Designing user-centered decision support systems for climate disasters: What information do communities and rescue responders need during floods?

Flooding events are the most common natural hazard globally, resulting in vast destruction and loss of life. An effective flood emergency response is necessary to lessen the negative impacts of flood disasters. However, disaster management and response efforts face a complex scenario. Simultaneously, regular citizens attempt to navigate the various sources of information being distributed and determine their best course of action. One thing is evident across all disaster scenarios: having accurate information and clear communication between citizens and rescue personnel is critical. This research aims to identify the diverse needs of two groups, rescue operators and citizens, during flood disaster events by investigating the sources and types of information they rely on and information that would improve their responses in the future. This information can improve the design and implementation of existing and future spatial decision support systems (SDSSs) during flooding events. This research identifies information characteristics crucial for rescue operators and everyday citizens' response and possible evacuation to flooding events by qualitatively coding survey responses from rescue responders and the public. The results show that including local input in SDSS development is crucial for improving higher-resolution flood risk quantification models. Doing so democratizes data collection and analysis, creates transparency and trust between people and governments, and leads to transformative solutions for the broader scientific community.

MetaQA: Enhancing human-centered data search using Generative Pre-trained Transformer (GPT) language model and artificial intelligence.

Accessing and utilizing geospatial data from various sources is essential for developing scientific research to address complex scientific and societal challenges that require interdisciplinary knowledge. The traditional keyword-based geosearch approach is insufficient due to the uncertainty inherent within spatial information and how it is presented in the data-sharing platform. For instance, the Gulf of Mexico Coastal Ocean Observing System (GCOOS) data search platform stores geoinformation and metadata in a complex tabular. Users can search for data by entering keywords or selecting data from a drop-down manual from the user interface. However, the search results provide limited information about the data product, where detailed descriptions, potential use, and relationship with other data products are still missing. Language models (LMs) have demonstrated great potential in tasks like question answering, sentiment analysis, text classification, and machine translation. However, they struggle when dealing with metadata represented in tabular format. To overcome these challenges, we developed Meta Question Answering System (MetaQA), a novel spatial data search model. MetaQA integrates end-to-end AI models with a generative pre-trained transformer (GPT) to enhance geosearch services. Using GCOOS metadata as a case study, we tested the effectiveness of MetaQA. The results revealed that MetaQA outperforms state-of-the-art question-answering models in handling tabular metadata, underlining its potential for user-inspired geosearch services.

Development of a Multi-target Protein Biomarker Assay for Circulating Tumor Cells.

We report a highly sensitive and selective CNT-switch liquid biopsy platform that detects and quantifies protein biomarker expressions from circulating tumor cells in blood for early detection of metastatic breast cancer and its relapse. This platform first isolates and enriches more than 99% of tumor cells with an off-chip micro-size membrane filtration technique and then conducts on-chip detection of the membrane and internal protein biomarkers of the tumor cells with high sensitivity and selectivity. High sensitivity is achieved with complete association of the antibody-antigen-antibody (Ab-Ag-Ab) complex by precisely and rapidly assembling carbon nanotubes (CNTs) across two parallel electrodes via sequential DC electrophoresis and dielectrophoresis (DEP) deposition. Each bridged CNT acts as a switch that connects the electrodes and closes the circuit to generate an electrical signal. The high selectivity is achieved with a critical hydrodynamic shear rate that irreversibly removes non-target linkers of the aligned CNTs. At present, we are able to detect the protein biomarkers from 5 spiked breast cancer tumor cells of different types within 7.5 ml of human blood samples. This demonstrates the potential of this platform as an inexpensive and noninvasive alternative to MRI scans and tissue biopsies currently used to detect early metastatic breast cancer and its relapse.

Modeling Spatiotemporal Pattern of Depressive Symptoms Caused by COVID-19 Using Social Media Data Mining.

By 29 May 2020, the coronavirus disease (COVID-19) caused by SARS-CoV-2 had spread to 188 countries, infecting more than 5.9 million people, and causing 361,249 deaths. Governments issued travel restrictions, gatherings of institutions were cancelled, and citizens were ordered to socially distance themselves in an effort to limit the spread of the virus. Fear of being infected by the virus and panic over job losses and missed education opportunities have increased people's stress levels. Psychological studies using traditional surveys are time-consuming and contain cognitive and sampling biases, and therefore cannot be used to build large datasets for a real-time depression analysis. In this article, we propose a CorExQ9 algorithm that integrates a Correlation Explanation (CorEx) learning algorithm and clinical Patient Health Questionnaire (PHQ) lexicon to detect COVID-19 related stress symptoms at a spatiotemporal scale in the United States. The proposed algorithm overcomes the common limitations of traditional topic detection models and minimizes the ambiguity that is caused by human interventions in social media data mining. The results show a strong correlation between stress symptoms and the number of increased COVID-19 cases for major U.S. cities such as Chicago, San Francisco, Seattle, New York, and Miami. The results also show that people's risk perception is sensitive to the release of COVID-19 related public news and media messages. Between January and March, fear of infection and unpredictability of the virus caused widespread panic and people began stockpiling supplies, but later in April, concerns shifted as financial worries in western and eastern coastal areas of the U.S. left people uncertain of the long-term effects of COVID-19 on their lives.

Towards perturbation prediction of biological networks using deep learning.

The mapping of the physical interactions between biochemical entities enables quantitative analysis of dynamic biological living systems. While developing a precise dynamical model on biological entity interaction is still challenging due to the limitation of kinetic parameter detection of the underlying biological system. This challenge promotes the needs of topology-based models to predict biochemical perturbation patterns. Pure topology-based model, however, is limited on the scale and heterogeneity of biological networks. Here we propose a learning based model that adopts graph convolutional networks to learn the implicit perturbation pattern factors and thus enhance the perturbation pattern prediction on the basic topology model. Our experimental studies on 87 biological models show an average of 73% accuracy on perturbation pattern prediction and outperforms the best topology-based model by 7%, indicating that the graph-driven neural network model is robust and beneficial for accurate prediction of the perturbation spread modeling and giving an inspiration of the implementation of the deep neural networks on biological network modeling.

A shear-enhanced CNT-assembly nanosensor platform for ultra-sensitive and selective protein detection.

Detection and quantification of low-concentration proteins in heterogeneous media are generally plagued by two distinct obstacles: lack of sensitivity due to high dissociation equilibrium constant KD and non-specificity due to an abundance of non-targets with similar KD. Herein, we report a nanoscale protein-sensing platform with a non-equilibrium on-off switch that employs dielectrophoretic and hydrodynamic shear forces to overcome these thermodynamic limitations with irreversible kinetics. The detection sensitivity is achieved with complete association of the antibody-antigen-antibody (Ab-Ag-Ab) complex by precisely and rapidly assembling carbon nanotubes (CNT) across two parallel electrodes via sequential DC electrophoresis and AC dielectrophoresis (DEP), and with single-CNT electron tunneling conductance. The high selectivity is achieved with a critical hydrodynamic shear rate between the activated dissociation shear rates of target and non-target linkers of the aligned CNTs. We are able to reach detection limits of 100 attomolar (aM) and 10 femtomolar (fM) in pure samples for two ELISA assays with low and high dissociation constant: biotin/streptavidin (10 fM) and HER2/HER2 antibody (0.44 ± 0.07nM), respectively. For both models, irreversible capture and shearing allow us to tune the dynamic range up to 5 decades by increasing the CNT numbers. We also demonstrate in spiked serum sample high selectivity towards target HER2 proteins against non-target HER2 isoform of a similar KD. The detection limit for HER2 in serum is lower than 100fM.

A comparison of atomistic and continuum approaches to the study of bonding dynamics in electrocatalysis: microcantilever stress and in situ EXAFS observations of platinum bond expansion due to oxygen adsorption during the oxygen reduction reaction.

Microcantilever stress measurements are examined to contrast and compare their attributes with those from in situ X-ray absorption spectroscopy to elucidate bonding dynamics during the oxygen reduction reaction (ORR) on a Pt catalyst. The present work explores multiple atomistic catalyst properties that notably include features of the Pt-Pt bonding and changes in bond strains that occur upon exposure to O2 in the electrochemical environment. The alteration of the Pt electronic and physical structures due to O2 exposure occurs over a wide potential range (1.2 to 0.4 V vs normal hydrogen electrode), a range spanning potentials where Pt catalyzes the ORR to those where Pt-oxide forms and all ORR activity ceases. We show that Pt-Pt surface bond strains due to oxygen interactions with Pt-Pt bonds are discernible at macroscopic scales in cantilever-based bending measurements of Pt thin films under O2 and Ar. Complementary extended X-ray absorption fine structure (EXAFS) measurements of nanoscale Pt clusters supported on carbon provide an estimate of the magnitude and direction of the in-operando bond strains. The data show that under O2 the M-M bonds elongate as compared to an N2 atmosphere across a broad range of potentials and ORR rates, an interfacial bond expansion that falls within a range of 0.23 (±0.15)% to 0.40 (±0.20)%. The EXAFS-measured Pt-Pt bond strains correspond to a stress thickness and magnitude that is well matched to the predictions of a mechanics mode applied to experimentally determined data obtained via the cantilever bending method. The data provide new quantitative understandings of bonding dynamics that will need to be considered in theoretical treatments of ORR catalysis and substantiate the subpicometer resolution of electrochemically mediated bond strains detected on the macroscale.

All-nanoparticle layer-by-layer surface modification of micro- and ultrafiltration membranes.

Layer-by-layer (LbL) deposition using primarily inorganic silica nanoparticles is employed for surface modification of polymeric micro- and ultrafiltration (MF/UF) membranes to produce novel thin film composite (TFC) membranes intended for nanofiltration (NF) and reverse osmosis (RO) applications. A wide variety of porous substrate membranes with different surface characteristics are successfully employed. This report gives detailed results for polycarbonate track etched (PCTE), polyethersulfone (PES), and sulfonated PES (SPEES) MF/UF substrates. Both spherical (cationic/anionic) and eccentric elongated (anionic) silica nanoparticles are deposited using conditions similar to those in prior works for solid substrates (e.g., Lee et al.). Appropriate selection of the pH for anionic and cationic particle deposition enables construction of nanoparticle-only layers 100-1200 nm in thickness atop the original porous membrane substrates. The surface layer thickness appears to vary linearly with the number of bilayers deposited, i.e., with the number of anionic/cationic deposition cycles. The deposition process is optimized to eliminate drying-induced cracking and improve mechanical durability via thickness control and postdeposition hydrothermal treatment. "Dead-end" permeation tests using dextran standards reveal the hydraulic characteristics and separations capability for the PCTE-based TFC membranes. The results show that nanoparticle-based LbL surface modification of MF and UF rated media can produce TFC membranes with NF capabilities.