Thesis (Ph. D.)--Harvard-MIT Division of Health Sciences and Technology, 2005. / This electronic version was submitted by the student author. The certified thesis is available in the Institute Archives and Special Collections. / Includes bibliographical references (p. 153-166). / The basal ganglia (BG) are a group of highly interconnected nuclei buried deep in the brain. They are involved in an important range of brain functions, including both lower-level movement control and higher-level cognitive decision making. Dysfunction of the BG has been linked to human neurological disorders such as Parkinson's disease, Huntington's disease, and schizophrenia. This thesis proposes a unified functional model of the BG, called multi-input multi-output adaptive switching (MIMOAS) model that attempts to account for the role of the BG in both higher-level rote behavior and lower-level motor control. In the model, BG circuitry effectively implements a large set of parallel noncompetitive logical OR and NOR circuits that can be driven by specific patterns of cortical activity. These afford selective gating of target thalamocortical neurons. This process can be viewed as a general mapping between binary context and response vectors. The mapping is proved to be learnable via a reinforcement mechanism that is consistent with actions commonly proposed for nigro-striatal dopaminergic pathways in the striatum and homeostasis in synaptic physiology. It appears that the cortico-striatal connections provide a biologically plausible realization of winner-take-all dynamics that is different from many engineering alternatives implementing the same function. With the winner-take-all units as functioning as a hidden layer, using corticostriatal weights as the only tunable parameters, the adaptive BG network can develop the capacity to perform universal binary mappings. In this way, the model can simulate important features of procedural learning in human experiments. At the same time, it can be shown that derangement of the winner-take-all dynamics could underlie the tremor and rigidity seen in Parkinson's disease. / by Zhi-Hong Mao. / Ph.D.

Thesis (Ph. D.)--Harvard University--MIT Division of Health Sciences and Technology, 2005. / Includes bibliographical references (leaves 112-125). / Speech is perhaps the most ecologically important acoustic stimulus to human beings, because it remains the primary means by which people interact socially. Despite many significant advances made in the development of cochlear implants, even the most successful cochlear-implant users do not hear as well as normal-hearing listeners. The differences in performance between normal-hearing listeners and cochlear-implant users are especially pronounced in understanding speech in complex auditory environments. For normal-hearing listeners, voice pitch or the fundamental frequency (FO) of voicing has long been thought to play an important role in the perceptual segregation of speech sources. The aim of this dissertation was to examine the role of voice pitch in speech perception in the presence of background interference, specifically simulating aspects of envelope-vocoder style implant processing. The findings of the studies show that FO encoded via envelope periodicity does not provide a sufficiently salient cue for the segregation of speech. This suggests that the poor speech reception performance of implant users in background interference may, at least in part, be due to the lack of salient voice pitch cues. When low-frequency fine-structure information was added to envelope- vocoder processed high-frequency information, some FO segregation benefits returned and the reception of speech in complex backgrounds improved. Taken as a whole, the dissertation suggests that low frequency fine-structure information is important to the task of speech segregation, and that every effort should be made to present such information to cochlear-implant users. / by Michael Kaige Qin. / Ph.D.

Thesis (S.M.)--Harvard-MIT Division of Health Sciences and Technology, 2005. / Includes bibliographical references (p. 36-37). / Failure to follow-up on abnormal tests is a common clinical concern comprising the quality of care. Although many clinicians track their patient follow-up by scheduling follow-up visits or by leaving physical reminders, most feel that automated, computerized systems to track abnormal test results would be useful. While existing clinical decision support systems and computerized clinical reminders focus on providing assistance with choosing the appropriate follow-up management, they fail by not tracking that follow-up effectively. We believe that clinicians do not want suggestions how to manage their patients, but instead want help tracking follow-up results once they have decided the management plan. We believe that a well-designed system can successfully track this follow-up and only require a small amount of information and time from the clinician. We have designed and implemented a complete tracking system including 1) an authoring tool to define tracking guidelines, 2) a query tool to search electronic medical records and identify patients without follow-up, and 3) a clinical tool to send reminders to clinicians and allow them to easily choose the follow-up management. Our tracking system has made improvements on previous reminder systems by 1) using our unique risk-management guideline model that more closely mirrors, yet does not attempt to replicate, the clinical decision process, 2) our use of massive population-based queries for tracking all patients simultaneously, and 3) our longitudinal approach that documents all steps in the patient follow-up cycle. With these developments, we are able to track 450 million pieces of clinical data for 1.8 million patients daily. / (cont.) Keyword follow-up tracking; reminder system; preventive medicine; computerized medical record system; practice guidelines; clinical decision support system / by Daniel Todd Rosenthal. / S.M.

Thesis (Ph. D.)--Harvard-MIT Division of Health Sciences and Technology, 2004. / Includes bibliographical references (p. 163-185). / The cardiovascular response to changes in posture has been the focus of numerous investigations in the past. Yet despite considerable, targeted experimental effort, the mechanisms underlying orthostatic intolerance (OI) following spaceflight remain elusive. The number of hypotheses still under consideration and the lack of a single unifying theory of the pathophysiology of spaceflight-induced OI testify to the difficulty of the problem. In this investigation, we developed and validated a comprehensives lumped-parameter model of the cardiovascular system and its short-term homeostatic control mechanisms with the particular aim of simulating the short-term, transient hemodynamic response to gravitational stress. Our effort to combine model building with model analysis led us to conduct extensive sensitivity analyses and investigate inverse modeling methods to estimate physiological parameters from transient hemodynamic data. Based on current hypotheses, we simulated the system-level hemodynamic effects of changes in parameters that have been implicated in the orthostatic intolerance phenomenon. Our simulations indicate that changes in total blood volume have the biggest detrimental impact on blood pressure homeostasis in the head-up posture. If the baseline volume status is borderline hypovolemic, changes in other parameters can significantly impact the cardiovascular system's ability to maintain mean arterial pressure constant. In particular, any deleterious changes in the venous tone feedback impairs blood pressure homeostasis significantly. This result has important implications as it suggests that al-adrenergic agonists might help alleviate the orthostatic syndrome seen post-spaceflight. / by Thomas Heldt. / Ph.D.

Thesis (S.M.)--Harvard-MIT Division of Health Sciences and Technology, 2005. / Includes bibliographical references (p. 31-33). / Background A number of important applications in medicine and biomedical research, including quality of care surveillance and identification of prospective study subjects, require identification of large cohorts of patients with specific clinical characteristics. Currently used conventional techniques are either labor-intensive or imprecise, while natural language processing-based applications are relatively slow and expensive. Specific Aims In this thesis we describe the design and formal evaluation of PACT - a suite of rapid, accurate, and easily portable software tools for identification of patients with specific clinical characteristics through analysis of the text of physician notes in the electronic medical record. Methods PACT algorithm is based on sentence-level semantic analysis. The major steps involve identification of word tags (e.g. name of the disease or medications exclusively used to treat the disease) specific for the clinical characteristics in the sentences of the physician notes. Sentences with word tags and negative qualifiers (e.g. "rule out diabetes") are excluded from consideration. PACT can also identify quantitative (e.g. blood pressure, height, weight) and semi-quantitative (e.g. compliance with medical treatment) clinical characteristics. PACT performance was evaluated against blinded manual chart review (the "gold standard") and currently used computational methods (analysis of billing data). Results Evaluation of PACT demonstrated it to be rapid and highly accurate. PACT processed 6.5 to 8.8x 10⁵ notes/hour (1.0 to 1.4 GB of text / hour). / (cont) When compared to the gold standard of manual chart review, PACT sensitivity ranged (depending on the patient characteristic being extracted from the notes) from 74 to 100%, and specificity from 86 to 100%. K statistic for agreement between PACT and manual chart review ranged from 0.67 to 1.0 and in most cases exceeded 0.75, indicating excellent agreement. PACT accuracy substantially exceeded the performance of currently used techniques (billing data analysis). Finally, index of patient non-compliance with physician recommendations computed by PACT was shown to correlate with the frequency of annual Emergency Department visits: patients in the highest quartile for the index of non-compliance had 50% as many annual visits as the patients in the lowest quartile. Conclusion PACT is a rapid, precise and easily portable suite of software tools for extracting focused clinical information out of free text clinical documents. It compares favorably with computation techniques currently available for the purpose (where ones exist). It represents an important advance in the field, and we plan to continue to develop this concept further to improve its performance and functionality. / by Alexander Turchin. / S.M.

Thesis (Ph. D.)--Harvard-MIT Division of Health Sciences and Technology, 2011. / Cataloged from PDF version of thesis. / Includes bibliographical references. / Active motility in outer hair cells (OHCs) amplifies basilar-membrane (BM) and auditory-nerve (AN) responses to low-level sounds. The recent finding that medial olivocochlear (MOC) efferents (which innervate OHCs) inhibit AN initial peak (ANIP) responses from mid-to-high-level clicks, but do not inhibit initial BM responses, suggests a coupling of OHC motility to inner-hair-cell (IHC) stereocilia that is not through the BM. The main thesis objective was to test whether different OHC mechanisms produce AN responses to low-level sounds versus ANIP from mid-to-high-level clicks by comparing the suppressive effects of low-frequency "bias-tones" on these responses. Bias tones suppress by pushing OHC stereocilia into low-slope regions of their mechanoelectric transduction functions thereby lowering OHC amplification, particularly for probe tones near an AN-fiber's characteristic frequency (CF). This suppression occurs at opposite bias-tone phases, with one suppression typically larger than the other. Bias-tone effects were measured on cat AN-fiber responses using 50 Hz bias tones. In the first thesis part, bias-tone suppressive effects on AN responses to low-level clicks and low-level CF-tones were found to be similar, as expected but never previously shown. Then, in the main thesis focus, bias-tone suppressions of AN responses to low-level clicks and ANIP responses were studied. Both responses were suppressed twice each bias-tone cycle, but their major suppressions were at opposite bias-tone phases, which indicates that both ANIP and low-level AN responses depend on the slope of OHCstereocilia mechanoelectric-transduction, but with some significant difference. In the last thesis part, bias-tone suppression effects on low-CF (<4 kHz) AN-fiber responses to low-level CF and off-CF (by >0.7 octaves) tones were studied. Previous work found differences in AN-response group delays between CF and off-CF frequency regions that might arise from two different IHC-drive mechanisms, and the objective was to test this hypothesis. Our results showed similar bias-tone effects in both regions. Overall, the results demonstrate differences and similarities in the OHC mechanisms that produce ANIP and traditional, low-level cochlear amplification, and the results are consistent with the ANIP drive coupling OHC motility to IHC stereocilia without going through BM motion. / by Hui S. Nam. / Ph.D.

Thesis (Ph. D. in Biomedical Engineering)--Harvard-MIT Program in Health Sciences and Technology, 2012. / Cataloged from PDF version of thesis. / Includes bibliographical references (p. 103-112). / The efficient isolation of specific bioparticles in lab-on-a-chip platforms is important for many applications in clinical diagnostics and biomedical research. The majority of microfluidic devices designed for specific particle isolation are constructed of solid materials such as silicon, glass, or polymers. Such devices are hampered by some critical challenges: the low efficiency of particle-surface interactions in affinity based particle capture, the difficulty in accessing sub-micron particles, and design inflexibility between platforms for different particle sizes. Existing porous materials do not offer the structural properties or patterning capabilities to address these challenges. This work introduces Vertically Aligned Carbon Nanotubes (VACNTs) as a new porous material in microfluidics, and demonstrates the different ways in which it can improve bioparticle separation across both the micro and nano size scales. Our devices are fabricated by integrating patterned VACNT forests with ultra-high (99%) porosity inside of microfluidic channels. We demonstrated both mechanical and chemical capture of particles ranging over three orders of magnitude in size using simple device geometries. Nanoparticles below the inter-nanotube spacing (80 nm) of the forest can penetrate inside the forest and interact with the large surface area created by individual nanotubes. For larger particles (>80 nm), the ultra-high porosity of the nanotube elements enhances particlestructure interactions on the outer surface of the patterned nanoporous elements. We showed using both modeling and experiment that this enhancement is achieved through two mechanisms: the increase of direct interception and the reduction of near-surface hydrodynamic resistance. We verified that the improvement of interception efficiency also results in an increase in capture efficiency when comparing nanoporous VACNT post arrays with solid PDMS post arrays of the same geometry, using both bacteria and cells as model systems. The technology developed in this thesis can provide improved control of bioseparation processes to access a wide range of bioparticles, opening new pathways for both research and point-of-care diagnostics. / by Grace Dongqing Chen. / Ph.D.in Biomedical Engineering

Thesis (Ph. D. in Speech and hearing Bioscience and technology)--Harvard-MIT Program in Health Sciences and Technology, 2012. / Cataloged from PDF version of thesis. / Includes bibliographical references (p. 86-91). / Unlike lower vertebrates, capable of spontaneous hair cell regeneration, mammals experience permanent sensorineural hearing loss following hair cell damage. Although low levels of hair cell regeneration have been demonstrated in the immature mammalian vestibular system, the cochlea has been thought to lack any spontaneous regenerative potential. Inhibition of the Notch pathway can stimulate hair cell generation in neonatal mammals, but the specific source of these new hair cells has been unclear. Here, using in vitro lineage tracing with the supporting cell markers Sox2 and Lgr5, we show that Lgr5-positive inner pillar and 3rd Deiter's cells in gentamicin-damaged organs of Corti from neonatal mice give rise to new hair cells following treatment with a Notch inhibitor. These new hair cells are generated primarily through direct transdifferentiation of supporting cells, although a small number show evidence of proliferation. Inner pillar cells show the greatest transdifferentation capability, giving rise to immature outer hair cells, and transdifferentiating in response to damage even in the absence of Notch inhibition. In vivo pharmacologic inhibition of Notch and in vivo lineage tracing with Sox2 during genetic Notch inhibition provide generally consistent results, although additional new hair cells develop in the inner hair cell region. These data suggest a spontaneous capacity for hair cell regeneration in the neonatal mammalian cochlea. In addition, the data identify Lgr5-positive supporting cells as potential hair cell progenitors, making them an attractive target for future hair cell regeneration treatments. / by Naomi F. Bramhall. / Ph.D.in Speech and hearing Bioscience and technology

Thesis (S.M.)--Harvard-MIT Division of Health Sciences and Technology, 2005. / Includes bibliographical references (p. 56-57). / Cancer development and progression typically features genomic instability frequently resulting in genomic changes involving DNA copy number gains or losses. Identifying the genomic location of these regional alterations provides important opportunities for the discovery of potential novel oncogenes and tumor suppressors. Recently, array based competitive genomic hybridization (array-CGH) has become available as a powerful approach for genome-wide detection of DNA copy number changes. Array-CGH assesses DNA copy number in tumor samples through competitive hybridization on microarrays containing probes for thousands of genes. The datasets generated are complex and require statistical methods to accurately define discrete and uniform copy number from the data and to identify transitions between genomic regions with altered copy number. Several approaches based on different statistical frameworks have been developed. However, a fundamental informatic issue in array-CGH analysis remains unsolved by these methods. In particular, sample-specific data compression, a result of tumor cells being commonly admixed with normal cells in many tumor types, must be accounted for in each sample analyzed. Additionally, in order to accurately assess deviations from normal copy number, the copy number readout must be shifted to faithfully represent the baseline copy number in each tumor sample. Failure to appropriately address these issues reduces the accuracy of the data in hard-threshold based high-level analysis. / (cont.) By using the natural framework Hidden Markov Models (HMM) to model the distribution of array-CGH signals, a method infer the absolute copy number and identify change points has been developed to address the above problems. This method has been validated on independent dataset and its utility in inference on array-CGH data is demonstrated here. / by Yunyu Zhang. / S.M.

Thesis (Ph. D.)--Harvard-MIT Division of Health Sciences and Technology, 2007. / Includes bibliographical references (p. 163-181). / High-throughput molecular data are revolutionizing biology by providing massive amounts of information about gene expression and regulation. Such information is applicable both to furthering our understanding of fundamental biology and to developing new diagnostic and treatment approaches for diseases. However, novel mathematical methods are needed for extracting biological knowledge from high-dimensional, complex and noisy data sources. In this thesis, I develop and apply three novel computational approaches for this task. The common theme of these approaches is that they seek to discover meaningful groups of genes, which confer robustness to noise and compress complex information into interpretable models. I first present the GRAM algorithm, which fuses information from genome-wide expression and in vivo transcription factor-DNA binding data to discover regulatory networks of gene modules. I use the GRAM algorithm to discover regulatory networks in Saccharomyces cerevisiae, including rich media, rapamycin, and cell-cycle module networks. I use functional annotation databases, independent biological experiments and DNA-motif information to validate the discovered networks, and to show that they yield new biological insights. Second, I present GeneProgram, a framework based on Hierarchical Dirichlet Processes, which uses large compendia of mammalian expression data to simultaneously organize genes into overlapping programs and tissues into groups to produce maps of expression programs. I demonstrate that GeneProgram outperforms several popular analysis methods, and using mouse and human expression data, show that it automatically constructs a comprehensive, body-wide map of inter-species expression programs. / (cont.) Finally, I present an extension of GeneProgram that models temporal dynamics. I apply the algorithm to a compendium of short time-series gene expression experiments in which human cells were exposed to various infectious agents. I show that discovered expression programs exhibit temporal pattern usage differences corresponding to classes of host cells and infectious agents, and describe several programs that implicate surprising signaling pathways and receptor types in human responses to infection. / by Georg Kurt Gerber. / Ph.D.