DISA tool: discriminative and informative subspace assessment with categorical and numerical outcomes

Motivation: Pattern discovery and subspace clustering play a central role in the biological domain, supporting for instance putative regulatory module discovery from omic data for both descriptive and predictive ends. In the presence of target variables (e.g. phenotypes), regulatory patterns should further satisfy delineate discriminative power properties, well-established in the presence of categorical outcomes, yet largely disregarded for numerical outcomes, such as risk profiles and quantitative phenotypes. Results: DISA (Discriminative and Informative Subspace Assessment), a Python software package, is proposed to assess patterns in the presence of numerical outcomes using well-established measures together with a novel principle able to statistically assess the correlation gain of the subspace against the overall space. Results confirm the possibility to soundly extend discriminative criteria towards numerical outcomes without the drawbacks well-associated with discretization procedures. A case study is provided to show the properties of the proposed method. Availability: DISA is freely available at https://github.com/JupitersMight/DISA under the MIT license.

Prognostic Performance of Alternative Lymph Node Classification Systems for Patients with Medullary Thyroid Cancer: A Single Center Cohort Study

Background Lymph node ratio (LNR) and the log odds of positive lymph nodes (LODDS) have been proposed as alternative lymph node (LN) classification schemes. Various cut-off values have been defined for each system, with the question of the most appropriate for patients with medullary thyroid cancer (MTC) still remaining open. We aimed to retrospectively compare the predictive impact of different LN classification systems and to define the most appropriate set of cut-off values regarding accurate evaluation of overall survival (OS) in patients with MTC. Methods 182 patients with MTC who were operated on between 1985 and 2018 were extracted from our medical database. Cox proportional hazards regression models and C-statistics were performed to assess the discriminative power of 28 LNR and 28 LODDS classifications and compare them with the N category according to the 8th edition of the AJCC/UICC TNM classification in terms of discriminative power. Regression models were adjusted for age, sex, T category, focality, and genetic predisposition. Results High LNR and LODDS are associated with advanced T categories, distant metastasis, sporadic disease, and male gender. In addition, among 56 alternative LN classifications, only one LNR and one LODDS classification were independently associated with OS, regardless of the presence of metastatic disease. The C-statistic demonstrated comparable results for all classification systems showing no clear superiority over the N category. Conclusion Two distinct alternative LN classification systems demonstrated a better prognostic performance in MTC patients than the N category. However, larger scale studies are needed to further verify our findings.

Prognostic Value of Routinely Measured Inflammatory Biomarkers in Older Cancer Patients: Pooled Analysis of Three Cohorts

Background: The prognostic assessment of older cancer patients is complicated by their heterogeneity. We aimed to assess the prognostic value of routine inflammatory biomarkers. Methods: A pooled analysis of prospective multicenter cohorts of cancer patients aged ≥70 was performed. We measured CRP and albumin, and calculated Glasgow Prognostic Score (GPS) and CRP/albumin ratio. The GPS has three levels (0 = CRP ≤ 10 mg/L, albumin ≥ 35 g/L, i.e., normal values; 1 = one abnormal value; 2 = two abnormal values). One-year mortality was assessed using Cox models. Discriminative power was assessed using Harrell’s C index (C) and net reclassification improvement (NRI). Results: Overall, 1800 patients were analyzed (mean age: 79 ± 6; males: 62%; metastases: 38%). The GPS and CRP/albumin ratio were independently associated with mortality in patients not at risk of frailty (hazard ratio [95% confidence interval] = 4.48 [2.03–9.89] for GPS1, 11.64 [4.54–29.81] for GPS2, and 7.15 [3.22–15.90] for CRP/albumin ratio > 0.215) and in patients at risk of frailty (2.45 [1.79–3.34] for GPS1, 3.97 [2.93–5.37] for GPS2, and 2.81 [2.17–3.65] for CRP/albumin ratio > 0.215). The discriminative power of the baseline clinical model (C = 0.82 [0.80–0.83]) was increased by adding GPS (C = 0.84 [0.82–0.85]; NRI events (NRI+) = 10% [2–16]) and CRP/albumin ratio (C = 0.83 [0.82–0.85]; NRI+ = 14% [2–17]). Conclusions: Routine inflammatory biomarkers add prognostic value to clinical factors in older cancer patients.

Assessing the relationship between systemic immune-inflammation index and mortality in patients with hypertrophic cardiomyopathy

Background: This study investigates the predictive value of the systemic immune-inflammation index (SII), which was calculated as platelet × neutrophil/lymphocyte ratio, for all-cause mortality in patients with hypertrophic cardiomyopathy (HCM). Methods: A total of 360 HCM patients were enrolled. They were divided into three groups based on the tertiles of baseline SII. The association between SII and all-cause mortality was analyzed. Results: There were 53 HCM patients who died during a mean follow-up time of 4.8 years (min: 6 days and max: 10.8 years), and the mortality rate was 3.0 per 100 person years. The cumulative mortality rate was significantly different among the three tertiles of SII (P = 0.004), and the mortality rate in tertile 3 was much higher than that in the first two tertiles. In reference to tertile 1, the fully adjusted hazard ratios of all-cause mortality were 1.02 for the tertile 2 (95% confidence interval [CI]: 0.45–2.31, P = 0.966) and 2.31 for tertile 3 (95% CI: 1.10–4.87, P = 0.027). No significant interactions between SII and other variables were observed during subgroup analysis. The discriminative power was better for mid-term outcome than that for short-term or long-term outcomes. Sensitivity analyses including patients with normal platelet and white blood cell count have revealed similar results. Conclusion: SII was a significant risk factor for all-cause mortality in HCM patients. However, the discriminative power was poor to moderate. It could be used in combination with other risk factors in mortality risk stratification in HCM.

Discriminative Power of Geometric Parameters of Different Cultivars of Sour Cherry Pits Determined Using Machine Learning

The aim of this study was to develop models based on linear dimensions or shape factors, and the sets of combined linear dimensions and shape factors for discrimination of sour cherry pits of different cultivars (‘Debreceni botermo’, ‘Łutówka’, ‘Nefris’, ‘Kelleris’). The geometric parameters were calculated using image processing. The pits of different sour cherry cultivars statistically significantly differed in terms of selected dimensions and shape factors. The discriminative models built based on linear dimensions produced average accuracies of up to 95% for distinguishing the pit cultivars in the case of ‘Nefris’ vs. ‘Kelleris’ and 72% for all four cultivars. The average accuracies for the discriminative models built based on shape factors were up to 95% for the ‘Nefris’ and ‘Kelleris’ pits and 73% for four cultivars. The models combining the linear dimensions and shape factors produced accuracies reaching 96% for the ‘Nefris’ vs. ‘Kelleris’ pits and 75% for all cultivars. The geometric parameters with high discriminative power may be used for distinguishing different cultivars of sour cherry pits. It can be of great importance for practical applications. It may allow avoiding the adulteration and mixing of different cultivars.

The illusion of having a large virtual body biases action-specific perception in patients with mild cognitive impairment

The illusion of having a large body makes us perceive objects as smaller than they really are. This action-specific perception effect occurs because we perceive the property of an object (i.e., size) differently according to our unique action capability (i.e., the affordance of body size). Although the body-ownership illusion contributing to this action-specific perception has been studied, its effects remain unclear in neurological patients. We examined the action-specific perception impairments of MCI patients by means of body-ownership illusion in a non-immersive virtual reality environment. Twenty healthy young adults, 21 healthy older adults, and 15 MCI patients were recruited. We assessed their “original-body action-specific perception” and “enlarged-body action-specific perception” using the original and enlarged sizes of their virtual bodies, respectively. The MCI patients’ original-body action-specific perception was no different than that of the healthy controls (p = 0.679). However, the enlarged-body action-specific perception of the MCI patients was significantly biased (p < 0.001). The inclusion of the enlarged-body action-specific perception provides additional discriminative power for early diagnosis of MCI (89.3% accuracy, 75.0% sensitivity, 100.0% specificity, and 87.5% balanced accuracy).

Effect of Fear of Falling on Mobility Measured During Lab and Daily Activity Assessments in Parkinson’s Disease

In chronic disorders such as Parkinson’s disease (PD), fear of falling (FOF) is associated with falls and reduced quality of life. With inertial measurement units (IMUs) and dedicated algorithms, different aspects of mobility can be obtained during supervised tests in the lab and also during daily activities. To our best knowledge, the effect of FOF on mobility has not been investigated in both of these settings simultaneously. Our goal was to evaluate the effect of FOF on the mobility of 26 patients with PD during clinical assessments and 14 days of daily activity monitoring. Parameters related to gait, sit-to-stand transitions, and turns were extracted from IMU signals on the lower back. Fear of falling was assessed using the Falls Efficacy Scale-International (FES-I) and the patients were grouped as with (PD-FOF+) and without FOF (PD-FOF−). Mobility parameters between groups were compared using logistic regression as well as the effect size values obtained using the Wilcoxon rank-sum test. The peak angular velocity of the turn-to-sit transition of the timed-up-and-go (TUG) test had the highest discriminative power between PD-FOF+ and PD-FOF− (r-value of effect size = 0.61). Moreover, PD-FOF+ had a tendency toward lower gait speed at home and a lower amount of walking bouts, especially for shorter walking bouts. The combination of lab and daily activity parameters reached a higher discriminative power [area under the curve (AUC) = 0.75] than each setting alone (AUC = 0.68 in the lab, AUC = 0.54 at home). Comparing the gait speed between the two assessments, the PD-FOF+ showed higher gait speeds in the capacity area compared with their TUG test in the lab. The mobility parameters extracted from both lab and home-based assessments contribute to the detection of FOF in PD. This study adds further evidence to the usefulness of mobility assessments that include different environments and assessment strategies. Although this study was limited in the sample size, it still provides a helpful method to consider the daily activity measurement of the patients with PD into clinical evaluation. The obtained results can help the clinicians with a more accurate prevention and treatment strategy.

Dissimilarity-Based Regularized Learning of Charts

Chart images exhibit significant variabilities that make each image different from others even though they belong to the same class or categories. Classification of charts is a major challenge because each chart class has variations in features, structure, and noises. However, due to the lack of affiliation between the dissimilar features and the structure of the chart, it is a challenging task to model these variations for automatic chart recognition. In this article, we present a novel dissimilarity-based learning model for similar structured but diverse chart classification. Our approach jointly learns the features of both dissimilar and similar regions. The model is trained by an improved loss function, which is fused by a structural variation-aware dissimilarity index and incorporated with regularization parameters, making the model more prone toward dissimilar regions. The dissimilarity index enhances the discriminative power of the learned features not only from dissimilar regions but also from similar regions. Extensive comparative evaluations demonstrate that our approach significantly outperforms other benchmark methods, including both traditional and deep learning models, over publicly available datasets.

The Attitudes of Polish Women towards Breastfeeding Based on the Iowa Infant Feeding Attitude Scale (IIFAS)

Background: The Iowa Infant Feeding Attitude Scale (IIFAS), which is used for the assessment of attitudes towards breastfeeding, has been found to be reliable and valid in a number of countries, but has not yet been psychometrically tested in Polish women. The purpose of the study was to report on the cultural adaptation of the IIFAS to Polish settings and on its validation, to evaluate the breastfeeding attitudes in Polish women who recently gave birth, and to identify the determinants of these attitudes. Methods: The study was performed in a group of 401 women in their first postpartum days. Results: Cronbach’s α for the scale was 0.725. Discriminative power coefficients of all questionnaire items were higher than 0.2. Subscales were strongly correlated with the total score, with a correlation coefficient of 0.803 for the “favorable toward breastfeeding” subscale (p < 0.001), and 0.803 for the “favorable toward formula feeding” subscale (p < 0.05). For the item “A mother who occasionally drinks alcohol should not breastfeed her baby”, the factor loading did not reach the criterion value, and so the item was not included in further analyses. The mean IIFAS score was 63.12 (±7.34). Conclusions: The Polish version of the IIFAS is a reliable and appropriate measure of women’s attitudes towards infant feeding in Polish settings, with acceptable psychometric properties and construct validity.

A machine learning based two-stage clinical decision support system for predicting patients’ discontinuation from opioid use disorder treatment: retrospective observational study

Background Buprenorphine is a widely used treatment option for patients with opioid use disorder (OUD). Premature discontinuation from this treatment has many negative health and societal consequences. Objective To develop and evaluate a machine learning based two-stage clinical decision-making framework for predicting which patients will discontinue OUD treatment within less than a year. The proposed framework performs such prediction in two stages: (i) at the time of initiating the treatment, and (ii) after two/three months following treatment initiation. Methods For this retrospective observational analysis, we utilized Massachusetts All Payer Claims Data (MA APCD) from the year 2013 to 2015. Study sample included 5190 patients who were commercially insured, initiated buprenorphine treatment between January and December 2014, and did not have any buprenorphine prescription at least one year prior to the date of treatment initiation in 2014. Treatment discontinuation was defined as at least two consecutive months without a prescription for buprenorphine. Six machine learning models (i.e., logistic regression, decision tree, random forest, extreme-gradient boosting, support vector machine, and artificial neural network) were tested using a five-fold cross validation on the input data. The first-stage models used patients’ demographic information. The second-stage models included information on medication adherence during the early phase of treatment based on the proportion of days covered (PDC) measure. Results A substantial percentage of patients (48.7%) who started on buprenorphine discontinued the treatment within one year. The area under receiving operating characteristic curve (C-statistic) for the first stage models varied within a range of 0.55 to 0.59. The inclusion of knowledge regarding patients’ adherence at the early treatment phase in terms of two-months and three-months PDC resulted in a statistically significant increase in the models’ discriminative power (p-value < 0.001) based on the C-statistic. We also constructed interpretable decision classification rules using the decision tree model. Conclusion Machine learning models can predict which patients are most at-risk of premature treatment discontinuation with reasonable discriminative power. The proposed machine learning framework can be used as a tool to help inform a clinical decision support system following further validation. This can potentially help prescribers allocate limited healthcare resources optimally among different groups of patients based on their vulnerability to treatment discontinuation and design personalized support systems for improving patients’ long-term adherence to OUD treatment.