Applied deep learning and data science with a human-centric and data-centric approach

The term Artificial intelligence acquired a lot of baggage since its introduction and in its current incarnation is synonymous with Deep Learning. The sudden availability of data and computing resources has opened the gates to myriads of applications. Not all are created equal though, and problems might arise especially for fields not closely related to the tasks that pertain tech companies that spearheaded DL. The perspective of practitioners seems to be changing, however. Human-Centric AI emerged in the last few years as a new way of thinking DL and AI applications from the ground up, with a special attention at their relationship with humans. The goal is designing a system that can gracefully integrate in already established workflows, as in many real-world scenarios AI may not be good enough to completely replace its humans. Often this replacement may even be unneeded or undesirable. Another important perspective comes from, Andrew Ng, a DL pioneer, who recently started shifting the focus of development from “better models” towards better, and smaller, data. He defined his approach Data-Centric AI. Without downplaying the importance of pushing the state of the art in DL, we must recognize that if the goal is creating a tool for humans to use, more raw performance may not align with more utility for the final user. A Human-Centric approach is compatible with a Data-Centric one, and we find that the two overlap nicely when human expertise is used as the driving force behind data quality. This thesis documents a series of case-studies where these approaches were employed, to different extents, to guide the design and implementation of intelligent systems. We found human expertise proved crucial in improving datasets and models. The last chapter includes a slight deviation, with studies on the pandemic, still preserving the human and data centric perspective.

Deep learning and embeddings for problems of computational biology

The development of Next Generation Sequencing promotes Biology in the Big Data era. The ever-increasing gap between proteins with known sequences and those with a complete functional annotation requires computational methods for automatic structure and functional annotation. My research has been focusing on proteins and led so far to the development of three novel tools, DeepREx, E-SNPs&GO and ISPRED-SEQ, based on Machine and Deep Learning approaches. DeepREx computes the solvent exposure of residues in a protein chain. This problem is relevant for the definition of structural constraints regarding the possible folding of the protein. DeepREx exploits Long Short-Term Memory layers to capture residue-level interactions between positions distant in the sequence, achieving state-of-the-art performances. With DeepRex, I conducted a large-scale analysis investigating the relationship between solvent exposure of a residue and its probability to be pathogenic upon mutation. E-SNPs&GO predicts the pathogenicity of a Single Residue Variation. Variations occurring on a protein sequence can have different effects, possibly leading to the onset of diseases. E-SNPs&GO exploits protein embeddings generated by two novel Protein Language Models (PLMs), as well as a new way of representing functional information coming from the Gene Ontology. The method achieves state-of-the-art performances and is extremely time-efficient when compared to traditional approaches. ISPRED-SEQ predicts the presence of Protein-Protein Interaction sites in a protein sequence. Knowing how a protein interacts with other molecules is crucial for accurate functional characterization. ISPRED-SEQ exploits a convolutional layer to parse local context after embedding the protein sequence with two novel PLMs, greatly surpassing the current state-of-the-art. All methods are published in international journals and are available as user-friendly web servers. They have been developed keeping in mind standard guidelines for FAIRness (FAIR: Findable, Accessible, Interoperable, Reusable) and are integrated into the public collection of tools provided by ELIXIR, the European infrastructure for Bioinformatics.

Deep learning and spatial transcriptomics to investigate thyroid tumors

There are many diseases that affect the thyroid gland, and among them are carcinoma. Thyroid cancer is the most common endocrine neoplasm and the second most frequent cancer in the 0-49 age group. This thesis deals with two studies I conducted during my PhD. The first concerns the development of a Deep Learning model to be able to assist the pathologist in screening of thyroid cytology smears. This tool created in collaboration with Prof. Diciotti, affiliated with the DEI-UNIBO "Guglielmo Marconi" Department of Electrical Energy and Information Engineering, has an important clinical implication in that it allows patients to be stratified between those who should undergo surgery and those who should not. The second concerns the application of spatial transcriptomics on well-differentiated thyroid carcinomas to better understand their invasion mechanisms and thus to better comprehend which genes may be involved in the proliferation of these tumors. This project specifically was made possible through a fruitful collaboration with the Gustave Roussy Institute in Paris. Studying thyroid carcinoma deeply is essential to improve patient care, increase survival rates, and enhance the overall understanding of this prevalent cancer. It can lead to more effective prevention, early detection, and treatment strategies that benefit both patients and the healthcare system.

Cation size effects-modified phase and PTCR development in Er3+ and Ca2+ co-doped BaTiO3 ceramics during sintering

Development of the positive temperature coefficient of resistivity (PTCR) in Er3+ and Ca2+ co-doped ferroelectric BaTiO3 was studied in this work, with Er3+ being used to act as a donor doping. Irrespective of all the materials showing high densities after sintering at 1200 to 1300 ºC, these revealed insulator at the lowest sintering temperature, changing to semiconducting and PTCR-type materials only when the sintering temperature was further increased. Observations from X-ray diffraction help correlating this effect with phase development in this formulated (Ba,Ca,Er)TiO3 system, considering the formation of initially two separated major (Ba,Ca)TiO3- and minor (Ca,Er)TiO3-based compounds, as a consequence of cation size-induced stress energy effects. Thus, appearance and enhancement here of the semiconducting and PTCR responses towards higher sintering temperatures particularly involve the incorporation of Er3+ into the major phase, rendering finally possible the generation and "percolative-like" migration of electrons throughout the whole material.

Neurobehavioral, reflexological and physical development of Wistar rat offspring exposed to ayahuasca during pregnancy and lactation

Ayahuasca is a hallucinogenic beverage prepared by the decoction of plants native to the Amazon Basin region. The beverage has been used throughout the world by members of some syncretic religious movements. Despite the recent legalization of ayahuasca in Brazil for religious purposes, there is little pre-clinical and clinical information attesting to its safety, particularly in relation to the use during pregnancy. The aim of the current work was to determine the effects of perinatal exposure to ayahuasca (from the 6th day of pregnancy to the 10th day of lactation) on physical, reflexology and neurobehavioral parameters of the Wistar rat offspring. The offspring showed no statistically significant changes in the physical and reflexology parameters evaluated. However, in adult rats, perinatally exposed to ayahuasca, an increase in frequency of entries in open arms in elevated plus-maze test, a decrease in total time of interaction in social interaction test, a decrease in time of latency for the animal to start swimming and a decrease of the minimum convulsant dose induced by pentylenetetrazol were observed. In conclusion, our results showed that the use of ayahuasca by mothers during pregnancy and lactation reduced the general anxiety and social motivation of the rat offspring. Besides, it promoted a higher sensitivity for initiation and spread of seizure activity.

Embryonic, Larval, and Juvenile Development of the Sea Biscuit Clypeaster subdepressus (Echinodermata: Clypeasteroida)

Sea biscuits and sand dollars diverged from other irregular echinoids approximately 55 million years ago and rapidly dispersed to oceans worldwide. A series of morphological changes were associated with the occupation of sand beds such as flattening of the body, shortening of primary spines, multiplication of podia, and retention of the lantern of Aristotle into adulthood. To investigate the developmental basis of such morphological changes we documented the ontogeny of Clypeaster subdepressus. We obtained gametes from adult specimens by KCl injection and raised the embryos at 26 degrees C. Ciliated blastulae hatched 7.5 h after sperm entry. During gastrulation the archenteron elongated continuously while ectodermal red-pigmented cells migrated synchronously to the apical plate. Pluteus larvae began to feed in 3 d and were similar to 20 d old at metamorphosis; starved larvae died 17 d after fertilization. Postlarval juveniles had neither mouth nor anus nor plates on the aboral side, except for the remnants of larval spicules, but their bilateral symmetry became evident after the resorption of larval tissues. Ossicles of the lantern were present and organized in 5 groups. Each group had 1 tooth, 2 demipyramids, and 2 epiphyses with a rotula in between. Early appendages consisted of 15 spines, 15 podia (2 types), and 5 sphaeridia. Podial types were distributed in accordance to Loven's rule and the first podium of each ambulacrum was not encircled by the skeleton. Seven days after metamorphosis juveniles began to feed by rasping sand grains with the lantern. Juveniles survived in laboratory cultures for similar to 9 months and died with <500 mu m wide, a single open sphaeridium per ambulacrum, aboral anus, and no differentiated food grooves or petaloids. Tracking the morphogenesis of early juveniles is a necessary step to elucidate the developmental mechanisms of echinoid growth and important groundwork to clarify homologies between irregular urchins.

A review of time-motion analysis and combat development in mixed martial arts matches at regional level tournaments

Mixed martial arts (MMA) have become a fast-growing worldwide expansion of martial arts competition, requiring high level of skill, physical conditioning, and strategy, and involving a synthesis of combat while standing or on the ground. This study quantified the effort-pause ratio (EP), and classified effort segments of stand-up or groundwork development to identify the number of actions performed per round in MMA matches. 52 MMA athletes participated in the study (M age = 24 yr., SD = 5; average experience in MMA = 5 yr., SD = 3). A one-way analysis of variance with repeated measurements was conducted to compare the type of action across the rounds. A chi-squared test was applied across the percentages to compare proportions of different events. Only one significant difference (p < .05) was observed among rounds: time in groundwork of low intensity was longer in the second compared to the third round. When the interval between rounds was not considered, the EP ratio (between high-intensity effort to low-intensity effort plus pauses) WE S 1:2 to 1:4. This ratio is between ratios typical for judo, wrestling, karate, and taekwondo and reflects the combination of ground and standup techniques. Most of the matches ended in the third round, involving high-intensity actions, predominantly executed during groundwork combat.

Collaborative e-learning and Learning Objects

The Learning Object (OA) is any digital resource that can be reused to support learning with specific functions and objectives. The OA specifications are commonly offered in SCORM model without considering activities in groups. This deficiency was overcome by the solution presented in this paper. This work specified OA for e-learning activities in groups based on SCORM model. This solution allows the creation of dynamic objects which include content and software resources for the collaborative learning processes. That results in a generalization of the OA definition, and in a contribution with e-learning specifications.

On Building a Memory Evolutive System for Application to Learning and Cognition Modeling

We address here aspects of the implementation of a memory evolutive system (MES), based on the model proposed by A. Ehresmann and J. Vanbremeersch (2007), by means of a simulated network of spiking neurons with time dependent plasticity. We point out the advantages and challenges of applying category theory for the representation of cognition, by using the MES architecture. Then we discuss the issues concerning the minimum requirements that an artificial neural network (ANN) should fulfill in order that it would be capable of expressing the categories and mappings between them, underlying the MES. We conclude that a pulsed ANN based on Izhikevich`s formal neuron with STDP (spike time-dependent plasticity) has sufficient dynamical properties to achieve these requirements, provided it can cope with the topological requirements. Finally, we present some perspectives of future research concerning the proposed ANN topology.

Effects of 7-Epiclusianone on Streptococcus mutans and Caries Development in Rats

The aim of this study was to evaluate the effects of 7-epiclusianone (7-epi) on specific virulence attributes of Streptococcus mutans in vitro and on development of dental caries in vivo. 7-Epi was obtained and purified from fruits of Rheedia brasiliensis. We investigated its influence on surface-adsorbed glucosyltransferase (Gtf) B activity, acid production, and viability of S. mutans in biofilms, as well as on caries development using a rodent model. 7-Epi (100 mu g/mL) significantly reduced the activity of surface-adsorbed GtfB (up to 48.0 +/- 1.8 of inhibition at 100 mu g/mL) and glyco-lytic pH-drop by S. mutans in biofilms (125 and 250 mu g/mL) (vs. vehicle control, p < 0.05). In contrast, the test compound did not significantly affect the bacterial viability when compared to vehicle control (15% ethanol, p > 0.05). Wistar rats treated topically with 7-epi (twice daily, 60-s exposure) showed significantly smaller number of and less severe smooth-and sulcal-surface carious lesions (p < 0.05), without reducing the S. mutans viable population from the animals` dental biofilms. In conclusion, the natural compound 7-epiclusianone may be a potentially novel pharmacological agent to prevent and control dental caries disease.

Calcium: magnesium ratio in amendments of soil acidity: nutrition and initial development of corn plants in a Humic Alic Cambisol

The variation in the Ca:Mg ratio in amendments used to neutralize soil acidity is one way of altering the availability of those nutrients to the plants in acid soils. The objective of the work was to evaluate the effect of different proportions of calcium and magnesium in the form of CaCO(3) and MgCO(3) Oil the nutrient uptake, and initial production of dry matter by corn plants. The study was carried out in greenhouse conditions, in Lages, SC, with a completely randomized experimental design, with three replications. The treatments were the application of equivalent to 21.0 t ha(-1) of lime, using mixtures of CaCO(3) and MgCO(3) in several proportions to obtain different Ca:Mg ratios (1: 1, 2:1, 4:1, 8:1, 16:1 and 32:1), on a Humic Alic Cambisol, with 310 g kg(-1) of clay. The application of treatments caused the following Ca:Mg ratios in the CEC: 1. 1: 1, 2.1:1, 4.0:1, 8.1:1, 16.4:1 and 31.8:1. The high concentrations of exchangeable Ca in soil caused by addition of lime with high Ca content inhibited the uptake of Mg and K by the corn plants. The increase in the soil Ca:Mg ratio reduced the dry matter production and height of plants in the initial stage of development.

Effects of isoflavone-supplemented soy yogurt on lipid parameters and atherosclerosis development in hypercholesterolemic rabbits: a randomized double-blind study

Background: There is increasing interest in natural treatments to control dyslipidemia and reduce the risk of cardiovascular disease. Previous studies have demonstrated the beneficial effects of soy yogurt fermented with Enterococcus faecium CRL 183 and of dietary isoflavones on the lipid profile. The purpose of the present study was to investigate the effects of isoflavone-supplemented soy yogurt, fermented with E. faecium CRL183, on lipid parameters and atherosclerosis development in rabbits with induced hypercholesterolemia. Methods: Forty-eight rabbits were randomly assigned to eight groups fed on the following diets for 60 days: C - control; IY - isoflavone-supplemented soy yogurt; H - hypercholesterolemic (1.0% cholesterol wt/wt diet); HY - hypercholesterolemic plus soy yogurt; HIY - hypercholesterolemic plus isoflavone-supplemented soy yogurt; HP - hypercholesterolemic plus placebo; HI hypercholesterolemic plus isoflavone and HE - hypercholesterolemic plus pure culture of E. faecium CRL 183. Serum lipids and autoantibodies against oxLDL (oxLDL Ab) were analyzed on days 0, 30 and 60 of the treatment and the atherosclerotic lesions were quantified at the end of the experiment. Results: Soy yogurt, soy yogurt supplemented with isoflavones and placebo promoted significant reductions in total cholesterol level (38.1%, 27.0% and 26.6%, respectively). Significant increases in serum HDL-C concentration relative to group H were detected in animals that ingested soy yogurt, with or without the isoflavone supplement (55.2%), E. faecium culture (43.3%) or placebo (35.8%). Intake of soy yogurt and soy yogurt supplemented with isoflavones prevented the rise of oxLDL Ab during the study period. The extent of atherosclerosis in the thoracic and abdominal aortas was reduced in the HIY, HY and HP groups. However, when the whole aorta was analyzed, animals treated with soy yogurt supplemented with isoflavones exhibited the greatest reduction (51.4%, P < 0.05) in atherosclerotic lesion area, compared to group H. Conclusion: Soy yogurt could be consumed as an alternative means of reducing the risk of cardiovascular disease by improving the lipid profile and inhibiting oxLDL Ab formation. Our findings also suggest that isoflavone supplementation may enhance the antiatherosclerotic effect of soy yogurt.

Inflorescence and floral development of Dahlstedtia species (Leguminosae: Papilionoideae: Millettieae)

Inflorescence and floral development of two tropical legume trees, Dahlstedtia pinnata and Dahlstedlia pentaphylla, occurring in the Atlantic Forest of south-eastern and southern Brazil, were investigated and compared with other papilionoids. Few studies have been made of floral development in tribe Millettieae, and this paper is intended to fill that gap in our knowledge. Dahlstedtia species have an unusual inflorescence type among legumes, the pseudoraceme, which comprises axillary units of three or more flowers, each with a subtending bract. Each flower exhibits a pair of opposite bractcoles. The order of flower initiation is acropetal; inception of the floral organs is as follows: sepals (5), petals (5), carpel (1) plus outer stamens (5) and finally inner stamens (5). Organ initiation in sepal, petal and inner stamen whorls is unidirectional; the carpel cleft is adaxial. The vexillum originates from a tubular-shaped primordium in mid-development and is larger than other petals at maturity, covering the keels. The filament tube develops later after initiation of inner-stamen primordia. Floral development in Dahlstedtia is almost always similar to other papilionoids, especially species of Phaseoleae and Sophoreae. But one important difference is the precocious ovule initiation (open carpel with ovules) in Dahlstedtia, the third citation of this phenomenon for papilionoids. No suppression, organ loss or anomalies occur in the order of primordia initiation or structure. Infra-generic differences in the first stages of ontogeny are rare; however, different species of Dahlstedtia are distinguished by the differing distribution pattern of secretory cavities in the flower. (C) 2009 Elsevier GmbH. All rights reserved.