Unsupervised machine learning for risk stratification and identification of relevant subgroups of ascending aorta dimensions using cardiac CT and clinical data

The potential of precision population health lies in its capacity to utilize robust patient data for customized prevention and care targeted at specific groups. Machine learning has the potential to automatically identify clinically relevant subgroups of individuals, considering heterogeneous data sources. This study aimed to assess whether unsupervised machine learning (UML) techniques could interpret different clinical data to uncover clinically significant subgroups of patients suspected of coronary artery disease and identify different ranges of aorta dimensions in the different identified subgroups. We employed a random forest-based cluster analysis, utilizing 14 variables from 1170 (717 men/453 women) participants. The unsupervised clustering approach successfully identified four distinct subgroups of individuals with specific clinical characteristics, and this allows us to interpret and assess different ranges of aorta dimensions for each cluster. By employing flexible UML algorithms, we can effectively process heterogeneous patient data and gain deeper insights into clinical interpretation and risk assessment.

Psychophysiological strategies for enhancing performance through imagery-skin conductance level analysis in guided vs. self-produced imagery

Athletes need to achieve their optimal level of arousal for peak performance. Visualization or mental rehearsal (i.e., Imagery) often helps to obtain an appropriate level of activation, which can be detected by monitoring Skin Conductance Level (SCL). However, different types of imagery could elicit different amount of physiological arousal. Therefore, this study aims: (1) to investigate differences in SCL associated with two instructional modalities of imagery (guided vs. self-produced) and six different scripts; (2) to check if SCL could differentiate respondents according to their sport expertise. Thirty participants, aged between 14 and 42 years (M = 22.93; SD = 5.24), with different sport levels took part in the study. Participants listened to each previously recorded script and then were asked to imagine the scene for a minute. During the task, SCL was monitored. We analysed the mean value, variance, slope and number of fluctuations per minute of the electrodermal signal. Unsupervised machine learning models were used for measuring the resemblance of the signal. The Wilcoxon signed-rank test was used for distinguishing guided and self-produced imagery, and The Mann-Whitney U test was used for distinguishing results of different level athletes. We discovered that among others, self-produced imagery generates lower SCL, higher variance, and a higher number of fluctuations compared to guided imagery. Moreover, we found similarities of the SCL signal among the groups of athletes (i.e. expertise level). From a practical point of view, our findings suggest that different imagery instructional modalities can be implemented for specific purposes of mental preparation.

Clusterization of co-morbidities and multi-morbidities among persons living with HIV: a cross-sectional study

BACKGROUND

Among people living with HIV (PLWH), the prevalence of non-HIV related co-morbidities is increasing. Aim of the present study is to describe co-morbidity and multi-morbidity, their clustering mode and the potential disease-disease interactions in a cohort of Italian HIV patients.

METHODS

Cross-sectional analysis conducted by the Coordinamento Italiano per lo Studio di Allergia e Infezioni da HIV (CISAI) on adult subjects attending HIV-outpatient facilities. Non-HIV co-morbidities included: cardiovascular disease, diabetes mellitus, hypertension, oncologic diseases, osteoporosis, probable case of chronic obstructive pulmonary disease (COPD), hepatitis C virus (HCV) infection, psychiatric illness, kidney disease. Multi-morbidity was defined as the presence of two or more co-morbidities.

RESULTS

One thousand and eighty-seven patients were enrolled in the study (mean age 47.9 ± 10.8). One hundred-ninety patients (17.5%) had no co-morbidity, whereas 285 (26.2%) had one condition and 612 (56.3%) were multi-morbid. The most recurrent associations were: 1) dyslipidemia + hypertension (237, 21.8%); 2) dyslipidemia + COPD (188, 17.3%); 3) COPD + HCV-Ab+ (141, 12.9%). Multi-morbidity was associated with older age, higher body mass index, current and former smoking, CDC stage C and longer ART duration.

CONCLUSIONS

More than 50% of PLHW were multi-morbid and about 30% had three or more concurrent comorbidities. The identification of common patterns of comorbidities address the combined risks of multiple drug and disease-disease interactions.

Evolutionary framework of the human interactome: Unicellular and multicellular giant clusters

The main contradiction of multicellularity (MCM) is between the unicellular (UC) and multicellular (MC) levels. In human interactome we revealed two giant clusters with MC and UC medians (and several smaller ones with MC medians). The enrichment of these clusters by phylostrata and by functions support the MC versus UC division. The total interactome and the giant clusters show a core-periphery evolutionary growth. From viewpoint of the MCM, the most important is the placement of genes, appearing at UC evolutionary stage, in the MC clusters. Thus, genes involved in vesicle-mediated transport, cell cycle, cellular responses to stress, post-translational modifications and many diseases appeared at UC evolutionary stage but are placed mostly in MC clusters. Genes downregulated with age are enriched in UC cluster, whereas the upregulated genes are preferentially placed in MC giant cluster. The tumor suppressor and pluripotency regulating pathways are also enriched in MC giant cluster. Therefore, this cluster probably operates as 'internal manager' constraining runaway unicellularity. The clusters have denser interactions within than between them, therefore they can serve as attractors (stable states of dynamic systems) of cellular programs. Importantly, the UC cluster have a higher inside/outside connection ratio compared with MC clusters, which suggests a stronger attractor effect and may explain why cells of MC organisms are prone to oncogenesis. The evolutionary clustering of human interactome elucidates the MC control over functions appearing at UC evolutionary stage and can build a framework for biosystems studies focusing on the interplay between UC and MC levels.

Kinetic model of Chlorella vulgaris growth with and without extremely low frequency-electromagnetic fields (EM-ELF)

Chlorella vulgaris was grown in two bench-scale photobioreactors with and without the application of a low intensity, low frequency electromagnetic field (EM-ELF) of about 3mT. Cell concentration and tendency of cells to form aggregates inside the reactor were recorded over a 30 days-time period at 0.5L-constant medium volume in the temperature range 289-304K. At 304K, after a cultivation period of 15 days, the rate of cell death became predominant over that of growth. In the temperature range 289-299K, a two step-kinetic model based on the mitotic division and the clusterization processes was developed and critically discussed. The best-fitted curves turned out to have a sigmoid shape, and the competition between mitosis and clusterization was investigated. Without EM-ELF, the temperature dependence of the specific rate constant of the mitotic step yielded an apparent total enthalpy of 15±6kJmol(-1), whose value was not influenced by the EM-ELF application. The electromagnetic field was shown to exert a significant effect on the exothermic clusterization step. The heat exchange due to binding between cells and liquid medium turned out to be -44±5kJmol(-1) in the absence of EM-ELF and -68±8kJmol(-1) when it was active. Optical microscopy observations were in agreement with the model predictions and confirmed that EM-ELF was able to enhance cell clusterization.

The Importance of Descriptor-Based Clusterization in QSAR Models Development: Tyrosine Kinases Inhibitors as a Key Study

Quantitative Structure Activity Relationship (QSAR) is a well known cheminformatic tool for the discovery of novel biologically active compounds. However, when large and heterogeneous datasets are mined, it is not possible to derive a QSAR equation able to predict in a satisfactory manner the activity of the compounds. Thus, QSAR models are often inadequate for virtual screening purpose. Herein we present a novel approach to multitarget classification QSAR models, useful to assess the selectivity profile of the tyrosine kinases inhibitors. A descriptor-based clusterization process was employed, that allowed the generation of models with high accuracies and independent from the chemical classification of the compounds (i.e. from the scaffold type). The herein proposed methodology can lead to QSAR models useful for virtual screening processes.