Oral Sessions

Growth in chronic diseases and epidemic viruses, combined with an aging population, urges the need for innovative, smart technologies to control the health condition of our indoor environment. Indoor environmental conditions play a significant role in protecting occupants’ health and well-being. The thermal condition is one of the primary factors of Indoor Environmental Qualities (IEQ) that can cause Sick Building Syndrome (SBS) and influence occupants’ health. Schedule-based and predefined environmental control is one of the main reasons for the current discomfort and dissatisfaction. These general standards make it impossible to consider people’s differences in thermal sensation. Recent research attempts to leverage occupants’ demand in the control loop of the buildings to consider the well-being of each individual based on their own physiological properties, which is called "personalized comfort models." In this study, we will develop personal comfort models by utilizing thermal cameras and a completely non-contact and non-intrusive approach. Elevated core body temperature is an indicator of heat stress. Thermal imaging as a technology can be used to specify the skin temperature. Recent research tries to utilize skin temperature and infrared thermal images for developing comfort models, but they have several limitations and inaccuracies. To capitalize on the potential and address the existing limitations, new solutions that take a more holistic approach to non-intrusive thermal scanning and health monitoring are required, leveraging the benefits of real-time artificial intelligence, multi-sensor fusion, and video data analytics. This project uses the recent advances in artificial intelligence (AI) to create smart, health-conscious environments that can assess and model the overall thermal health of individuals from a distance without the need for direct person-to-person contact. Our research primarily focuses on thermal comfort assessment, as a major reason for unhealthy buildings, and combines it with other environmental factors to (1) enhance the accuracy of facial skin temperature assessment, and (2) integrate multimodal sensing systems. In contrast to existing research, which has limited data quality, this project proposes a holistic solution to enable scalable, reliable body temperature assessment from a distance and in different head positions. To create prediction algorithms, we have gathered a facial thermography dataset from ten healthy adult individuals. The dataset includes more than 10,000 thermal images of the same subjects in various thermal and physical conditions, including the environmental temperature, distance from the camera, head position, and their reported subjective thermal comfort. To assess the efficacy of machine learning algorithms for predicting thermal comfort, we trained three previously proven algorithms for predicting personalized thermal comfort from the literature. Random Forest, Support Vector Machine (SVM), and K-Nearest Neighbor (KNN) were trained and tested with each subject's personalized physiological and subjective data. It was found that although the prediction performance of images at closer distances was superior to those at larger distances, by increasing the number of data frames, we could obtain very high accuracy predictions from lower quality images. Another finding of this study is the better performance of the Random Forest and KNN prediction algorithms.

Genomic instability is a key driving force for cancer development and age-related disease. While DNA damage responses are known to regulate genome integrity and cell fates, there is mounting evidence that genomic instability may trigger inflammatory and/or antitumor responses via pattern recognition receptors (PRRs) located in the cytosol. One such PRR, cyclic GMP-AMP synthase (cGAS), canonically detects foreign viral or bacterial DNA resulting from infection, initiating a signaling cascade resulting in the production and secretion of Type I interferons. However, we hypothesize that cGAS contributes to the initiation of innate immune responses following DNA damage outside of its canonical role of cytosolic DNA recognition in the central nervous system (CNS). In the present study, we utilized CRISPR/Cas9 technology to efficiently knock down cGAS expression in a human microglial cell line then induced DNA damage using ionizing radiation and an oxidative stress inducing system. We then measured the production of key inflammatory cytokines and chemokines via enzyme-linked immunosorbent assays (ELISAs). We found that our wild-type microglia showed robust production and secretion of interleukin 8, while in our cGAS knock-down microglia, this response was significantly diminished at 24 hours post treatment. Taken together, our data supports the contention that cGAS plays a critical role in the initiation of proinflammatory immune responses following DNA damage outside of its canonical role in antiviral immunity.

Deception cues are a particular form of indicators that hint a statement is not true or a person is lying. To equip people with deception cues, training is a promising solution. In this regard, this study aims to investigate the effect of training on online deception detection in the context of online phishing attacks, assuming that training is to improve people’s deception detection ability. Additionally, this study examines the relationship between email topic familiarity and detection accuracy. We expect one’s familiarity with email topics to be a primary factor that affects his/her deception detection performance. Thus, we have the following four hypotheses. H1. Verbal-cue-based training will improve the deception detection accuracy of trainees. H2. Familiarity with the communication topic is positively associated with online deception detection accuracy. H3-1. Detectors’ confidence will improve online deception detection accuracy. H3-2. Detectors’ self-efficacy will improve online deception detection accuracy. The findings provide insights on how to design effective training to improve online deception detection performance. The results suggest a promising approach to mitigating the victimization of online deception through the proper design of verbal-cue-based training.

Plasmodium vivax malaria is a neglected tropical disease, despite being more geographically widespread than any other form of malaria. P. vivax primarily utilizes the Duffy Binding Protein (DBP) for red blood cell invasion. Documentation of P. vivax infections in different parts of Africa where Duffy-negative individuals are dominant suggested that there are alternative pathways for P. vivax to invade human RBCs. After P. vivax enters a reticulocyte, the parasite feeds on the host hemoglobin to grow from an initial ring to a trophozoite. As it further develops into the schizont stage, the parasite begins to asexually reproduce forming 8-12 merozoites inside the host RBC and express invasion-specific transcripts. We have successfully obtained 10 P. vivax transcriptomes from in vitro culture for the identification of schizont-specific transcripts. However, mixed parasite stages among samples and a lack of reliable deconvolution methods as well as reference model impede the identification of these transcripts. Previous studies used P. falciparum homologues to infer P. vivax stage-specific transcripts and do not accurately reflect the gene expression profile of P. vivax. Thus, this study utilizes a reference-free deconvolution method Linseed that uses linear dependency of stage-specific reference genes to determine clusters of parasite stage-specific reads, which allows an unbiased identification of schizont-specific transcripts. Our ongoing work compares the outcomes of different reference and non-reference modeling methods using both the Ethiopian and Cambodian datasets, with the goal to provide a more accurate bioinformatic pipeline and identify top 20 expressed genes in each life cycle.

Based on the 2017 report by WHO Global Antimicrobial Surveillance system, antibiotic resistant bacteria (ARB) have become a worldwide challenge with growing need to be tackled. Current approaches to eliminate ARBs include antibiotic cocktails, engineered bacteriophages, monoclonal antibodies and antimicrobial peptides aimed at targeting multiple pathways in bacteria. However, the use of antibiotic cocktails is limited by the increased risk of developing resistance, whereas antibodies, bacteriophages and peptides are susceptible to proteolytic degradation compromising their antimicrobial efficacy. A nanoparticle (NP)-based alternative has been recently explored as a promising solution to overcome the current challenges to eliminate ARBs. Silver nanoparticles (AgNPs) have gained a lot of attention for this purpose in the last few years. AgNP’s antibacterial property is associated with the leaching of silver ion (Ag+), which leads to bacterial cell membrane disruption affording the killing of bacteria. However, the use of AgNPs is limited due to the slow release of Ag+. In this study, a photosensitizer (protoporphyrin (PpIX)) conjugated on the surface of AgNPs enhances Ag+ release leading to their increased antibacterial effect. The generation of singlet oxygen under light irradiation provides a highly oxidizing environment for AgNPs that lead to enhanced Ag+ release. PpIX molecule was chemically attached to AgNPs using cysteamine as the linker. The final product (cysPpIX-AgNPs) was characterized using UV-vis spectroscopy, FT-IR, DLS and Zeta-potential. cysPpIX-AgNP were electrostatically coated with polyethyleneimine (PEI) to render a positive surface charge to increase bacterial interaction (PEI-cysPpIX-AgNP). ICP-OES was used to evaluate the Ag+ release kinetic in water and DPBS under light irradiation. This synergistic effect was evaluated in two antibiotic resistant bacterial strains: Methicillin resistant Staphylococcus aureus (MRSA) and wild type Escherichia coli (MDR E. coli) resistant to ampicillin, sulfamethoxazole, ciprofloxacin, and tetracycline. The antibacterial effect of AgNP, cysPpIX-AgNP and PEI-cysPpIX-AgNP to eliminate ARB was demonstrated in DPBS using the colony count method. Among the three nanoparticles, cysPpIX-AgNP achieved the highest Ag+ release in DPBS with ~ 575 ug/L, almost 3 times higher than AgNP in DPBS. The conjugation of cysPpIX allow AgNPs to release Ag+ with ~75% efficiency. In the presence of light, cysPpIX-AgNP achieved >6 log inactivation of MRSA and >7 log inactivation of MDR E. coli at a low NP concentration (1.5ug/mL or 1uM). Bacterial inactivation correlates to the Ag+ release in DPBS pointing to the important role of silver ions in killing ARBs. All NPs showed negligible cytotoxicity to mammalian cells at tested concentration of 1.5 µg/mL (in terms of AgNP) or 1uM (in terms of cysPpIX) after 20mins light exposure and 24 hours incubation. This study provides new evidence to support the accelerated conversion of AgNP to Ag+ using light and PpIX and demonstrates their promising role in wound healing by killing ARBs at concentrations that are harmless to mammalian cells.