67 research outputs found
Enforcing Fairness via Constraint Injection with FaUCI
No full textThe problem of fairness in AI can be tackled by minimising bias in the data (pre-processing), in the algorithms (in-processing), or in the results (post-processing). In the particular case of in-processing applied to supervised machine learning, state-of-the-art solutions rely on a few well-known fairness metrics – e.g., demographic parity, disparate impact, or equalised odds – optimised during training—which, however, mostly focus on binary attributes and their effects on binary classification problems. Accordingly, in this work we propose FaUCI as a general purpose framework for injecting fairness constraints into neural networks (or, any model trained via stochastic gradient descent), supporting attributes of many sorts—there including binary, discrete, or continuous features. To evaluate its effectiveness and efficiency, we test FaUCI against several sorts of features and fairness metrics. Furthermore, we compare FaUCI with state-of-the-art solutions for in-processing, demonstrating its superiority
Ovulation induction in young girls with menometrorragia: a safe and effective treatment
No full textThe prevalence of menometrorrhagia in fertile women is 11.4-13.2% and increases with aging. The presence of metrorrhagia is a relatively common cause of concern among adolescents and their parents, as well as a frequent cause of visits to emergency departments, gynaecologists, and pediatricians. Clomiphene is a selective estrogen receptor modulator (SERM) that increases the production of gonadotropins by inhibiting negative feedback on the hypothalamus.
Clomiphene is primarily used in the treatment of female infertility for ovulation induction to reverse oligoovulation or anovulation, as occurs in polycystic ovary syndrome.
Objective: The aim of our study was to evaluate the use of clomiphene citrate in ovulation induction, and therefore, in the cessation of bleeding in adolescents with menometrorrhagia in the absence of uterine, ovarian, or systemic pathologies.
Design: Cohort study.
Materials and methods: The study group was comprised of 50 subjects (age range, 13-16 years) with menometrorrhagia (bleeding >7 days in length with an average blood loss >80 ml). The treatment with clomiphene citrate at a dose of 50 mg/day for 5 days during the attack cycle, and from days 3 to 7 of three subsequent cycles, was offered to the patients.
Results: After three cycles of therapy, all patients had resolution of the menometrorrhagia and resumption of ovulatory cycles. No patient reported unwanted side effects.
Conclusion: We propose that low-dose clomiphene should be the first step in the treatment of adolescent dysfunctional bleeding (DUB)
Obesity related lipid profileand altered insulin incretion in adolescent with policystic ovary syndrome
No full textNon
KINS: Knowledge Injection via Network Structuring
No full textWe propose a novel method to inject symbolic knowledge in form of Datalog formulæ into neural networks (NN), called KINS (Knowledge Injection via Network Structuring). The idea behind our method is to extend NN internal structure with ad-hoc layers built out the injected symbolic knowledge. KINS does not constrain NN to any specific architecture, neither requires logic formulæ to be ground. Moreover, it is robust w.r.t. both lack of data and imperfect/incomplete knowledge. Experiments are reported to demonstrate the potential of KINS
Il-6 Serum Levels and Production Is Related to an Altered Immune Response in Polycystic Ovary Syndrome Girls with Insulin Resistance
No full textPolycystic ovarian syndrome (PCOS) is frequently characterized by obesity and metabolic diseases including hypertension, insulin resistance, and diabetes in adulthood, all leading to an increased risk of atherosclerosis. The present study aimed to evaluate serum and production of inflammatory markers in adolescent Sardinian PCOS. On the basis of HOMA findings, patients were divided into noninsulin resistant (NIR) and insulin resistant (IR), and were weight- and age-matched with healthy girls. Inflammatory cytokines (TNF-α, IL-6, Il-10, TGF-β) and lipokines (leptin, adiponectin), the reactant hs-CRP, and in vitro inflammatory lympho-monocyte response to microbial stimulus were evaluated. In healthy and PCOS subjects, leptin and hs-CRP were correlated with BMI, whereas adiponectin was significantly reduced in all PCOS groups. Although cytokines were similar in all groups, Interleukin-6 (IL-6) was significantly higher in IR PCOS. Moreover, in the latter group lipopolysaccharide-activated monocytes secreted significantly higher levels of IL-6 compared to NIR and control subjects. To conclude, IR PCOS displayed increased IL-6 serum levels and higher secretion in LPS-activated monocytes, whilst revealing no differences for other inflammatory cytokines. These results suggest that in PCOS patients an altered immune response to inflammatory stimuli is present in IR, likely contributing towards determining onset of a low grade inflammation
Symbolic knowledge extraction for explainable nutritional recommenders
No full textBackground and objective:This paper focuses on nutritional recommendation systems (RS), i.e. AI-powered automatic systems providing users with suggestions about what to eat to pursue their weight/body shape goals. A trade-off among (potentially) conflictual requirements must be taken into account when designing these kinds of systems, there including: (i) adherence to experts’ prescriptions, (ii) adherence to users’ tastes and preferences, (iii) explainability of the whole recommendation process. Accordingly, in this paper we propose a novel approach to the engineering of nutritional RS, combining machine learning and symbolic knowledge extraction to profile users—hence harmonising the aforementioned requirements. MethodsOur contribution focuses on the data processing workflow. Stemming from neural networks (NN) trained to predict user preferences, we use CART Breiman et al.(1984) to extract symbolic rules in Prolog Körner et al.(2022) form, and we combine them with expert prescriptions brought in similar form. We can then query the resulting symbolic knowledge base via logic solvers, to draw explainable recommendations. ResultsExperiments are performed involving a publicly available dataset of 45,723 recipes, plus 12 synthetic datasets about as many imaginary users, and 6 experts’ prescriptions. Fully-connected 4-layered NN are trained on those datasets, reaching ∼86% test-set accuracy, on average. Extracted rules, in turn, have ∼80% fidelity w.r.t. those NN. The resulting recommendation system has a test-set precision of ∼74%. The symbolic approach makes it possible to devise how the system draws recommendations. ConclusionsThanks to our approach, intelligent agents may learn users’ preferences from data, convert them into symbolic form, and extend them with experts’ goal-directed prescriptions. The resulting recommendations are then simultaneously acceptable for the end user and adequate under a nutritional perspective, while the whole process of recommendation generation is made explainable.Interactive Intelligenc
The quantitative insulin sensitivity check index is not able to detect early metabolic alterations in young patients with polycystic ovarian syndrome
No full textObjective. To verify whether QUICKY is a suitable method for the identification of metabolic deterioration in normal weight patients affected by polycystic ovarian syndrome (PCOS). Design. Prospective clinical study. Patient(s). Seventy-nine PCOS normal weight adolescent subjects, 50 eumenorrheic, normal weight, non-hirsute controls matched for age and BMI. Method(s). Quantitative insulin sensitivity check index (QUICKY) and integrated secretory area under the curve of insulin values (I-AUC) during oral glucose tolerance test were calculated. Result(s). Seventy-nine PCOS and 50 controls were studied. Normal insulin sensitivity was defined as upper control 95th percentile by QUICKY values <0.31, I-AUC at 180 min < 16,645. When applying the calculated I-AUC cut-off, 41 PCOS were classified as normoinsulinemic and 38 as hyperinsulinemic, whereas using the calculated QUICKY cut-off, only 19 PCOS could be classified as insulin resistant (IR). Fifteen out of the 60 non-IR PCOS presented hyperinsulinemia; fasting glucose and insulin levels and QUICKY were not sufficient to identify these subjects. Thus, QUICKY displayed a low sensitivity (44%) and specificity (91%) in the diagnosis of the metabolic disorder disclosed by I-AUC. Conclusions. In young normal weight patients with PCOS the prevalence of early alterations of insulin metabolism are not detectable by QUICKY studies
Conventional transbronchial needle aspiration for the staging of lung cancer
No full textConventional transbronchial needle aspiration for the staging of lung cance
A Computational Approach to Poetic Structure, Rhythm and Rhyme
No full textIn this paper we present SPARSAR, a system for the automatic analysis of English and Italian poetry. The system can work on any type of poem and produces a set of parameters that are then used to compare poems with one another, of the same author or of different authors. In this paper, we will concentrate on the second module, which is a rule-based system to represent and analyze poetic devices. Evaluation of the system on the basis of a manually created dataset - including poets from Shakespeare's time down to T.S.Eliot and Sylvia Plath - has shown its high precision and accuracy approximating 90%
- …
