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.

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.

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.

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.

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.

Menopause-specific Quality of Life Questionnaire: factor and Rasch analytic approach

OBJECTIVE

The study aimed to assess the psychometric properties of the Menopause-specific Quality of Life Questionnaire (MENQOL) in Serbian climacteric women.

METHODS

Two hundred mid-aged (40-65 years) women filled out a general sociodemographic and medical questionnaire, the MENQOL, the Short Form-36 questionnaire (SF-36), and Beck's Depression Inventory (BDI). The MENQOL was translated according to recommended methodology and its psychometric properties (internal consistency, discriminant, construct and criterion validity, factor and Rasch analyses) were assessed.

RESULTS

Cronbach's alpha coefficient for the Serbian MENQOL was 0.957 (Vasomotor =0.917, Psychosocial =0.907, Physical =0.928, Sexual =0.913). Corrected item-total correlation coefficients were >0.50 for all items. Factor analysis extracted six domains with total variance of 72.02%. Beside the four original domains, two new domains were observed: Attractiveness and Pain. Factor loadings for all items were appropriate (>0.4). The MENQOL correlated with the SF-36 and BDI total scores, suggesting good criterion validity. Rasch analysis showed adequate reliability (item =0.91; person =0.79) and separation (item =1.95; person =3.21) indexes, proving good reproducibility and reliability of the Serbian MENQOL. All items had infit and outfit mean squares in the standard accepted range.

CONCLUSION

Principal component analysis showed somewhat different constructs from the original scale. However, Rasch analysis confirmed that the MENQOL represents a good metric instrument for menopausal symptoms.

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.

Lipid peroxidation as risk factor for endothelial dysfunction in antiphospholipid syndrome patients

The aim of this study was to evaluate oxidative stress markers and it relations to endothelial damage as risk factor for thrombosis in patients with primary (PAPS) and secondary (SAPS) antiphospholipid syndrome (APS) in correlation to traditional risk factors. Flow-mediated (FMD) and nitroglycerine (NMD)-induced dilation of the brachial artery were studied in 140 APS patients (90 PAPS, 50 SAPS) and 40 controls matched by age, sex, and conventional risk factors for atherosclerosis. Markers of oxidative stress, lipid hydroperoxydes (LOOH), advanced oxidation protein products (AOPP), total sulfhydryl groups (tSHG), and paraoxonase 1 activity (PON1) were determined by spectrophotometric method. Oxidative stress dominates in APS patients. LOOH and AOPP correlate to lipid fractions (p < 0.05), unlike PON1, tSHG that correlated to antiphospholipid antibody positivity (p < 0.05). FMD was lower in APS patients comparing to controls (p < 0.001). Cholesterol is independent variable for FMD impairment in control group (p = 0.011); LOOH in PAPS (p = 0.004); LOOH, aCL, and triglycerides in SAPS patients (p = 0.009, p = 0.049, and p = 0.012, respectively). Combined predictive of aCL and LOOH is better for FMD impairment than LOOH alone in both PAPS and SAPS patients (AUC 0.727, p = 0.001, 95 % CI 0.616-0.837 and AUC 0.824, p˂0.001, 95 % CI 0.690-0.957, respectively). Lipid peroxidation is independent predictor for endothelial dysfunction in APS patients. We demonstrated synergistic effect of aCL and LOOH as risk for endothelial impairment in both PAPS and SAPS patients.

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.

Serbian version of the Women's Health Questionnaire: psychometric properties

OBJECTIVE

The aim of this study was to translate the Women's Health Questionnaire (WHQ) into the Serbian language and assess its validity and reliability in a population of Serbian menopausal women.

METHODS

The study included peri- and postmenopausal women from two Community Health Centers in Belgrade. Women filled out the WHQ, the Short Form-36 questionnaire (SF-36) and Beck's Depression Inventory (BDI). The WHQ was translated according to recommended methodology for cultural adaptation of questionnaires and its psychometric characteristics (internal consistency, inter-rater reliability, factor analysis, sensitivity, discriminant, construct and criterion validity) were tested.

RESULTS

In the Serbian population, the mean values of the WHQ domains were mostly comparable with reference Mediterranean countries. Whole-scale Cronbach's α was 0.838. Moreover, five WHQ domains had a value of Cronbach's α above the acceptable limit. There were no significant differences in WHQ scores between our two investigators. On exploratory factor analysis, we obtained ten factors (two items formed a new factor - 'Menstrual pathology'). Almost all SF-36 domains were significantly associated with WHQ domains, while the BDI was associated with domains: depressive mood, anxiety and sleep problems. Based on ROC analysis, WHQ is slightly more reliable for perimenopausal than postmenopausal Serbian women.

CONCLUSION

The Serbian version of the WHQ showed very good reliability and validity in assessment of quality of life among menopausal women. The WHQ is applicable for both peri- and postmenopausal women.

A simple method to assess freezing of gait in Parkinson's disease patients

Freezing of gait (FOG) can be assessed by clinical and instrumental methods. Clinical examination has the advantage of being available to most clinicians; however, it requires experience and may not reveal FOG even for cases confirmed by the medical history. Instrumental methods have an advantage in that they may be used for ambulatory monitoring. The aim of the present study was to describe and evaluate a new instrumental method based on a force sensitive resistor and Pearson's correlation coefficient (Pcc) for the assessment of FOG. Nine patients with Parkinson's disease in the "on" state walked through a corridor, passed through a doorway and made a U-turn. We analyzed 24 FOG episodes by computing the Pcc between one "regular/normal" step and the rest of the steps. The Pcc reached ±1 for "normal" locomotion, while correlation diminished due to the lack of periodicity during FOG episodes. Gait was assessed in parallel with video. FOG episodes determined from the video were all detected with the proposed method. The computed duration of the FOG episodes was compared with those estimated from the video. The method was sensitive to various types of freezing; although no differences due to different types of freezing were detected. The study showed that Pcc analysis permitted the computerized detection of FOG in a simple manner analogous to human visual judgment, and its automation may be useful in clinical practice to provide a record of the history of FOG.

Muscle stretch-induced modulation of noxiously activated dorsal horn neurons of feline spinal cord

The present work was designed to check for the possibility of interactions between mechanical innocuous and chemically induced noxious muscle afferent inputs on discharge behavior of nociceptive superficial dorsal horn neurons (SDHNs) of the spinal cord in decerebrated cats. The innocuous and noxious stimuli were applied separately and in combination, so that the effects of the innocuous stimulus on nociceptive processing could be evaluated. The innocuous stimulus consisted of ramp-and-hold stretches of the gastrocnemius muscles, whereas the noxious stimulus consisted of i.a. injections of bradykinin (BK; 0.5-1 ml, 50 microg/ml) into the arterial circulation of same muscles. Only neurons up to approximately 1mm depth and those that responded to noxious pinch of the gastrocnemius muscles were selected for further analysis. The activity of 16 dorsal horn neurons was recorded extracellularly with high-impedance glass microelectrodes, out of which seven responded to stretch, while 12 neurons responded to bradykinin injections. The bradykinin injections induced three types of responses: excitatory, inhibitory and mixed. The majority of the neurons that showed excitatory and mixed responses to bradykinin were also influenced by stretches applied directly after the bradykinin injection. In these neurons, the stretch usually counteracted the bradykinin-induced response, i.e. shortening and reducing bradykinin-induced excitation and re-exciting the cells after bradykinin-induced inhibition. The mechanism of the stretch modulation is proposed to reside in a segmental spinal control of the nociceptive transmission.

Fatigue-related changes in electomyogram activity of the cat gastrocnemius during frequency-modulated efferent stimulation

Changes in the compound muscle action potentials of cat gastrocnemius muscle were studied during low- and high-frequency fatigue. Fatiguing session consisted of 25-28 repetitions of the standard single fatigue tests (1.5min interval between the tests) that included the part of continuous frequency-modulated stimulation preceded and followed by single stimuli evoking twitch contractions in the muscle. The rate of the continuous part was changed in accordance with symmetrical double-trapezoidal signal, including three successive phases of constant rate at 10, 40 and 10s(-1); between these phases of 4s duration the rate changed linearly within a 2s interval. During fatigue relative changes in compound muscle action potential waves were usually smaller than changes in tension. Within the same fatigue procedure applied to a fresh muscle, the drop in tension was as much as 35% for high-rate stimulation and 59-71% for low-rate stimulation, whereas the decrease of the peak-to-peak compound muscle action potential waves amplitudes did not exceed 10-20%. Compound muscle action potential waves underwent the most pronounced depression during high-rate stimulation, the decrease proceeding during the following phase of low-rate stimulation. The tension changes during long-lasting activation were different for low- and high-frequency fatigue, with more pronounced depression during low-rate stimulation. As a rule, compound muscle action potential waves changes followed opposite patterns. Compound muscle action potential waves progressively split up, which was probably associated with a continuous slowing of the action potentials in the most fatigable motor units and the subsequent disappearance of the reactions at least in part of the motor units. Hysteresis effects in muscle contraction seem to be able, at least in part, to compensate for some of the depressive effects appearing during conduction of action potentials in muscle fibres. Changes in the compound muscle action potentials were studied during development of the muscle fatigue. These changes showed pronounced dependency on stimulation rate allowing differentiating effects of low- and high-frequency stimulation of the efferents supplying muscle under study. At the same time the fatigue-related changes in the action potentials were noticeably smaller than changes in tension, thus supporting existing concepts in the field arguing that fatigue effects are mainly connected with corresponding activity-dependent changes in muscle contraction machinery.

Fatigue effects in the cat gastrocnemius during frequency-modulated efferent stimulation

Effects of low- and high-frequency fatigue were studied on muscle dynamics in isometric conditions of the cat gastrocnemius. Fatiguing sessions consisted of 25-28 repetitions of the standard tests that included an 18-s interval of continuous frequency-modulated stimulation preceded and followed by single stimuli evoking twitch contractions. The rate of the continuous part was changed in accordance with a symmetrical double-trapezoidal signal, including three successive phases of constant rate at 10, 40 and 10s(-1); between these phases, each lasting for 4s, the rate changed linearly within a 2-s interval. The following modes of muscle activation were applied: (i) stimulation of single filaments constituting approximately one-fifth to one-seventh of the total cross-section of the L(7) and S(1) ventral roots; (ii) the distributed stimulation of five similar filaments; and (iii) direct stimulation of muscle through bipolar wire electrodes. A relative drop in tension, the fatigue index, expressed as the ratio at the end of a fatigue session over its value at the beginning of the test, was used to quantify fatigue effects. The fatigue indices during low-rate stimulation were 0.56+/-0.03 (mean+/-S.D.) at the first phase and 0. 64+/-0.02 at the third phase, while during high-rate stimulation this parameter was only 0.32+/-0.02. The high-rate stimulation noticeably increased the mean tension during low-rate stimulation; the ratio between the reactions at the third and the first phases could be as much as two to three times greater than that at the beginning of the fatigue session. It was demonstrated that the potentiation was connected with after-effects of the rate-tension hysteresis. The hysteresis decreased with fatigue, the fatigue index for the rate-tension loop areas ranging from 0.39 to 0.52 (0.45+/-0. 05, mean+/-S.D.). The fatigue processes developed more quickly and intensively in the previously fatigued muscles: the obtained fatigue indices were 0.73+/-0.05 and 0.70+/-0.10 at the first and third phases, and 0.62+/-0.06 (mean+/-S.D.) at the second phase of stimulation, respectively. In the cases of distributed and direct stimulation applied to muscles in a fresh state, fatigue dynamics did not differ significantly from those observed during single-filament stimulation. In experiments with distributed stimulation applied to previously fatigued muscles, a powerful depression of the high-rate components was registered in several cases, which seemed to be connected with depressive effects at the level of nerve-muscle synaptic transmission. The effects of low- and high-frequency fatigue were studied in isometric conditions of muscle contraction. In addition to the well-known differentiation between low- and high-frequency fatigue effects, the complex pattern of efferent stimulation used allowed us to identify additional fatigue-related changes in the rate-tension hysteresis. This hysteresis seems to be one of the possible mechanisms directed to compensate for low-frequency fatigue in the muscle contraction.

Changes in movement final position associated with agonist and antagonist muscle fatigue

We have tested the hypothesis that agonist and antagonist muscle fatigue could affect the final position of rapid, discrete movements. Six subjects performed consecutive elbow flexion and extension movements between two targets, with their eyes closed prior to, and after fatiguing the elbow extensor muscles. The results demonstrate that elbow extension movements performed in the post-test period systematically undershot the final position as compared to pre-test movements. However, attainment of the aimed final position in elbow flexion movements was unaffected by fatiguing of the extensor muscles. Undershoot of the final position obtained in extension movements was associated with agonist muscle fatigue, a result that was expected from the point of view of current motor control theories, and that could be explained by a reduced ability of the shortening muscle to exert force. On the other hand, the absence of the expected overshoot of the final position when the antagonist is fatigued, indicates the involvement of various reflex and/or central mechanisms operating around the stretched muscle that could contribute to returning the limb to the standard final position after a brief prominent overshoot.