Dartmouth Institute for Health Policy and Clinical Practice
Dartmouth Digital Commons (Dartmouth College)Not a member yet
8214 research outputs found
Sort by
Interpretable Learning in Multivariate Big Data Analysis for Network Monitoring
Get PDFThere is an increasing interest in the development of new data-driven models useful to assess the performance of communication networks. For many applications, like network monitoring and troubleshooting, a data model is of little use if it cannot be interpreted by a human operator. In this paper, we present an extension of the Multivariate Big Data Analysis (MBDA) methodology, a recently proposed interpretable data analysis tool. In this extension, we propose a solution to the automatic derivation of features, a cornerstone step for the application of MBDA when the amount of data is massive. The resulting network monitoring approach allows us to detect and diagnose disparate network anomalies, with a data-analysis workflow that combines the advantages of interpretable and interactive models with the power of parallel processing. We apply the extended MBDA to two case studies: UGR’16, a benchmark flow-based real-traffic dataset for anomaly detection, and Dartmouth’18, the longest and largest Wi-Fi trace known to date
De novo Synthetic Studies on Tetracyclic Triterpenoid Systems: Stereoselective Preparation, Biological Evaluation and Total Syntheses of Euphol and Tirucallol
Get PDFTetracyclic triterpenoids have a rich history in biology, medicine, and chemistry. In terms of pharmaceutical relevance, compounds with this scaffold are one of the most successful classes with over 100 FDA approved examples. Despite this success, the vast majority of compounds within this class are prepared through semisynthesis. While this approach is undeniably successful, judged by the sheer number of medications generated, it also has limitations. Namely, semisynthesis requires the use of complex starting materials that have fixed absolute stereochemistry and often sparse functionality. One approach to address the inherent limitations with semisynthesis is through de novo synthesis. Recently, our group has put significant effort into establishing novel de novo synthetic organic chemistry to access unique tetracyclic triterpenoid systems. This thesis outlines my contributions to this body of work, including: (1) the development of a stereoselective Friedel–Crafts cyclization that allows access to C9-C13 anti-substituted tetracyclic triterpenoid systems (2) the design and synthesis of the enantiomeric estranes for glucocorticoid receptor modulation and (3) the first asymmetric total syntheses of (+)-euphol and (+)-tirucallol. It is our belief that the chemistry developed through these studies have the potential to be useful for the preparation of biologically relevant systems that would be difficult to access through any other means. Additionally, it is our hope that the work done in this thesis will be helpful to inspire new research to be done on these fascinating and complex systems
Open Source Supply Chain Security: a Cost-Benefit Analysis of Achieving Various Security Thresholds in Build Environments
Get PDFOpen source software has become a cornerstone of modern software development, offering unparalleled opportunities for innovation and collaboration. However, its widespread adoption has also introduced a host of security vulnerabilities, particularly in the software supply chain. This paper provides a comprehensive cost-benefit analysis of achieving various security thresholds to harden the build environment, focusing on isolated, hermetic, reproducible, and bootstrappable builds. For each build type, we provide a clear definition and outline the steps required for implementation. We then evaluate the associated costs and benefits of each build, emphasizing their roles in strengthening the build environment and enhancing supply chain security. The paper concludes with recommendations for stakeholders, including startups, large corporations, and government agencies, and proposes future research directions to enhance build environment security
Exploring Tokenization Techniques to Optimize Patch-Based Time-Series Transformers
Get PDFTransformer architectures have revolutionized deep learning, impacting natural language processing and computer vision. Recently, PatchTST has advanced long-term time-series forecasting by embedding patches of time-steps to use as tokens for transformers. This study examines and seeks to enhance PatchTST\u27s embedding techniques. Using eight benchmark datasets, we explore explore novel token embedding techniques. To this end, we introduce several PatchTST variants, which alter the embedding methods of the original paper. These variants consist of the following architectural changes: using CNNs to embed inputs to tokens, embedding an aggregate measure like the mean, max, or sum of a patch, adding the exponential moving average (EMA) of prior tokens to any given token, and adding a residual between neighboring tokens. Our findings show that CNN-based patch embeddings outperform PatchTST’s linear layer strategy, andsimple aggregate measures, particularly embedding just the mean of a patch, provide comparable results to PatchTST for some datasets. These insights highlight the potential for optimizing time-series transformers through improved embedding strategies. Additionally, they point to PatchTST\u27s inefficiency at exploiting all information available in a patch during the token embedding process
Behind the Rhododendron
Get PDFLike a rhododendron, man most often presents to the observer his ornamental self. Yet, what keeps in the shadows of all he presents to the world? What remains beneath in the dark soils of past experience from where he’s come? What must perish and decay to yield hearty growth and an individual expression that reaches humbly, and more deeply, toward some divine light? And what of this seemingly intuitive notion of something sacred he may struggle to comprehend and know?
Behind the Rhododendron is an intensely personal collection of poems that earnestly strives to scratch at the surface of such existential concerns from the fluid perspectives of fathers, sons, and humans being spiritual. If there was any singular guiding principle in the creation of this work, it was just to attempt to artfully and honestly bring with the poetic form some of what is ugly and shaded into harmonious union with all that which is beautiful and enlightening
Automated Glacier Classification in High Mountain Asia using Machine Learning and a Random Forest Classifier
Get PDFHigh Mountain Asia (HMA) is home to the largest mass of glaciers and ice outside the north and south polar regions. HMA glaciers are projected to experience accelerated mass loss from higher greenhouse gas emissions through the end of the century. Many studies of glacier mass balance and mass loss in HMA obtain glacier area from the Randolph Glacier Inventory (RGI). However, the RGI is designed to show glacier area across the world that is accurate to the year 2000 and, as a result, is not an accurate representation of the current state of glacier area in HMA. Additionally, glacier outlines in HMA are delineated using various manual and semi-automated mapping methods which are labor- and time-intensive making them difficult to repeat on a large spatial or temporal scale.
To address this issue, I developed an automated method to classify glacier area, leveraging machine learning via a random forest classifier model. I used remotely sensed data including Landsat-7 Enhanced Thematic Mapper Plus multispectral imagery, a digital elevation model, and glacier velocity data to train the model. I tested two different geographic regions, the Karakoram Mountains and the Pamir Alay Mountains, for model training and accuracy. Within each region, the model was trained and tested on different-sized tiles. Overall, the model trained in the Karakoram Mountains more accurately identified glacier cover than the model trained in the Pamir Alay Mountains. Test 05a, in the Karakoram Mountains with a tile size of 0.75° by 0.5°, was the best at classifying glacier cover based on its ice-covered classification results compared to the other tiles and had an accuracy score of 0.9615. The most important feature for identifying glacier cover in test 05a was glacier velocity.
I then applied the trained model to classify glacier area across the Tian Shan Mountains using satellite imagery from two time periods, 1999-2010 (Epoch 1) and 2015-2023 (Epoch 2) to assess change in glacier area since around the year 2000. I calculated a 10.1% decrease in glacier area from 10,183 ± 601 km2 in Epoch 1 to 9,156 ± 540 km2 in Epoch 2. Despite some persistent areas that the model misclassified as glacier-covered that were not glaciated, the model results overall appeared to agree well with the RGI polygons
Research: The Little Insects That Could: New Hampshire\u27s Arctic Butterflies
Get PDFNew Hampshire’s arctic butterflies, the White Mountain fritillaries
Proterozoic to Paleozoic Bedrock Geology of the Islesboro Block, Penobscot Bay Maine, USA
No full textThe Islesboro block is a fault-bounded crustal fragment isolated between the St. Croix and Ellsworth terranes in Penobscot Bay, Maine. We present the results of new geological mapping of the Islesboro 7.5´ quadrangle, which is coupled with new laser ablation-inductively coupled plasma mass spectrometry (LA-ICPMS) U-Pb zircon data to support revised unit subdivisions and provide an assessment of the age and provenance of the Islesboro block. Based on these data, the Islesboro block consists of: (1) the Paleoproterozoic–Mesoproterozoic(?) Seven Hundred Acre Island Formation, composed of marble, quartzite, and garnet-mica schist, and intruded by highly retrogressed Neoproterozoic (ca. 670–655 Ma) amphibolite and pegmatite, the latter of which is dated herein by LA-ICPMS U-Pb analyses on zircon; (2) the Neoproterozoic–Cambrian(?) Islesboro Formation, a heterogenous package of metaclastic and metacarbonate rocks subdivided here into seven new distinct members and submembers; and (3) the informal Cambrian(?) Turtle Head Cove formation, consisting of metaclastic rocks subdivided here into three members. The stratigraphic relationships among these primary formations are commonly obscured by faults, but repeatable map patterns suggest they were once in depositional contact. The Islesboro block is bounded by two major dextral strike-slip faults (Turtle Head and Penobscot Bay faults), which are linked by N-S- and NE-SW-striking faults that offset and overprint at least one previous phase of contractional or transpressional deformation. All units of the Islesboro block record complex deformation, including abundant subvertical faults, open to isoclinal and predominantly asymmetrical folds, and narrow to diffuse zones of cataclasis and mylonitization. Additionally, the Seven Hundred Acre Island Formation contains a highly retrogressed, peak mineral assemblage of garnet- kyanite-staurolite-rutile, indicating that it was subjected to higher amphibolite-facies P-T conditions prior to ca. 655 Ma. Ongoing petrologic, geochronologic, and structural analyses will help clarify the origin and tectonic evolution of the Islesboro block and its significance in the northern Appalachian orogen, but our preliminary results suggest strong affinities to both the adjacent St. Croix and Ellsworth terranes
Enhancing Terrain Creation in Unity With a Hybrid Modular Tool: Integrating Procedural Generation and Manual Input for Optimized Design Flexibility and Efficiency
Get PDFDigital terrain creation in Unity, especially for games, typically requires extensive manual effort, which is time-consuming and inefficient. Although procedural generation offers a systematic alternative, it often lacks the precision needed for specific design requirements, such as exact path or water body placements. This study introduces a novel modular tool that integrates manual input capabilities with automated procedural generation, aiming to combine the efficiency of procedural techniques with the precision of manual methods. The tool is designed for use within Unity, allowing for detailed customization and adjustments. Comprehensive user testing was conducted with 21 participants. The effectiveness of the tool was evaluated both qualitatively and quantitatively. The overall feedback for the modular terrain building tool was highly positive, both quantitatively and qualitatively. Users praised the tool’s enjoyable using process, intuitive design, and the high quality of the terrains created, noting its efficiency in reducing terrain creation time. However, there were suggestions for more customization options and performance improvements. Overall, the tool was seen as a valuable asset for terrain building, effectively streamlining and enhancing the terrain creation process in Unity
Chromatin regulation by SWI/SNF remodelers in somatic stem cell maintenance and transformation
No full textCell identity is defined by the epigenome, whereby chromatin regulators work in concert to promote gene expression programs that serve a cell’s specialized purpose. Mutations in chromatin regulators are amongst the most frequent drivers of human disease, underscoring the importance of understanding their activities in maintaining cell identity and tissue function. In particular, mutations in subunits of the evolutionarily conserved SWI/SNF chromatin remodeling complexes drive diseases across human tissues in both development and adult tissue maintenance5. Three major SWI/SNF complexes exist: BAF, PBAF, and GBAF, which differ in their composition and genomic targeting but share an ATP-dependent catalytic activity to bind and mobilize nucleosomes. SWI/SNF complexes control chromatin access at cis-regulatory elements known as enhancers, promoters, and insulators which are dense in transcription factor binding sites and undergo dynamic regulation by lineage-specific chromatin regulators. Thus, the sheer biochemical complexity and genome wide functions have made it difficult to grasp the rules that govern SWI/SNF functions in vivo. Despite their enrichment at lineage-specific regulatory elements, SWI/SNF chromatin functions have been largely investigated in cancer cell lines that suffer from already compromised genomes and epigenomes. No deep investigation has been made into their homeostatic contributions to somatic stem cells, which underpin disease origins. We sought to address these gaps in knowledge using CRISPR-Cas9 gene editing and state-of-the-art chromatin assays in human colon organoids. Colorectal cancers and inflammatory bowel diseases frequently incur ARID1A mutations in the colon epithelium. This presented a relevant paradigm to model BAF complex function in lineage competent adult stem cells. Using organoid models, we identified robust cellular and molecular evidence for the pathogenic relevance of ARID1A loss-of-function in inflammatory bowel diseases and colorectal cancers. We also revealed temporal dynamics of BAF complex cooperation with lineage-specific transcription factors NFIX, ELF3, HNF4A, CDX2, and AP-1 during colonic stem cell renewal and differentiation. Loss of ARID1A impaired stem cell differentiation to both absorptive and secretory lineages. These phenotypes are further inspected by single cell RNA and chromatin accessibility profiles showing loss of open chromatin at cell-type-specific enhancers. Taken together, these studies underscore ARID1A as a crucial regulator of the colonic epithelium and support a pleiotropic disease driving mechanism for ARID1A loss of function in human disease