Jointly Predicting Future Content in Multiple Social Media Sites Based on Multi-task Learning

User-generated contents (UGC) in social media are the direct expression of users’ interests, preferences, and opinions. User behavior prediction based on UGC has increasingly been investigated in recent years. Compared to learning a person’s behavioral patterns in each social media site separately, jointly predicting user behavior in multiple social media sites and complementing each other (cross-site user behavior prediction) can be more accurate. However, cross-site user behavior prediction based on UGC is a challenging task due to the difficulty of cross-site data sampling, the complexity of UGC modeling, and uncertainty of knowledge sharing among different sites. For these problems, we propose a Cross-Site Multi-Task (CSMT) learning method to jointly predict user behavior in multiple social media sites. CSMT mainly derives from the hierarchical attention network and multi-task learning. Using this method, the UGC in each social media site can obtain fine-grained representations in terms of words, topics, posts, hashtags, and time slices as well as the relevances among them, and prediction tasks in different social media sites can be jointly implemented and complement each other. By utilizing two cross-site datasets sampled from Weibo, Douban, Facebook, and Twitter, we validate our method’s superiority on several classification metrics compared with existing related methods.

Analisis Keamanan Sistem Informasi Akademik (SIAKAD) Universitas XYZ Menggunakan Metode Vulnerability Assessment

The use of academic information systems (siakad) has become mandatory for universities in providing user convenience in online academic administrative activities. However, sometimes college siakad has security holes that irresponsible people can take advantage of by hacking. This study aims to identify security vulnerabilities at XYZ Siakad University. The method used in this study is a vulnerability assessment method. A university syakad will conduct an initial vulnerability assessment by doing footprinting to get information related to XYZ syakad after that a vulnerability scan is carried out using vulnerability assessment tools to identify vulnerabilities and the level of risk found. Based on the vulnerability of the XYZ university's vulnerabilities, it is quite good, with a high risk level of 1, a medium risk level of 6 and a low risk level of 14. Researchers provide recommendations for improvements related to the findings of security holes in XYZ university Siakad from XSS (Cross Site Scripting) attacks, Clickjacking, Brute Force, Cross-site Request Forgery (CSRF) and Sniffing.

STRUMM: An Intelligent, Adaptive Software Package for Older Adults With a Cognitive Impairment

Many older adults have a cognitive impairment (CI), which negatively impacts on their quality of life and threatens their independence. In this presentation, we provide an overview of the conceptual framework, structure, and processes of our multi-site Center, ENHANCE, which is focused on developing technology support for aging adults with a CI. ENHANCE has two cross-site research projects, two cross-site development projects, training, and dissemination components. A core battery of measures is collected across all projects. We also discuss the Supportive Technology Resources through Usability & Machine-learning Methods (STRUMM) research project, which focuses on an innovative intelligent adaptive software package aimed at providing cognitive and social support, and support for resource access to aging adults with a CI. STRUMM is designed to meet the user’s varying cognitive needs. Finally, we present preliminary data regarding the perceived usability and value of STRUMM from our clinical partners and potential user groups.

The impact of Juvenile Drug Treatment Courts (JDTC) implementing Federal Evidence-Based Guidelines on recidivism and substance use: multisite Randomized Controlled Trial (RCT) and Regression Discontinuity (RDD) Designs

Background Juvenile drug treatment courts (JDTC) have struggled to define themselves since their inception in 1995. Early courts followed a format similar to adult drug courts, but these did not address the unique needs of juveniles, which led to the creation of 16 Strategies by a consensus panel of practitioners and researchers. But, like the first JDTCs, research with courts following these strategies failed to provide convincing evidence that this “model” was associated with significant reductions in recidivism or drug use. More recently, a new set of evidence-based guidelines were developed through meta-analyses commissioned by the Office of Juvenile Justice and Delinquency Prevention (OJJDP, 2016). Method OJJDP provided funding for a rigorous multi-site evaluation of the guidelines. This study protocol paper for the Juvenile Drug Treatment Court (JDTC) Guidelines Cross-Site Evaluation presents research designs for the comparison of youth outcomes from 10 JDTCs compared with 10 Traditional Juvenile Courts (TJCs) in the same jurisdictions. Two sites opted into a randomized controlled trial (RCT) and eight sites chose to follow a regression discontinuity design (RDD). Youth data are captured at baseline, and at 6- and 12-month follow-ups using an interview comprised of several standardized assessments. The youths’ official records also are abstracted for recidivism and substance use information. The degree to which the evidence-based guidelines are implemented at each site is assessed via an in-depth court self-assessment collected at baseline and again 2 years later and via structured site visits conducted once during implementation. Discussion As a field-based trial, using both RCT and RDD designs, findings will provide important, policy-relevant information regarding the implementation of the OJJDP evidence-based guidelines, including the degree to which JDTCs adopted and/or modified these practices, their relative impact on recidivism and substance use, as well as the degree to which JDTCs differ from TJCs. Specific inferences may be drawn about whether following or not following specific guidelines differentially impact youth outcomes, yielding recommendations about the translation of this information from research-to-practice for potentiating the broader adoption of these guidelines by JDTCs nationwide. Clinical trials registration This was not an NIH supported trial. The funder, OJJDP/NIJ, instead required publishing the design with even more information at https://www.ojp.gov/ncjrs/virtual-library/abstracts/juvenile-drug-treatment-court-jdtc-guidelines-cross-site-evaluation.

Using OPM-MEG in contrasting magnetic environments

Magnetoencephalography (MEG) has been revolutionised in recent years by optically pumped magnetometers (OPMs). OPM-MEG offers higher sensitivity, better spatial resolution and lower cost than conventional instrumentation based on superconducting quantum interference devices (SQUIDS). Moreover, OPMs offer the possibility of motion robustness and lifespan compliance, dramatically expanding the range of MEG applications. However, OPM-MEG remains nascent technology; it places stringent requirements on magnetic shielding, and whilst a number of viable systems exist, most are custom made and there have been no cross-site investigations showing the reliability of data. In this paper, we undertake the first cross-site OPM-MEG comparison, using near identical commercial systems scanning the same participant. The two sites are deliberately contrasting, with different magnetic environments: a green field campus university site with an OPM-optimised shielded room (low interference) and a city centre hospital site with a standard (non-optimised) MSR (high interference). We show that despite a 25-fold difference in background field, and a 30-fold difference in low frequency interference, using dynamic field control and software-based suppression of interference we can generate comparable noise floors at both sites. In human data recorded during a visuo-motor task and a face processing paradigm, we were able to generate similar data, with source localisation showing that brain regions could be pinpointed with just ~10 mm spatial discrepancy and temporal correlations of > 80%. Overall, our study demonstrates that plug-and-play OPM-MEG systems exist and can be sited even in challenging magnetic environments.

51 A novel cross-site analysis of Vectra® Polaris™ multiplex fluorescence PD-1/PD-L1 immunohistochemistry on colorectal cancer with high and low microsatellite instability

BackgroundColorectal cancer (CRC) is the third most diagnosed cancer in the United States with a projected 52,980 deaths in 2021.1 Microsatellite instability-high (MSI-H) CRCs with deficiencies in mismatch repair (MMR) are significantly associated with positive response to immunotherapy and improved outcomes when treated with immune checkpoint inhibitors. Programmed cell death ligand-1 (PD-L1) is an effective biomarker of MSI-H status to identify CRC patients who will respond to treatment, however, reproducible quantification of programmed cell death receptor-1 (PD-1)/PD-L1 in the tumor microenvironment (TME) across laboratory sites has been under-reported.2–3 In this study, our group directly addressed this issue by interrogating PD-1/PD-L1 cross-site at Akoya Biosciences and NeoGenomics Laboratories by employing the MOTiF™ PD-1/PD-L1 Panel kit along with the Vectra Polaris imaging system.MethodsSerial sections from 40 CRC samples with known MSI status were stained at Akoya and NeoGenomics Laboratories using a modified MOTiF PD-1/PD-L1 Lung Panel Kit on the Leica BOND RX. Sections were scanned using the Vectra Polaris imaging system at both sites. Inter-site staining reproducibility was assessed using image analysis algorithms developed with inForm tissue analysis software. Cell counts and densities were calculated using the R-script package PhenoptrReports and correlations were plotted per marker.ResultsThe average signal intensity for all markers/Opal fluorophores was within the recommended ranges of 10–30 normalized counts, with the exception of Polaris 780, which has an advised range of 1–10. This indicates the protocol stained successfully and reproducibly across all serial sections at both sites. Inter-site concordance analysis of cell densities for each marker yielded an average R2 value of ≥0.70. H-Score of PD-L1 quantified at the cell membrane trended with MSI status (stable/high).ConclusionsThis study demonstrated that the MOTiF PD-1/PD-L1 Panel kit imaged in conjunction with the Vectra Polaris is not only a reliable assay that can be run across different sites, based on the concordant cross-site data, but that re-optimization of the kit allows for the assay panel to be successfully adapted to other cancers, such as CRC, which can then capture biological differences across a multitude of samples.ReferencesAmerican Cancer Society https://www.cancer.org/cancer/colon-rectal-cancer/about/key-statistics.htmlYi M, Jiao D, Xu H, Liu Q, Zhao W, Han X, et al. Biomarkers for predicting efficacy of PD-1/PD-L1 inhibitors. Mol Cancer 2018;17(1):129Lemery S, Keegan P, Pazdur R. First FDA approval agnostic of cancer site - when a biomarker defines the indication. N Engl J Med 2017;377(15):1409–12.