Deep Neural Network Regression to Assist Non-Invasive Diagnosis of Portal Hypertension

Portal hypertension is a complex medical condition characterized by elevated blood pressure in the portal venous system. The conventional diagnosis of such disease often involves invasive procedures such as liver biopsy, endoscopy, or imaging techniques with contrast agents, which can be uncomfortable for patients and carry inherent risks. This study presents a deep neural network method in support of the non-invasive diagnosis of portal hypertension in patients with chronic liver diseases. The proposed method utilizes readily available clinical data, thus eliminating the need for invasive procedures. A dataset composed of standard laboratory parameters is used to train and validate the deep neural network regressor. The experimental results exhibit reasonable performance in distinguishing patients with portal hypertension from healthy individuals. Such performances may be improved by using larger datasets of high quality. These findings suggest that deep neural networks can serve as useful auxiliary diagnostic tools, aiding healthcare professionals in making timely and accurate decisions for patients suspected of having portal hypertension.

A machine-learning based bio-psycho-social model for the prediction of non-obstructive and obstructive coronary artery disease

BACKGROUND

Mechanisms of myocardial ischemia in obstructive and non-obstructive coronary artery disease (CAD), and the interplay between clinical, functional, biological and psycho-social features, are still far to be fully elucidated.

OBJECTIVES

To develop a machine-learning (ML) model for the supervised prediction of obstructive versus non-obstructive CAD.

METHODS

From the EVA study, we analysed adults hospitalized for IHD undergoing conventional coronary angiography (CCA). Non-obstructive CAD was defined by a stenosis < 50% in one or more vessels. Baseline clinical and psycho-socio-cultural characteristics were used for computing a Rockwood and Mitnitski frailty index, and a gender score according to GENESIS-PRAXY methodology. Serum concentration of inflammatory cytokines was measured with a multiplex flow cytometry assay. Through an XGBoost classifier combined with an explainable artificial intelligence tool (SHAP), we identified the most influential features in discriminating obstructive versus non-obstructive CAD.

RESULTS

Among the overall EVA cohort (n = 509), 311 individuals (mean age 67 ± 11 years, 38% females; 67% obstructive CAD) with complete data were analysed. The ML-based model (83% accuracy and 87% precision) showed that while obstructive CAD was associated with higher frailty index, older age and a cytokine signature characterized by IL-1β, IL-12p70 and IL-33, non-obstructive CAD was associated with a higher gender score (i.e., social characteristics traditionally ascribed to women) and with a cytokine signature characterized by IL-18, IL-8, IL-23.

CONCLUSIONS

Integrating clinical, biological, and psycho-social features, we have optimized a sex- and gender-unbiased model that discriminates obstructive and non-obstructive CAD. Further mechanistic studies will shed light on the biological plausibility of these associations.

CLINICAL TRIAL REGISTRATION

NCT02737982.

The Use of Machine Learning for Inferencing the Effectiveness of a Rehabilitation Program for Orthopedic and Neurological Patients

Advance assessment of the potential functional improvement of patients undergoing a rehabilitation program is crucial in developing precision medicine tools and patient-oriented rehabilitation programs, as well as in better allocating resources in hospitals. In this work, we propose a novel approach to this problem using machine learning algorithms focused on assessing the modified Barthel index (mBI) as an indicator of functional ability. We build four tree-based ensemble machine learning models and train them on a private training cohort of orthopedic (OP) and neurological (NP) hospital discharges. Moreover, we evaluate the models using a validation set for each category of patients using root mean squared error (RMSE) as an absolute error indicator between the predicted mBI and the actual values. The best results obtained from the study are an RMSE of 6.58 for OP patients and 8.66 for NP patients, which shows the potential of artificial intelligence in predicting the functional improvement of patients undergoing rehabilitation.

Supervised and unsupervised learning to classify scoliosis and healthy subjects based on non-invasive rasterstereography analysis

The aim of our study was to classify scoliosis compared to to healthy patients using non-invasive surface acquisition via Video-raster-stereography, without prior knowledge of radiographic data. Data acquisitions were made using Rasterstereography; unsupervised learning was adopted for clustering and supervised learning was used for prediction model Support Vector Machine and Deep Network architectures were compared. A M-fold cross validation procedure was performed to evaluate the results. The accuracy and balanced accuracy of the best supervised model were close to 85%. Classification rates by class were measured using the confusion matrix, giving a low percentage of unclassified patients. Rasterstereography has turned out to be a good tool to distinguish subject with scoliosis from healthy patients limiting the exposure to unnecessary radiations.

MOSES: A New Approach to Integrate Interactome Topology and Functional Features for Disease Gene Prediction

Disease gene prediction is to date one of the main computational challenges of precision medicine. It is still uncertain if disease genes have unique functional properties that distinguish them from other non-disease genes or, from a network perspective, if they are located randomly in the interactome or show specific patterns in the network topology. In this study, we propose a new method for disease gene prediction based on the use of biological knowledge-bases (gene-disease associations, genes functional annotations, etc.) and interactome network topology. The proposed algorithm called MOSES is based on the definition of two somewhat opposing sets of genes both disease-specific from different perspectives: warm seeds (i.e., disease genes obtained from databases) and cold seeds (genes far from the disease genes on the interactome and not involved in their biological functions). The application of MOSES to a set of 40 diseases showed that the suggested putative disease genes are significantly enriched in their reference disease. Reassuringly, known and predicted disease genes together, tend to form a connected network module on the human interactome, mitigating the scattered distribution of disease genes which is probably due to both the paucity of disease-gene associations and the incompleteness of the interactome.

A machine-learning-based bio-psycho-social model for the prediction of non-obstructive and obstructive coronary artery disease

Background

Although cardiovascular disease is the leading cause of mortality in both females and males, women are more likely to have non-obstructive ischemic heart disease (IHD) than men. However, the underlying sex- and gender-specific mechanisms and differences in IHD manifestations are still not fully understood.

Aim

To develop an interpretable machine learning (ML) model to gain insight on the clinical, functional, biological and psychosocial features playing a major role in the supervised prediction of non-obstructive versus obstructive CAD.

Methods

From the EVA study, we analyzed a consecutive unselected cohort of adults hospitalized for IHD undergoing coronary angiography. Non-obstructive CAD was defined by a coronary stenosis at the angiogram &lt;50%. Baseline clinical and psycho-socio-cultural characteristics were used for computing a frailty index based on Rockwood and Mitnitsky model, and gender score according to GENESIS-PRAXY methodology. The serum concentration of inflammatory cytokines was measured with a multiplex flow cytometric assay. An XGBoost classifier combined to an explainable artificial intelligence tool (SHAP) was employed to identify the most influential features in discriminating obstructive versus non-obstructive CAD.

Results

Among the overall EVA cohort (n=509), 311 individuals (mean age 67±11 years, 38% females; 67% obstructive CAD) with complete data were analyzed. The ML-based model (83% accuracy and 87% precision) revealed that while obstructive CAD associated with higher frailty index (i.e., lower physiological reserve), older age and a cytokine signature characterized by IL-1β, IL-12p70 and IL-33, non-obstructive CAD was more likely associated with higher gender score (i.e., social characteristics traditionally ascribed to women, regardless of biological sex) and with a cytokine signature characterized by IL-18, IL-8, IL-23.

Conclusions

Integrating clinical, biological and psycho-social features, we have optimized a sex- and gender-unbiased model that discriminates obstructive and non-obstructive CAD. Further mechanistic studies will shed light on the biological plausibility of the observed associations.

Funding Acknowledgement

Type of funding sources: Public Institution(s). Main funding source(s): Italian Ministry of Education, Research and University, Scientific Independence of young Researcher (SIR)

An Integer Black-Box Optimization Model for Repairable Spare Parts Management

Spare parts management affects significantly costs and service level for supply chains. This paper deals with an inventory management problem for multi-item repairable systems via a systemic perspective based on a new efficient integer black-box optimization model. With respect to the traditionally used marginal allocation that considers items individually, the proposed black-box optimization model is a holistic approach in the fact that it exploits relationships among items. The authors propose a derivative-free algorithm specifically tied to the application which exploits a new selection strategy for choosing entire subsets of items with the aim to get the best expected improvement in the objective function. The approach has been tested on a real case study for optimizing stocks in an airline's inventory network. The case study provides evidence about the good behavior of the exploratory geometry of the proposed approach in finding quickly a feasible and optimal solution for inventory control.

Machine Learning Use for Prognostic Purposes in Multiple Sclerosis

The course of multiple sclerosis begins with a relapsing-remitting phase, which evolves into a secondarily progressive form over an extremely variable period, depending on many factors, each with a subtle influence. To date, no prognostic factors or risk score have been validated to predict disease course in single individuals. This is increasingly frustrating, since several treatments can prevent relapses and slow progression, even for a long time, although the possible adverse effects are relevant, in particular for the more effective drugs. An early prediction of disease course would allow differentiation of the treatment based on the expected aggressiveness of the disease, reserving high-impact therapies for patients at greater risk. To increase prognostic capacity, approaches based on machine learning (ML) algorithms are being attempted, given the failure of other approaches. Here we review recent studies that have used clinical data, alone or with other types of data, to derive prognostic models. Several algorithms that have been used and compared are described. Although no study has proposed a clinically usable model, knowledge is building up and in the future strong tools are likely to emerge.

Considering patient clinical history impacts performance of machine learning models in predicting course of multiple sclerosis

Multiple Sclerosis (MS) progresses at an unpredictable rate, but predictions on the disease course in each patient would be extremely useful to tailor therapy to the individual needs. We explore different machine learning (ML) approaches to predict whether a patient will shift from the initial Relapsing-Remitting (RR) to the Secondary Progressive (SP) form of the disease, using only "real world" data available in clinical routine. The clinical records of 1624 outpatients (207 in the SP phase) attending the MS service of Sant'Andrea hospital, Rome, Italy, were used. Predictions at 180, 360 or 720 days from the last visit were obtained considering either the data of the last available visit (Visit-Oriented setting), comparing four classical ML methods (Random Forest, Support Vector Machine, K-Nearest Neighbours and AdaBoost) or the whole clinical history of each patient (History-Oriented setting), using a Recurrent Neural Network model, specifically designed for historical data. Missing values were handled by removing either all clinical records presenting at least one missing parameter (Feature-saving approach) or the 3 clinical parameters which contained missing values (Record-saving approach). The performances of the classifiers were rated using common indicators, such as Recall (or Sensitivity) and Precision (or Positive predictive value). In the visit-oriented setting, the Record-saving approach yielded Recall values from 70% to 100%, but low Precision (5% to 10%), which however increased to 50% when considering only predictions for which the model returned a probability above a given "confidence threshold". For the History-oriented setting, both indicators increased as prediction time lengthened, reaching values of 67% (Recall) and 42% (Precision) at 720 days. We show how "real world" data can be effectively used to forecast the evolution of MS, leading to high Recall values and propose innovative approaches to improve Precision towards clinically useful values.

Data of patients undergoing rehabilitation programs

In this data article, we present a dataset made up of personal, social and clinical records related to patients undergoing a rehabilitation program. Data refers to records registered in the “Acceptance/Discharge Report for the rehabilitation area” (ADR) which implements the Italian law (DGR 731/2005) and refer to hospitalization at the rehabilitation hospital of Rome “San Raffaele” in the years from 2015 to 2018 of patients suffering from orthopedic and neurological pathologies. For each ADR report, the clinical status of the patient at the date of acceptance and discharge is reported using, among other, the Barthel index as a measure of the Activities Daily Living of the patient.

These data can be used to understand the influence of many different factors in the rehabilitation progress of clinical patients.

A convergent hybrid decomposition algorithm model for SVM training

Training of support vector machines (SVMs) requires to solve a linearly constrained convex quadratic problem. In real applications, the number of training data may be very huge and the Hessian matrix cannot be stored. In order to take into account this issue, a common strategy consists in using decomposition algorithms which at each iteration operate only on a small subset of variables, usually referred to as the working set. Training time can be significantly reduced by using a caching technique that allocates some memory space to store the columns of the Hessian matrix corresponding to the variables recently updated. The convergence properties of a decomposition method can be guaranteed by means of a suitable selection of the working set and this can limit the possibility of exploiting the information stored in the cache. We propose a general hybrid algorithm model which combines the capability of producing a globally convergent sequence of points with a flexible use of the information in the cache. As an example of a specific realization of the general hybrid model, we describe an algorithm based on a particular strategy for exploiting the information deriving from a caching technique. We report the results of computational experiments performed by simple implementations of this algorithm. The numerical results point out the potentiality of the approach.

Factors affecting insufficiency in activity daily living in the elderly

BACKGROUND

In a "case-control" study we investigated the correlations among twenty-four clinical signs of "functional impairment" and probability of "activity daily living insufficiency".

METHODS

The study involved 788 randomised inpatients, aged 65 years and over, of nineteen long-stay hospitals of an Italian region (Lazio, Rome). We measured self care autonomy, mobility and continence, on a modified Barthel's scale; the score on Barthel's scale, Barthel Index (BI), was correlated to twenty-four signs of "functional impairment" (explicative variables). Of these variables entered in stepwise regression only "cognitive impairment" (coef. B-22), "paralysis" (coef. B-21), "body weight reduction over 10 kg vs ideal weight" (coef. B-12), "joint deformation" (coef. B-7) and "visual impairment" (coef. B-5). Insufficiency in daily living is defined by BI < 100. The presence of these five clinical signs leads to the likelihood of "activity daily living insufficiency" to 0.996. The trend of cognitive impairment to rise with age could be responsible for the inverse regression between age and BI.

RESULTS

There was no significant correlation between BI and sex. Hearing impairment, serum creatinine level > or = 4 mg/dl, bronchospasm, obstructive and restrictive ventilation disorders, precordial pain on stress or spontaneous and dyspnea are not significantly correlated to the Barthel Index Score and to the likelihood of insufficiency in daily living activity.

Evidence that a salt bridge in the light chain contributes to the physical stability difference between heavy and light human ferritins

Human ferritin, a multimeric iron storage protein, is composed by various proportions of two subunit types: the H- and L-chains. The biological functions of these two genic products have not been clarified, although differences in reactivity with iron have been shown. Starting from the hypothesis that the high stability typical of ferritin is an important property which may be relevant for its iron storage function, we studied ferritin homopolymers of H- and L-chains in different denaturing conditions. In addition we analyzed 13 H-chain variants with alterations in regions conserved within mammalian H-chains. In all the denaturation experiments H-chain ferritin showed lower stability than L-chain ferritin. The difference was greater in guanidine HCl denaturation experiments, where the end products are fully unfolded peptides, than in acidic denaturation experiments, where the end products are peptides with properties analogous to "molten globule." The study on H-chain variants showed: (i) ferritin stability was not affected by alterations of regions exposed to the inner or outer surface of the shell and not involved in intra- or inter-chain interactions; (ii) stability was reduced by alterations of sequences involved in inter-subunit interactions such as the deletion of the N-terminal extension or substitutions along the hydrophobic and hydrophilic channels; (iii) stability was increased by the substitution of 2 amino acids inside the four-helix bundle with those of the homologous L-chain. One of the residues is involved in a salt bridge in the L-chain, and we concluded that the stability difference between H- and L-ferritins is to a large extent due to the stabilizing effect of this salt bridge on the L-subunit fold.

[Arterial pressure and heart rate: postural variations in the elderly]

The behaviour of arterial blood pressure and heart rate after an active change from the supine to the upright posture was studied in the aged. A significant decrease in the average values for systolic and diastolic blood pressure and a significant increase in the average values for heart rate in the first minute of the upright posture were observed. The systolic and diastolic pressures are reduced in a higher percentage of the cases in the 80-94 year age class. In the same class the PAS is more frequently reduced in the male and the PAD in the female. The heart rate, in the upright posture, is higher in 80-94 year-old males without a significant difference between the sexes.

[Vectorcardiographic study of an unusual form of hypertrophic cardiomyopathy]

The many aspects of heart electrical activity in hypertrophic cardiomyopathy have recently been described in numerous occasions. A vectorcardiographic study was made of two sisters with an echocardiographic diagnosis of hypertrophy of the free wall of the left ventricle. Both the ECG and the VCG shown evidence of pseudonecrosis and pseudoischaemia associated with left intraventricular conduction disturbances and left atrial dilatation. It is felt that echocardiography, and the two-dimensional examination in particular, is essential in the diagnosis of myocardial hypertrophy, but it is pointed out that ECG, and above all VCG, may sometimes indicate its site and extent with a good degree of approximation.