Developing a machine learning model for bleeding prediction in patients with cancer-associated thrombosis receiving anticoagulation therapy.

BACKGROUND: Only 1 conventional score is available for assessing bleeding risk in patients with cancer-associated thrombosis (CAT): the CAT-BLEED score. OBJECTIVES: Our aim was to develop a machine learning-based risk assessment model for predicting bleeding in CAT and to evaluate its predictive performance in comparison to that of the CAT-BLEED score. METHODS: We collected 488 attributes (clinical data, biochemistry, and International Classification of Diseases, 10th Revision, diagnosis) in 1080 unique patients with CAT. We compared CAT-BLEED score, Ridge and Lasso logistic regression, random forest, and Extreme Gradient Boosting (XGBoost) algorithms for predicting major bleeding or clinically relevant nonmajor bleeding occurring 1 to 90 days, 1 to 365 days, and 90 to 455 days after venous thromboembolism (VTE). RESULTS: The predictive performances of Lasso logistic regression, random forest, and XGBoost were higher than that of the CAT-BLEED score in the prediction of bleeding occurring 1 to 90 days and 1 to 365 days after VTE. For predicting major bleeding or clinically relevant nonmajor bleeding 1 to 90 days after VTE, the CAT-BLEED score achieved a mean area under the receiver operating characteristic curve (AUROC) of 0.48 ± 0.13, while Lasso logistic regression and XGBoost both achieved AUROCs of 0.64 ± 0.12. For predicting bleeding 1 to 365 days after VTE, the CAT-BLEED score achieved a mean AUROC of 0.47 ± 0.08, while Lasso logistic regression and XGBoost achieved AUROCs of 0.64 ± 0.08 and 0.59 ± 0.08, respectively. CONCLUSION: This is the first machine learning-based risk model for bleeding prediction in patients with CAT receiving anticoagulation therapy. Its predictive performance was higher than that of the conventional CAT-BLEED score. With further development, this novel algorithm might enable clinicians to perform personalized anticoagulation strategies with improved clinical outcomes.

Diagnostic value of VEGF in peri-implantitis and its correlation with titanium particles: A controlled clinical study.

OBJECTIVES: VEGF is prototypic marker of neovascularization, repeatedly proposed as intrinsic characteristic of peri-implantitis. This study aimed to assess pattern of VEGF in peri-implantitis, its correlation with titanium particles (TPs) and capacity as respective biomarker. MATERIAL AND METHODS: Pathological specificity of VEGF was assessed in peri-implant granulations using immunohistochemistry, periodontal granulations represented Ti-free positive controls. VEGF was correlated to TPs, identified using scanning electron microscopy coupled with dispersive x-ray spectrometry. Diagnostic accuracy, sensitivity and specificity of VEGF were estimated in PICF specimens from peri-implantitis, peri-implant mucositis (PIM) and healthy peri-implant tissues (HI) using machine learning algorithms. RESULTS: Peri-implantitis exhibited rich neovascular network with expressed density in contact zones toward neutrophil infiltrates without specific pattern variations around TPs, identified in all peri-implantitis specimens (mean particle size 8.9 ± 24.8 µm2; Ti-mass (%) 0.380 ± 0.163). VEGF was significantly more expressed in peri-implantitis (47,065 ± 24.2) compared to periodontitis (31,14 ± 9.15), and positively correlated with its soluble concentrations in PICF (p = 0.01). VEGF was positively correlated to all clinical endpoints and significantly increased in peri-implantitis compared to both PIM and HI, but despite high specificity (96%), its overall diagnostic capacity was average. Two patient clusters were identified in peri-implantitis, one with 8-fold higher VEGF values compared to HI, and second with lower values comparable to PIM. SIGNIFICANCE: VEGF accurately reflects neovascularization in peri-implantitis that was expressed in contact zones toward implant surface without specific histopathological patter variation around TPs. VEGF answered requests for biomarker of peri-implantitis but further research is necessary to decrypt its exact underlying cause.

Effect of argon plasma abutment activation on soft tissue healing: RCT with histological assessment.

OBJECTIVE: To assess the peri-implant soft tissue profiles between argon plasma treatment (PT) and non-treated (NPT) healing abutments by comparing clinical and histological parameters 2 months following abutment placement. MATERIALS AND METHODS: Thirty participants were randomly assigned to argon-plasma treatment abutments group (PT) or non-treated abutments (NPT) group. Two months after healing abutment placement, soft peri-implant tissues and abutment were harvested, and histological and clinical parameters including plaque index, bleeding on probing, and keratinized mucosa diameter (KM) were assessed. Specialized stainings (hematoxylin-eosin and picrocirious red) coupled with immunohistochemistry (vimentin, collagen, and CK10) were performed to assess soft tissue inflammation and healing, and the collagen content keratinization. In addition to standard statistical methods, machine learning algorithms were applied for advanced soft tissue profiling between the test and control groups. RESULTS: PT group showed lower plaque accumulation and inflammation grade (6.71% vs. 13.25%, respectively; p-value 0.02), and more advanced connective tissue healing and integration compared to NPT (31.77% vs. 23.3%, respectively; p = 0.009). In the control group, more expressed keratinization was found compared to the PT group, showing significantly higher CK10 (>47.5%). No differences in KM were found between the groups. SIGNIFICANCE: PT seems to be a promising protocol for guided peri-implant soft tissue morphogenesis reducing plaque accumulation and inflammation, and stimulating collagen and soft tissue but without effects on epithelial tissues and keratinization.

Resolution of peri-implant mucositis following standard treatment: A prospective split-mouth study.

BACKGROUND: Peri-implant mucositis (PIM) is a pathological precursor of peri-implantitis, but its pattern of conversion to peri-implantitis is unclear and complicated to diagnose clinically, while none of the available protocols yield complete disease resolution. The aim of this study was the evaluation of PIM responsiveness to standard anti-infective mechanical treatment (AIMT) at clinical and biomarker levels, and estimation of the diagnostic capacity of bone markers as surrogate endpoints and predictors. METHODS: Systemically healthy outpatients presenting one implant exhibiting clinical signs of inflammation confined within the soft tissue (PIM) and one healthy control (HC) implant at a non-adjacent position were included. Clinical parameters and peri-implant crevicular fluid samples were collected baseline and 6 months following mechanical therapy, to assess the levels of RANKL, OPG, and IGFBP2. PIM clustering was performed using machine learning algorithms. RESULTS: Overall, 38 patients met the inclusion criteria. Therapy resulted in the reduction of all clinical and biological indicators, but respective values remained significantly higher compared to HC. Clinical examination noted 30% disease resolution at the 6-month follow-up, while 43% showed no active bone resorption. OPG showed positive prognostic value for treatment outcome, while the clustering based on active bone resorption did not differ in terms of therapeutic effectiveness. CONCLUSION: AIMT is effective in reducing the clinical and biological indicators of PIM, but complete clinical resolution was achieved in only 30% of the cases. Around one third of PIM patients exhibited active bone resorption bellow clinical detectability that was not associated with disease progression and poor treatment responsiveness.

Machine learning in scientific grant review: algorithmically predicting project efficiency in high energy physics.

As more objections have been raised against grant peer-review for being costly and time-consuming, the legitimate question arises whether machine learning algorithms could help assess the epistemic efficiency of the proposed projects. As a case study, we investigated whether project efficiency in high energy physics (HEP) can be algorithmically predicted based on the data from the proposal. To analyze the potential of algorithmic prediction in HEP, we conducted a study on data about the structure (project duration, team number, and team size) and outcomes (citations per paper) of HEP experiments with the goal of predicting their efficiency. In the first step, we assessed the project efficiency using Data Envelopment Analysis (DEA) of 67 experiments conducted in the HEP laboratory Fermilab. In the second step, we employed predictive algorithms to detect which team structures maximize the epistemic performance of an expert group. For this purpose, we used the efficiency scores obtained by DEA and applied predictive algorithms - lasso and ridge linear regression, neural network, and gradient boosted trees - on them. The results of the predictive analyses show moderately high accuracy (mean absolute error equal to 0.123), indicating that they can be beneficial as one of the steps in grant review. Still, their applicability in practice should be approached with caution. Some of the limitations of the algorithmic approach are the unreliability of citation patterns, unobservable variables that influence scientific success, and the potential predictability of the model.

BargCrEx: A System for Bargaining Based Aggregation of Crowd and Expert Opinions in Crowdsourcing.

Crowdsourcing and crowd voting systems are being increasingly used in societal, industry, and academic problems (labeling, recommendations, social choice, etc.) due to their possibility to exploit "wisdom of crowd" and obtain good quality solutions, and/or voter satisfaction, with high cost-efficiency. However, the decisions based on crowd vote aggregation do not guarantee high-quality results due to crowd voter data quality. Additionally, such decisions often do not satisfy the majority of voters due to data heterogeneity (multimodal or uniform vote distributions) and/or outliers, which cause traditional aggregation procedures (e.g., central tendency measures) to propose decisions with low voter satisfaction. In this research, we propose a system for the integration of crowd and expert knowledge in a crowdsourcing setting with limited resources. The system addresses the problem of sparse voting data by using machine learning models (matrix factorization and regression) for the estimation of crowd and expert votes/grades. The problem of vote aggregation under multimodal or uniform vote distributions is addressed by the inclusion of expert votes and aggregation of crowd and expert votes based on optimization and bargaining models (Kalai-Smorodinsky and Nash) usually used in game theory. Experimental evaluation on real world and artificial problems showed that the bargaining-based aggregation outperforms the traditional methods in terms of cumulative satisfaction of experts and crowd. Additionally, the machine learning models showed satisfactory predictive performance and enabled cost reduction in the process of vote collection.

Is the personalized approach the key to improve clinical diagnosis of peri-implant conditions? The role of bone markers.

BACKGROUND: Study objectives were 1) to estimate diagnostic capacity of clinical parameters, receptor activator of nuclear factor kappa-B (RANKL) and osteoprotegerin (OPG) to diagnose healthy peri-implant condition (HI), peri-implant mucositis (PIM) and peri-implantitis (PIMP) by assessing respective diagnostic accuracy, sensitivity, specificity and diagnostic ranges 2) to develop personalized diagnostic model (PDM) for implant monitoring. METHODS: Split-mouth study included 126 patients and 252 implants (HI = 126, PIM = 57, and PIMP = 69). RANKL and OPG concentrations were estimated in peri-implant crevicular fluid using enzyme-linked immunosorbent assay method and assessed with clinical parameters using routine statistics, while the diagnostic capacity of individual parameters and overall clinical diagnosis were estimated using classifying algorithms. PDM was developed using decision trees. RESULTS: Bleeding on probing (BOP), plaque index, and probing depth (PD) were confirmed reliable discriminants between peri-implant health and disease, while increase in PD (PD > 4 mm) and suppuration were good discriminants amongst PIM/PIMP. Bone turnover markers (BTMs) demonstrated presence of bone resorption in PIM; between comparable diagnostic ranges PIM/PIMP, PIMP was clinically distinguished from PIM in about 60% of patients while 40% remained diagnosed as false negatives. PDM demonstrated highest diagnostic capacity (accuracy: 96.27%, sensitivity: 95.00%, specificity: 100%) and defined HI: BOP ≤0.25%; PIM: BOP >0.25%, PD ≤4.5 mm; PIMP: BOP >0.25%, PD >4.5 mm and RANKL ≤19.9 pg/site; PIM: BOP >0.25%, PD >4.5 mm, and RANKL >19.9 pg/site. CONCLUSIONS: BTMs demonstrated capacity to substantially improve clinical diagnosis of peri-implant conditions. Considering lack of difference in BTMs between PIM/PIMP and cluster of PIM with exceeding BTMs, a more refined definition of peri-implant conditions according to biological characteristics is required for further BTMs validation and appropriate PIMP management.

IDPpi: Protein-Protein Interaction Analyses of Human Intrinsically Disordered Proteins.

Intrinsically disordered proteins (IDPs) are characterized by the lack of a fixed tertiary structure and are involved in the regulation of key biological processes via binding to multiple protein partners. IDPs are malleable, adapting to structurally different partners, and this flexibility stems from features encoded in the primary structure. The assumption that universal sequence information will facilitate coverage of the sparse zones of the human interactome motivated us to explore the possibility of predicting protein-protein interactions (PPIs) that involve IDPs based on sequence characteristics. We developed a method that relies on features of the interacting and non-interacting protein pairs and utilizes machine learning to classify and predict IDP PPIs. Consideration of both sequence determinants specific for conformational organizations and the multiplicity of IDP interactions in the training phase ensured a reliable approach that is superior to current state-of-the-art methods. By applying a strict evaluation procedure, we confirm that our method predicts interactions of the IDP of interest even on the proteome-scale. This service is provided as a web tool to expedite the discovery of new interactions and IDP functions with enhanced efficiency.

How frequent does peri-implantitis occur? A systematic review and meta-analysis.

OBJECTIVES: The objective of this study is to estimate the overall prevalence of peri-implantitis (PI) and the effect of different study designs, function times, and implant surfaces on prevalence rate reported by the studies adhering to the case definition of Sanz & Chapple 2012. MATERIAL AND METHODS: Following electronic and manual searches of the literature published up to February 2016, data were extracted from the studies fitting the study criteria. Meta-analysis was performed for estimation of overall prevalence of PI while the effects of the study design, function time, and implant surface type on prevalence rate were investigated using meta-regression method. RESULTS: Twenty-nine articles were included in this study. The prevalence rate in all subset meta-analyses was always higher at patient level when compared to the prevalence rate at the implant level. Prevalence of PI was 18.5% at the patient level and 12.8% at the implant level. Meta-regression analysis did not identify any association for different study designs and function times while it was demonstrated the significant association between moderately rough surfaces with lower prevalence rate of PI (p = 0.011). CONCLUSIONS: The prevalence rate of PI remains highly variable even following restriction to the clinical case definition and it seems to be affected by local factors such as implant surface characteristics. The identification of adjuvant diagnostic markers seems necessary for more accurate disease classification. CLINICAL RELEVANCE: The occurrence of PI is affected by local factors such as implant surface characteristics hence the careful assessment of the local factors should be performed within treatment planning.

The predictive value of microbiological findings on teeth, internal and external implant portions in clinical decision making.

AIM: The primary aim of this study was to evaluate 23 pathogens associated with peri-implantitis at inner part of implant connections, in peri-implant and periodontal pockets between patients suffering peri-implantitis and participants with healthy peri-implant tissues; the secondary aim was to estimate the predictive value of microbiological profile in patients wearing dental implants using data mining methods. MATERIAL AND METHODS: Fifty participants included in the present case─control study were scheduled for collection of plaque samples from the peri-implant pockets, internal connection, and periodontal pocket. Real-time polymerase chain reaction was performed to quantify 23 pathogens. Three predictive models were developed using C4.5 decision trees to estimate the predictive value of microbiological profile between three experimental sites. RESULTS: The final sample included 47 patients (22 healthy controls and 25 diseased cases), 90 implants (43 with healthy peri-implant tissues and 47 affected by peri-implantitis). Total and mean pathogen counts at inner portions of the implant connection, in peri-implant and periodontal pockets were generally increased in peri-implantitis patients when compared to healthy controls. The inner portion of the implant connection, the periodontal pocket and peri-implant pocket, respectively, presented a predictive value of microbiologic profile of 82.78%, 94.31%, and 97.5% of accuracy. CONCLUSION: This study showed that microbiological profile at all three experimental sites is differently characterized between patients suffering peri-implantitis and healthy controls. Data mining analysis identified Parvimonas micra as a highly accurate predictor of peri-implantitis when present in peri-implant pocket while this method generally seems to be promising for diagnosis of such complex infections.

Building interpretable predictive models for pediatric hospital readmission using Tree-Lasso logistic regression.

OBJECTIVES: Quantification and early identification of unplanned readmission risk have the potential to improve the quality of care during hospitalization and after discharge. However, high dimensionality, sparsity, and class imbalance of electronic health data and the complexity of risk quantification, challenge the development of accurate predictive models. Predictive models require a certain level of interpretability in order to be applicable in real settings and create actionable insights. This paper aims to develop accurate and interpretable predictive models for readmission in a general pediatric patient population, by integrating a data-driven model (sparse logistic regression) and domain knowledge based on the international classification of diseases 9th-revision clinical modification (ICD-9-CM) hierarchy of diseases. Additionally, we propose a way to quantify the interpretability of a model and inspect the stability of alternative solutions. MATERIALS AND METHODS: The analysis was conducted on >66,000 pediatric hospital discharge records from California, State Inpatient Databases, Healthcare Cost and Utilization Project between 2009 and 2011. We incorporated domain knowledge based on the ICD-9-CM hierarchy in a data driven, Tree-Lasso regularized logistic regression model, providing the framework for model interpretation. This approach was compared with traditional Lasso logistic regression resulting in models that are easier to interpret by fewer high-level diagnoses, with comparable prediction accuracy. RESULTS: The results revealed that the use of a Tree-Lasso model was as competitive in terms of accuracy (measured by area under the receiver operating characteristic curve-AUC) as the traditional Lasso logistic regression, but integration with the ICD-9-CM hierarchy of diseases provided more interpretable models in terms of high-level diagnoses. Additionally, interpretations of models are in accordance with existing medical understanding of pediatric readmission. Best performing models have similar performances reaching AUC values 0.783 and 0.779 for traditional Lasso and Tree-Lasso, respectfully. However, information loss of Lasso models is 0.35 bits higher compared to Tree-Lasso model. CONCLUSIONS: We propose a method for building predictive models applicable for the detection of readmission risk based on Electronic Health records. Integration of domain knowledge (in the form of ICD-9-CM taxonomy) and a data-driven, sparse predictive algorithm (Tree-Lasso Logistic Regression) resulted in an increase of interpretability of the resulting model. The models are interpreted for the readmission prediction problem in general pediatric population in California, as well as several important subpopulations, and the interpretations of models comply with existing medical understanding of pediatric readmission. Finally, quantitative assessment of the interpretability of the models is given, that is beyond simple counts of selected low-level features.

Distinguishing predictive profiles for patient-based risk assessment and diagnostics of plaque induced, surgically and prosthetically triggered peri-implantitis.

OBJECTIVE: To investigate whether specific predictive profiles for patient-based risk assessment/diagnostics can be applied in different subtypes of peri-implantitis. MATERIALS AND METHODS: This study included patients with at least two implants (one or more presenting signs of peri-implantitis). Anamnestic, clinical, and implant-related parameters were collected and scored into a single database. Dental implant was chosen as the unit of analysis, and a complete screening protocol was established. The implants affected by peri-implantitis were then clustered into three subtypes in relation to the identified triggering factor: purely plaque-induced or prosthetically or surgically triggered peri-implantitis. Statistical analyses were performed to compare the characteristics and risk factors between peri-implantitis and healthy implants, as well as to compare clinical parameters and distribution of risk factors between plaque, prosthetically and surgically triggered peri-implantitis. The predictive profiles for subtypes of peri-implantitis were estimated using data mining tools including regression methods and C4.5 decision trees. RESULTS: A total of 926 patients previously treated with 2812 dental implants were screened for eligibility. Fifty-six patients (6.04%) with 332 implants (4.44%) met the study criteria. Data from 125 peri-implantitis and 207 healthy implants were therefore analyzed and included in the statistical analysis. Within peri-implantitis group, 51 were classified as surgically triggered (40.8%), 38 as prosthetically triggered (30.4%), and 36 as plaque-induced (28.8%) peri-implantitis. For peri-implantitis, 51 were associated with surgical risk factor (40.8%), 38 with prosthetic risk factor (30.4%), 36 with purely plaque-induced risk factor (28.8%). The variables identified as predictors of peri-implantitis were female sex (OR = 1.60), malpositioning (OR = 48.2), overloading (OR = 18.70), and bone reconstruction (OR = 2.35). The predictive model showed 82.35% of accuracy and identified distinguishing predictive profiles for plaque, prosthetically and surgically triggered peri-implantitis. The model was in accordance with the results of risk analysis being the external validation for model accuracy. CONCLUSIONS: It can be concluded that plaque induced and prosthetically and surgically triggered peri-implantitis are different entities associated with distinguishing predictive profiles; hence, the appropriate causal treatment approach remains necessary. The advanced data mining model developed in this study seems to be a promising tool for diagnostics of peri-implantitis subtypes.

Cloud based metalearning system for predictive modeling of biomedical data.

Rapid growth and storage of biomedical data enabled many opportunities for predictive modeling and improvement of healthcare processes. On the other side analysis of such large amounts of data is a difficult and computationally intensive task for most existing data mining algorithms. This problem is addressed by proposing a cloud based system that integrates metalearning framework for ranking and selection of best predictive algorithms for data at hand and open source big data technologies for analysis of biomedical data.