Bea Yu

Vocal biomarkers of depression based on motor incoordination

In Major Depressive Disorder (MDD), neurophysiologic changes can alter motor control [1, 2] and therefore alter speech production by influencing the characteristics of the vocal source, tract, and prosodics. Clinically, many of these characteristics are associated with psychomotor retardation, where a patient shows sluggishness and motor disorder in vocal articulation, affecting coordination across multiple aspects of production [3, 4]. In this paper, we exploit such effects by selecting features that reflect changes in coordination of vocal tract motion associated with MDD. Specifically, we investigate changes in correlation that occur at different time scales across formant frequencies and also across channels of the delta-mel-cepstrum. Both feature domains provide measures of coordination in vocal tract articulation while reducing effects of a slowly-varying linear channel, which can be introduced by time-varying microphone placements. With these two complementary feature sets, using the AVEC 2013 depression dataset, we design a novel Gaussian mixture model (GMM)-based multivariate regression scheme, referred to as Gaussian Staircase Regression, that provides a root-mean-squared-error (RMSE) of 7.42 and a mean-absolute-error (MAE) of 5.75 on the standard Beck depression rating scale. We are currently exploring coordination measures of other aspects of speech production, derived from both audio and video signals.

On the relative importance of vocal source, system, and prosody in human depression

In Major Depressive Disorder (MDD), neurophysiologic changes can alter motor control [1][2] and therefore alter speech production by influencing vocal fold motion (source), the vocal tract (system), and melody (prosody). In this paper, we use a database of voice recordings from 28 depressed subjects treated over a 6-week period [3] to compare correlations between features from each of the three speech-production components and clinical assessments of MDD. Toward biomarkers for audio-based continuous monitoring of depression severity, we explore the contextual dependence of these correlations with free-response and read speech, and show tradeoffs across categories of features in these two example contexts. Likewise, we also investigate the context-and speech component-dependence of correlations between our vocal features and assessment of individual symptoms of MDD (e.g., depressed mood, agitation, energy). Finally, motivated by our initial findings, we describe how context may be useful in “on-body” monitoring of MDD to facilitate identification of depression and evaluation of its treatment.

Multispectral vegetation detection for improved SAR CCD

Synthetic Aperture Radar Coherent Change Detections (SAR CCD) sensitivity to changes in ground surface height is coupled with sensitivity to other environmental changes such as minor movement in vegetation. The CCD Clutter Location, Estimation and Negation (CLEAN) algorithm decreases the false alarm rate in SAR CCD change pattern detection algorithms using intensity information in SAR images to discriminate false alarms from changes of interest. Unfortunately, CLEAN has difficulty identifying vegetation using only SAR imagery and vegetation is problematic in SAR CCD. In this paper, we propose an extension to CLEAN that fuses information from multispectral imagery with SAR intensity information for more robust vegetation classification. Experimental results show that our algorithm significantly improves change identification in SAR CCD.

Experiments and simulations on short chain fatty acid production in a colonic bacterial community

Understanding how production of specific metabolites by gut microbes is modulated by interactions with surrounding species and by environmental nutrient availability is an important open challenge in microbiome research. As part of this endeavor, this work explores interactions between F. prausnitzii, a major butyrate producer, and B. thetaiotaomicron, an acetate producer, under three different in vitro media conditions in monoculture and coculture. In silico Genome-scale dynamic flux balance analysis (dFBA) models of metabolism in the system using COMETS (Computation of Microbial Ecosystems in Time and Space) are also tested for explanatory, predictive and inferential power. Experimental findings indicate enhancement of butyrate production in coculture relative to F. prausnitzii monoculture but defy a simple model of monotonic increases in butyrate production as a function of acetate availability in the medium. Simulations recapitulate biomass production curves for monocultures and accurately predict the growth curve of coculture total biomass, using parameters learned from monocultures, suggesting that the model captures some aspects of how the two bacteria interact. However, a comparison of data and simulations for environmental acetate and butyrate changes suggest that the organisms adopt one of many possible metabolic strategies equivalent in terms of growth efficiency. Furthermore, the model seems not to capture subsequent shifts in metabolic activities observed experimentally under low-nutrient regimes. Some discrepancies can be explained by the multiplicity of possible fermentative states for F. prausnitzii. In general, these results provide valuable guidelines for design of future experiments aimed at better determining the mechanisms leading to enhanced butyrate in this ecosystem. Importance Studies associating butyrate levels with human colonic health have inspired research on therapeutic microbiota consortia that would optimize butyrate production if implanted in the human colon. Faecalibacterium prausnitzii is commonly observed in human fecal samples and produces butyrate as a product of fermentation. Previous studies indicate that Bacteroides thetaiotaomicron, also commonly found in human fecal samples, may enhance butyrate production in F. prausnitzi when the two species are co-localized. This possibility is investigated here under different environmental conditions using experimental methods paired with computer simulations of the whole metabolism of bacterial cells. Initial findings indicate that interactions between these two species result in enhanced butyrate production. However, results also paint a nuanced picture, suggesting the existence of a multiplicity of equivalently efficient metabolic strategies and complex interactions between acetate and butyrate production in these species that appear highly dependent on specific environmental conditions.

A Framework for Predicting Design Failures in Engineered Genetic Codes

Extreme engineering of an organism’s genetic code could impart true genetic incompatibility, even blocking effects of horizontal gene transfer and viral infection. Recent experiments exploring this possibility demonstrate that such radical genome engineering achievements are plausible. However, it is unclear when the modifications will compromise the fitness of an organism. Efforts to reformat an entire genome are difficult and expensive; computational methods predicting fruitful experimental trajectories could play a pivotal role in advancing such efforts. We present a framework for building in silico models to assist genome-scale engineering. Genetic code engineering requires choosing from many possible codon-usage schemes, to find a design that is viable and effective. We use machine learning to identify which alternative codon-usage schemes are likely to result in no observed viable cells. Our data-driven approach employs observations of how modifying codon usage in individual genes impacted observed viability in E. coli, revealing salient features for early identification of problematic genetic code designs. We achieved an average area under the receiver operating characteristic of 0.72 on out-ofsample data. Author Summary As machine learning and artificial intelligence play an increasingly central role in science and engineering, it will be important to establish standardized techniques that facilitate the dialogue between experimentation and modeling. Biological experimental techniques are concurrently evolving at a rapid pace, providing unique opportunities to collect high-quality, novel information that was previously unobtainable. This work navigates the landscape of this vast, new territory, identifies interesting landmarks for exploration and posits new approaches towards advancing our research efforts in these areas. In this work, we show that, using a small dataset of 47 observations and rigorous nested cross validation techniques, we can build a model that makes better-than-random predictions of how codon usage changes in essential genes influence viability in E. coli. These predictions can be used to inform experimental trajectories in both genetic code and codon optimization experiments. We discuss ways to improve this model, iteratively, by performing high value experiments that decrease uncertainty in predictions and extrapolation error. Finally, we present novel visualization methods to aid in developing intuitions for how re-coding impacts groups of genes. These methods are also useful tools in building important insights into how well machine learning algorithms can generalize to new data.

Using contextual information to improve SAR CCD: Bayesian contextual coherent change detection (BC CCD)

Semi-automated, subtle, ground surface change detection using synthetic aperture radar coherent change detection (SAR CCD) suffers from a high false alarm rate. Errors due to dispersion in coherence estimates propagate into change estimates based on coherence values. Low coherence values indicating ground surface changes also map to non-salient world states, such as leaf movement and radar shadows. In this paper, we address these issues by incorporating contextual information from a multispectral land-cover classification into SAR CCD using Bayesian approach, called Bayesian Contextual Coherent Change Detection (BC CCD). We demonstrate improved change detection performance of BC CCD with data over diverse areas using ROC curves. BC CCD shows substantial improvements over unprocessed SAR CCD and the SAR CCD false alarm reduction method Clutter Location Estimation and Negation (CLEAN) CCD. Supervised classification on the multispectral (MS) image coupled with a maximum entropy prior stipulation yields the highest performance gains.