinese subjects (P-values from 0.02 to <0.001). According to the IPSS-M, 21 patients (2.5%) were very low, 138 (16.2%) were low, 125 (14.7%) were moderate low, 113 (13.3%) were moderate high, 170 (20.0%) were high and 192 (22.5%) were very high, with a 3-year survival of 100%, 80%, 67.3%, 56.3%, 42%, 24.5% (P<0.0001, Figure 1). When analyzing the reclassification of patients from IPSS-R to IPSS-M, 247 (70.4%) were upstaged and 104 (29.6%) were downstaged. The median OS of intermediate IPSS-R patients reclassified as moderate, high and very high IPSS-M was not reached, 34 months and 13 months (P=0.014). The C-index for the IPSS-R and IPSS-M were 0.67 and 0.68 in the whole group (Figure 2). Then we compared the discriminative power of the two models in younger group (age≤ 60 years) and older group (age> 60 years). IPSS-R had a higher C-index in younger group (0.67 vs. 0.65) while IPSS-M had a higher C-index in the elderly (0.69 vs. 0.66, Figure 2), suggesting that the predictive power of IPSS-M was more reliable than IPSS-R in older group. To further analyze the differences between the younger and older ones, we compared the clinical, hematologic, cytogenetic and genetic data for both age groups. The older group was more likely to be male (72.1% vs. 60.2%, P<0.0001), had higher percentage of bone marrow blasts (P<0.0001) and worse IPSS-R karyotypes (P<0.0001). Considering the risk stratification, no significant difference was observed in IPSS-R categories between the younger and older ones (P=0.332). However, the elderly was enriched for very high IPSS-M category (27.9% vs. 19.4%, P=0.027). The incidence and distribution of mutations among the two age groups were strikingly different. The average number of genes (2.0 vs. 1.6, P=0.003) and IPSS-M mutated genes (1.6 vs. 1.3, P=0.006) in the older ones was higher than the younger ones. The most frequent genes were ASXL1 in older MDS patients (21.5%) and U2AF1 in the younger ones (26.1%). The elderly had a higher incidence of RUNX1 (16.0% vs. 9.4%, P=0.004), TP53 (14.1% vs. 7.6%, P=0.002), TET2 (11.9% vs. 5.9%, P=0.002), SRSF2 (11.2% vs. 3.0%, P<0.0001), DNMT3A (10.9& vs. 6.7%, P=0.030), STAG2 (7.1% vs. 1.9%, P<0.0001), EZH2 (5.8% vs. 2.6%, P=0.019) and DDX41 (4.8% vs. 0.9%, P<0.0001), with U2AF1 more frequently occurring in younger ones (26.1% vs. 15.4%, P<0.0001). Conclusion: The risk estimation differences between the IPSS-R and IPSS-M models in our cohort could be explained by study-specific factors, like unique ethnic groups and different population age. The IPSS-M could better stratify patients within the IPSS-R scoring system. With respect to an increased number of putative mutations and a different molecular profile in the elderly, IPSS-M model was more discriminative compared with IPSS-R in patients older than 60 years, but it had not been found to be more predictive in younger patients. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal

Synthetic Data Generation By Artificial Intelligence to Accelerate Translational Research and Precision Medicine in Hematological Malignancies

Background In hematological malignancies there is a growing demand for real-world, comprehensive data including clinical and genomic information to build powerful models to improve diagnosis, prognosis and personalized treatment choice. However, collecting such information in large patient populations is challenging and there are many issues concerning patient privacy that need to be accounted for. One approach that can circumvent these issues is the creation of synthetic data that captures the complexities of the original data set (distributions, non-linear relationships, and noise) without including any real patient information. Aims 1) Apply advanced synthetic data generation methods to real-world datasets of different hematological malignancies. 2) Develop a Synthetic Validation Framework to evaluate the quality of synthetic data and perform data augmentation. 3) Test the capability of synthetic data to accelerate translational research. Methods Here we implemented a Conditional Tabular Wasserstein Generative Adversarial Networks (GAN) architecture with Gradient Penalty to generate synthetic data. Use cases were different cohorts of patients with myelodysplastic syndrome (MDS) and acute myeloid leukemia (AML) with available clinical and molecular features. We created a Synthetic Validation Framework to evaluate the quality of generated synthetic data: Clinical Synthetic Fitness (CSF) and Genomic Synthetic Fitness (GSF) scores were calculated as the average of multiple metric tests adopted. Patients were stratified by Hierarchical Dirichlet (HD) clustering. Explainability analysis was carried out by SHapley Additive exPlanations approach (SHAP). Survival analyses were performed by Kaplan-Meier curves and CoxPH models (Experimental plan is reported in Figure 1). Results We first created a synthetic copy of a MDS cohort (n=2,043) using all the real data for training the model. We compared synthetic vs. real data, obtaining high fitness performances for both clinical and genomic features (CSF=93%; GSF=90%). HD were then applied to define clusters capturing broad dependencies among genomic features, showing comparable results in synthetic vs. real data; SHAP analysis indicated that similar features drive patients’ classification in both datasets. Finally, synthetic patients had comparable survival with respect to real ones; when applying conventional scoring system (IPSS-R), the probability of survival of the 5 risk categories was comparable between synthetic and real data. In the second experiment setting, we analysed synthetic MDS datasets with different size generated with model trained on a real dataset. Interestingly, when generating a synthetic augmented dataset (200%) we obtained high fitness performance for both clinical and genomic features (CSF=91%; GSF=89%). Moreover, all the performances showed a similar trend when considering a cohort of 1,002 patients with AML (CSF=92%; GSF=89%) thus proving evidence for high generalizability of the model across different clinical settings. Finally, we investigated if the generation of synthetic data can accelerate translational research in hematology. Since the first publication on clinical relevance of gene mutations in MDS (Leukemia 2014;28:241), it took several years to collect data in large patient populations for generating a molecular classification (JCO 2021;39:1223) and prognostic score (IPSS-M, NEJM Evid 2022;1:7). Starting from the MDS cohort available in 2014 (n=944, Leukemia 2014;28:241), we generated 300% augmented synthetic dataset. HD were applied to synthetic data to define genomic-based clinical entities, resulting in the identification of the same 8 subgroups described in a real cohort of 2,043 patients many years later. Moreover, we applied a CoxPH model to the synthetic dataset to generate a molecular prognostic score (IPSS-M_Syn). The model was based on similar molecular features as the real IPSS-M and identified 6 risk categories in which the probability of survival was similar to that of IPSS-M risk groups (Figure 2). Conclusion GAN-generated synthetic data recapitulate statistical properties and complexity of clinical and genomic features in different hematological malignancies, replicate reliable survival estimates and allow effective data augmentation. The implementation of this technology seems to accelerate precision medicine research in hematology. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal

Clinical and Microbiological Evaluation of the Chemomechanical Caries Removal Agents in Primary Molars.

Chemomechanical caries removal (CMCR) is a noninvasive procedure that uses a chemical substance to remove the diseased dentin. The natural dental architecture is also preserved using this technique, preventing patient discomfort and pulp irritation. This method of eliminating caries is based on disintegration. This method removes soft carious dentin using chemical agents and non-traumatic mechanical force. This study was carried out to evaluate the clinical and microbiological evaluation of the chemomechanical caries removal agents in primary molars. For the elimination of caries, teeth in category I (polymer bur category) were treated with Smartburs II® (SS White Dental, Lakewood, New Jersey, United States) (n = 40). Teeth treated with the new Carisolv® technology (Mediteam, Sweden) to remove caries (n = 40). Teeth were treated for removal cavities in category III (conventional group) using excavators and carbon steel low-speed round burs from Dentsply Maillefer in Switzerland (n = 40). There was an evaluation of the efficacy of caries removal, the time required for caries removal, patient satisfaction, and microbial culture assessment. In comparison to the other two categories, the conventional category median caries detector dye values were considerably lower (p value<0.05). There was no substantial difference observed between study participants belonging to category I and category II with a p-value greater than 0.05. Time taken during the removal of caries was substantially greater in category I (455.46±73.7) as compared to category II (129.21±44.18) and the traditional category (113.26±37.7). The value of significance was less than 0.05. Contrarily, no discernible difference was observed between category I and category III (the p-value was greater than 0.05). In comparison to the other two categories, the conventional group's median facial expression scale scores were substantially higher (p = 0.02). In comparison, there was no discernible difference between categories I and II (p = 0.65). It was observed that there was no substantial variation in three categories regarding the number of colonies of bacteria prior to the eradication of caries (p-value greater than 0.05). After caries had been removed, the number of living bacterial colonies in category I was noticeably greater than those in the other two categories. However, there was no discernible variation between category II and category 1 (p-value greater than 0.05). The mechanical approach has the highest clinical efficacy for removing caries. Carisolv required the most time to remove cavities. Patient satisfaction levels were greater with Carisolv than with the mechanical approach. It was also observed that Carisolv as well as the mechanical technique had greater antibacterial effectiveness.