Local and regional public health departments in the U.S. are legally required to make their services available in languages other than English. Most health departments outsource translation of health and safety information materials (e.g. websites and flyers on vaccinations, emergency preparedness, etc.). This is costly and time-consuming and makes it impossible for health departments to translate many important documents. We designed a web-based collaborative system to enable the production of multilingual health communication materials by bilingual public health workers. Our system could reduce the cost of translation and increase access to public health information for people in the U.S. with limited English proficiency. The system is based on a workflow where a source document in English is machine-translated and then revised and corrected by humans to produce a finalized translation. We designed the system for public health professionals who are bilingual domain experts but not necessarily trained translators. We initially gathered data to inform the design of our system through interviews and focus groups with local and regional public health departments. Based on the design recommendations extracted from the data, we implemented a web-based prototype collaborative translation management system. We further refined the system through an iterative design process that included informal user testing with multilingual participants. Future work will include usability studies with public health workers and the integration of additional collaborative features.

 A new non-visual method of numeric entry into a smartphone is designed, implemented, and tested. Users tap the smartphone screen with one to three fingers or swipe the screen in order to enter numbers. No buttons are used—only simple, easy-to-remember gestures. A preliminary evaluation compares the method to a standard accessible numeric keyboard with a VoiceOver-like screen reader interface. Preliminary results indicate that users can enter numbers faster and with higher accuracy when using this method. The Tapulator, a complete calculator based on this non-visual numeric entry that uses simple gestures for arithmetic operations and other calculator actions is described.

Pharmacogenetics is the study of how variations in a person’s DNA affect their response to drugs. It is the foundation of ‘personalized medicine’ and could one day enable drug therapies for individual patients based on genetic composition. Glucocorticoids are a group of cholesterol-based signaling molecules including endogenous hormones such as cortisol, which the body produces naturally in response to various forms of stress, and commonly prescribed anti-inflammatory drugs, which are used to treat diseases ranging from asthma and cancer to autoimmune disorders and psoriasis. For a glucocorticoid to produce its physiologic effects in any cell of the body, it must first bind to the glucocorticoid receptor (GR) and translocate with the GR to the cell nucleus. Both of these biologic processes are dependent on the molecular chaperone protein hsp90. A necessary first step in glucocorticoid-based pharmacogenetics research is better understanding the genetic variation associated with alterations in endogenous cortisol signaling. Our goal was to identify correlations between individuals’ variations in endogenous cortisol signaling and underlying genetic variability. Salivary cortisol (CORT) concentrations were measured in healthy volunteers (n=66) before and after CORT stimulation to determine variability in circulating CORT levels and CORT response. Statistical analysis revealed that inter-participant CORT variability in concentration or percent change was not exclusively correlated to participant demographics, health history, or GR single nucleotide polymorphisms (SNPs). Interestingly, nuclear localization of hsp90 correlated with an elevated pre-stimulus CORT concentration and a decreased post-activity CORT concentration. To investigate the observed CORT response and hsp90 localization association further, coding regions of the hsp90 gene were sequenced to identify possible genetic linkages. Eight hsp90 SNPs were detected, four previously unreported. All participants expressed between two and five of these SNPs. These data demonstrate inter-participant differences in CORT response, nuclear hsp90 concentrations, and GR/hsp90-associated SNPs are readily detected and potentially interrelated.

Natural Language Processing is a subfield of Computer Science that inspects the intersection of computation and human language. Idiomatic language is in turn ubiquitous in human language—when someone is a “diamond in the rough”, they are not literally an uncut diamond, rather, they are a person whose goodness is hidden by their surface appearance. However, a “diamond in the rough” is a phrase that can literally mean an uncut diamond—this is to say, it has two different senses, one that is idiomatic and one that is literal. Given a dictionary entry, a human can easily distinguish between idiomatic and literal definitions, however, doing this in an automatic fashion is difficult because it requires asking whether the meaning represented in the definition corresponds with the literal meaning of the phrase. This research leverages Wiktionary, an extremely large, collaboratively authored dictionary, to perform idiom identification in a scalable manner through machine learning algorithms. To do this identification, we are developing two sets of features—traits that we describe a definition-phrase pair with—selectional preference features and graph-based features. Selectional preference features describe traits of language that are used in a way that violates its literal meaning while graph-based features describe where the entry occurs in relation to the other pages within Wiktionary (what links does this page contain?). We are using two kinds of machine learning algorithms for idiom identification—supervised methods and semi-supervised methods. Supervised methods allow us to verify the validity of the features that we develop, and semi-supervised methods allow us to use these features for knowledge discovery because they are ideal for discovering definitions that are not yet marked as idiomatic but that ought to be.

The field of Brain Computer Interface (BCI) systems is concerned with the study of how thoughts alone can be used to interface with and control the actions of external systems. Such BCI systems have been actively studied since 1969 when the seminal work of Fitz, et. al. at the University of Washington showed that monkeys could voluntarily control the deflection of an external meter with their thoughts alone. To date, there has been extensive research in this area as concerns thought-control of external devices in focused-activity environments, such as video gaming, but research into the effectiveness of BCI methods in more-realistic, distracted-environments is somewhat sparse, with notable exceptions including the work of Lalor, et. al., who performed a non-controlled and non-randomized study of the effects of Steady-State Visual Evoked Potentials (SSVEP) on real-time video-gaming control and Friedricha, et. al. who performed a study on the effects of auditory distractions on the accuracy of BCIs. Our present study represents a refinement of the Lalor approach in a broader scope than video-game control through the addition of more carefully defined controls, and we further extend this research through the study of the effects of simultaneous auditory and visual environmental conditioning. Based on the study of Friedricha, et. al. involving auditory distractions, we expect a constant auditory tone to have no impact on BCI performance.

Although every cell in the human body has nearly the same DNA sequence, there are many different cell types performing different functions; for example, lymphoblast cells which differentiate into white blood cells or endothelial cells which provide a lining throughout the human vascular system. But cells do not always function as they should, for example, overproduction of white blood cells can be caused by lymphoblastic leukemia, a type of cancer. I intend to use data on the sensitivity of parts of the genome to deoxyribonuclease (DNase) digestion to explain the differences in cell behavior. Regions of the genome structure that are hypersensitive to DNase have already been shown to affect gene expression. I want to determine if portions of the genome which are mildly sensitive, or mesosensitive, to DNase play a role in cell differentiation. My hypothesis is that DNase can identify large scale regions, on the order of thousands of base pairs, that harbor cell type specific genes. I used Segway, software which uses a dynamic Bayesian network, to segment the genome based on DNase sensitivity data into regions of low sensitivity, mesosensitivity, and hypersensitivity on six different cell types. From my segmentations I removed the hypersensitive regions from considerations, and compared the low and mesosensitive regions to gene expression data. I found that regions Segway labels mesosensitive had higher levels of expression than low sensitivity regions. I expect to find that many of the regions responsible for cell type specific behavior are mesosensitive. If this is the case, I hope the results will be robust enough such that anyone with access to DNase data can determine what genes are active in a population of cells, enabling them to better predict the behavior of the cells. In furthering our understanding of gene regulation, we may better understand diseases such as leukemia.

Two strains of Botryococcus braunii algae were grown in three different sets of conditions to compare growth characteristics and lipid production. One species, UTEX LB 572, was a Race A Botryococcus braunii from the culture collection of the University of Texas in Austin. The second species, AC 762, was a Race B Botryococcus braunii from the culture collection of the University of Caen, France. UTEX 572 was found to produce two hydrocarbon chain lengths of C:16 and C:18 in a 2:3 ratio respectively. The AC 762 was found to produce hydrocarbons called botryococcenes of chain lengths between C:30 and C:36. The lipid chain lengths of both strains were evaluated by mass spectrometry. The UTEX 572 lipid content was quantified by analytical spectrometry after direct transesterification of the lipid content to biodiesel. The AC 762 lipid content was quantified by a photospectrometry technique developed by UC Berkeley researchers Eroglu and Melis (1985). Cells of both species were harvested during exponential growth phase from 400 mL of growth media and then freeze dried and analyzed gravimetrically. Nine flasks of each species were grown with 3000 lux average light intensity on a 12h:12h light:dark cycle at room temperature averaging 22° C. Three flasks of each species were grown with 1.5% CO2 in BG-11 media, three flasks of each species were grown in BG-11 media with air only and three control flasks were grown in respective carrier medias as those used in the algae collection of origin. Knowledge of the rates at which AC 762 and UTEX 572 produce hydrocarbons as well as the specific lengths and proportions of those hydrocarbons contributes to data needed for comparison of economic considerations in industrial applications.

Can a group of men and women be paired together to guarantee no divorces or scandals? In 1962, two economists posed this question mathematically and offered a solution. A 2012 Nobel Prize went to two of the key researchers that worked on this problem. Their solution took the form of an algorithm, a systematic procedure that could be performed by a computer. The solution to the "stable marriage problem" has both interesting theoretical properties and real-world applications, such as the assignment of medical students to hospitals. In the original version of the problem, each man and woman know their preferences over all their potential partners. We are considering a variant in which they only know some of their preferences. The algorithm not only needs to figure out whom to match with whom, but also what information to gather on preferences. We present our preliminary results on efficient algorithms for this problem. This is joint work with Anna Karlin and Jamie Morgenstern.

Acoustic levitation is a method of levitating matter against the influence of gravity using sound waves.  Small droplets of fluid can, for example, be suspended in standing wave fields produced by high-power ultrasonic transducers at locations where the gradients of the sound field potentials counterbalance the gravitational forces acting on the droplets, creating a simulated microgravity environment.  These environments could be useful in simulating and studying how chemical and biochemical reactions might take place in space.  Our research builds on earlier work by J.Santa Cruz and J. Griffith who successfully achieved acoustic levitation at EdCC in 2011.  Santa Cruz and Griffith were able to experimentally demonstrate rough preliminary agreement with Bjerknes theory for levitation stability points in a planar field approximation, but they also observed stable particle orbits completely beyond explanation by the planar theory.  Here, we compare our experimental findings to the more general Gor’kov theory of levitation to explain both one- and two-dimensional levitation stability subspaces.

Exoplanets are planets outside our solar system, and the current explosion in exoplanet discoveries is revolutionizing our understanding of the potential for extraterrestrial life. This prolific era of detections has stemmed largely from the unprecedented observing capabilities of NASA's Kepler Space Telescope. The Kepler Spacecraft collects high precision time-series photometric data on a fixed group of approximately 160,000 stars. The data are represented by temporal lightcurves (i.e. brightness vs. time) that can be used to detect transiting exoplanets, the topic of our research. Transits are events where an orbiting planet partially eclipses its host star, casting a small shadow on the telescope. To detect transit signals, we rely on the Quasi-Periodic Automated Transit Search Algorithm (QATS). As an automated tool, QATS provides a crucial means to reduce the Kepler dataset to a manageable size. However, since the algorithm is sensitive to stellar variability, eclipsing binary stars, and systematic artifacts of the spacecraft, additional analysis is required to separate true detections from false positives. Determining the best way to do this is the present focus of our work. Concurrently, we are exploring the potential for QATS not only to determine orbital period, but also to constrain transit depth and duration (properties related to the size of the exoplanet and to the density of the stellar host). While this increases the complexity of the QATS algorithm and the amount of output to manage, it provides greater potential for a fully automated transit search process with results that are more descriptive of the exoplanet systems detected.

Software systems are often difficult to comprehend. When a system behaves in an unexpected manner, or when developers must modify the code written by someone else, they have the challenge of understanding the system behavior. A typical way developers use to gain insight into a system is logging. Logging is a method that captures program activities and states during an execution into a log file (a text file), which developers then inspect. Unfortunately, the size and complexity of logs are often too large for manual inspection. Synoptic is a tool that supports the task of inspecting execution logs. Synoptic takes a log as input and infers a finite state machine model of the process that generated the log. The model that Synoptic produces is compact and captures essential behavior traits of the subject system. However, there is a missing piece of information. Synoptic is only interested in system activities, thus the models it infers are activity-based models. System states, which are also often captured in a log, are ignored. This project presents an approach to utilize state in model inference. We extend Synoptic to model data values in a subject system by annotating states in the inferred model with conditions on data values. We use Daikon, an implementation of dynamic detection of likely invariants, to determine those conditions from an input log that captures data values at each state of the system. We argue that this new model helps developers reason about their systems better.