Dartmouth Institute for Health Policy and Clinical Practice
Dartmouth Digital Commons (Dartmouth College)Not a member yet
8214 research outputs found
Sort by
HeartFEV1: A MOBILE ELECTROCARDIOGRAM BASED SYSTEM FOR INFERRING FORCED EXPIRATORY VOLUME IN ONE SECOND FROM PATIENTS WITH CHRONIC OBSTRUCTIVE PULMONARY DISEASE
Get PDFChronic Obstructive Pulmonary Disease (COPD), characterized by chronic airway inflammation and airflow obstruction, is the third leading cause of death globally. Patients with COPD experience exacerbated symptoms like breathlessness and cough, significantly impacting their quality of life and leading to costly hospitalizations. Early detection of COPD exacerbations is crucial for mitigating these negative effects.
The most critical element for early detection of COPD exacerbations is daily monitoring of lung function, particularly forced expiratory volume in one second (FEV1), a key metric of lung function. By tracking declines in FEV1, COPD exacerbations can be predicted up to two weeks in advance, allowing for timely interventions and potentially reducing hospital admissions. However, the gold standard for remote lung function monitoring, at-home spirometry using a handheld spirometer, requires specialized hardware and a physically demanding maneuver, making it difficult and unreliable for daily monitoring.
This thesis introduces HeartFEV1, a novel system that addresses these challenges by utilizing readily available mobile electrocardiogram (ECG) signals acquired during quiet breathing for FEV1 estimation. To achieve this, HeartFEV1 utilizes an ensemble approach, combining two models: a machine learning model and a residual neural network. The machine learning model uses features extracted from ECG-derived respiratory signals to predict FEV1, whereas the residual neural network is a deep learning approach that directly estimates FEV1 from ECG signals. By averaging the predictions from both models, the HeartFEV1 system achieves accurate FEV1 prediction. Using a dataset of twenty-five patients with obstructive lung disease, the HeartFEV1 system demonstrates a strong correlation (r = 0.78, R^2 = 0.61) with hospital-grade spirometry for FEV1 estimation. Additionally, it reports a low mean absolute percentage error (MAPE) of 20.77% and minimal bias (0.01) in FEV1 predictions.
These findings highlight the feasibility of using mobile ECG signals collected during quiet breathing as a convenient method for daily lung function monitoring
THE FALL OF THE FEMALE HEAD COACH IN THE POST TITLE IX ERA — WILL SHE RISE AGAIN?
Get PDFIn the United States, the percentage of women in leadership positions is rising across nearly all professions, except for women in athletics. While Title IX dramatically increased the number of opportunities for female student-athletes, it was the catalyst behind female coaches\u27 decline in sport. This study aims to explore published studies, current events, and case studies that examine the barriers women face when pursuing coaching as a full-time profession. The main obstacles studied include the unintended consequences of Title IX, current societal and sociological structures, the role of bias in the under-representation of women in leadership positions, the exorbitant cost and exclusivity associated with coaching certifications and professional development (specifically in soccer), and public expenditure (or lack thereof) on women’s sport and gender inequity among recipients of that public expenditure.
A systemic review of published studies conducted in this field indicates that while society may be on the verge of tearing down discriminatory hiring practices and bridging the gender gap in leadership across athletics, it is not quite there yet. The main objective of this research is to create awareness around the barriers and obstacles women face when pursuing a career in sports, as well as encourage female-supportive programming that will improve the experience of young female coaches as they navigate a male-dominated industry
Germanium-Tin on Silicon for Integrated Photonics and Integrated Quantum Materials
Get PDFGroup IV GeSn alloys are attracting attention due to their compatibility with the complementary metal-oxide-semiconductor (CMOS) process. On one hand, Ge-rich GeSn alloys with a tunable direct bandgap are well-suited to infrared (IR) photonic applications such as image sensors. On the other hand, Sn-rich GeSn alloys in diamond cubic α phase are topological quantum materials (TQM) holding potential for important quantum applications. However, directly growing GeSn on Si remains challenging due to the lattice mismatch. Regular epitaxial GeSn grown on a Ge buffer layer is not applicable to many photonic applications including CMOS image sensors (CIS) because the buffer layer prevents the direct transfer of photoelectrons from GeSn to Si storage wells. Meanwhile, recent theories predicted that short-range order (SRO) exists in GeSn and can impact its band structure, but characterizing SRO is difficult due to the requirement of atomic resolution. Therefore, to tackle these challenges, this thesis introduces innovative approaches for the growth, characterization, and integration of GeSn on Si, culminating in the prototype of the first GeSn/Si IR CIS.
We first demonstrate the direct growth of crystallized Ge-rich GeSn on Si by physical vapor deposition (PVD) and construct a meta-stable phase diagram for GeSn thin films to guide the design and fabrication of GeSn devices. Two novel methods based on a Ge seed layer or Ge doping are then presented to grow Sn-rich GeSn on Si, for integrated TQM. Next, a new algorithm based on Poisson statistics is developed to achieve three-dimensional nano-scale mapping of SRO in GeSn using atom probe tomography (APT). Surface termination and precursors are investigated to control SRO towards bandgap engineering. At last, photoresponse from a 32x32 GeSn/Si CIS pixel array at 1310-1854 nm wavelength is demonstrated under thermoelectric cooling at -60 or -65 oC for the first time using back-end-of-line (BEOL) processing, paving the path towards further optimization. Overall, these advancements in GeSn on Si bring exciting opportunities for integrated photonics and integrated quantum materials
Terrestrial Exchange of Atmospheric Metals: Insights from Fallout Radionuclides
Get PDFForests mediate the exchange of gases and particulate matter (PM2.5 and PM10) between terrestrial ecosystems and the atmosphere over 30% of global land area. Because forest foliage efficiently absorbs PM with persistent pollutants including metals Pb and Hg, as well as CO2 and gaseous elemental mercury (GEM), this exchange profoundly influences the composition of the atmosphere as well as terrestrial biogeochemical cycles. The processes by which PM is absorbed remain enigmatic, however, due to the complexity of micrometeorological physics and submicron physical scale of the interaction. Here, measurements of fallout radionuclides (FRNs) beryllium-7 and lead-210, which are quintessential tracers of PM but also radioactive and decay with known rates, provide insights into fundamental questions regarding the role of wet vs dry process in PM deposition, the strength of PM retention by forest canopies, and the timescales over which PM metals are cycled to underlying soils. I combine long-term timeseries of FRNs and trace metals in bulk wet deposition, paired openfall and throughfall event-based deposition, and foliage collections, with annual litterfall and whole-tree mass balance, to describe processes and timescales that govern atmospheric metal dynamics in forest canopies. FRNs and metals accumulate efficiently and permanently in live and senesced vegetation, coupling them to the fate of organic matter. Foliar uptake occurs primarily through wet deposition (~80%), but efficiencies of wet (55%) and dry absorption (53%) by the canopy are similar. While the FRNs 7Be and 210Pb are fractionated at the single-leaf scale, with a deficit of 7Be possibly attributable to hyperacidity of dry-deposited PM, there is no discernible fractionation at the whole-canopy scale due to buffering capacity of the canopy and surfeit of organic surfaces for absorption. FRNs and metals accumulate in the canopy to the equivalent of many decades of deposition, primarily in non-foliar surfaces including lichen, moss, mold, and bark (collectively phyllosphere). These inventories are slowly recycled to underlying soils in association with dissolved organic carbon (DOC) and fine particulate organic matter (FPOM). Cumulatively the forest has a long memory with respect to atmospheric deposition, storing vast quantities of legacy pollutants for multi-decadal timescales, and thereby strongly influencing biogeochemical cycles of critical pollutants
Under the Influence: The Dark Psychology Behind the Power of Social Conformity and Obedience
Get PDFMy lesson, titled Under the Influence: The Dark Psychology Behind the Power of Social Conformity and Obedience intends to inform my students about the darker aspects of social psychology and the deeper, unconscious social influences that can lead good people to do terrible things in moments of conformity, obedience and group think. This course will cover professional laboratory experiments, real life historical examples, and psychological methods and tactics to give a broad range of information regarding this aspect of social psychology. With this information, the students will then have enough material to engage in group discussions regarding the deeper themes of human psychology, morality and ethics and reflect on their own social lives and their place in organizations and institutions
Embracing Your Curly Hair: History, Confidence, and Care
Get PDFThis lesson focuses on empowering students with curly or wavy hair to understand and care for their natural curls. We begin by exploring the history and cultural context of curly hair, highlighting its journey from stigmatization to celebration. Students will then learn to identify their specific curl type and understand the role of porosity in hair care, allowing them to make informed product choices. Practical, hands-on sections will guide students through selecting the right products and using basic styling techniques tailored to their hair type. By the end of the lesson, students will create a personalized hair care plan, gaining both confidence in their hair care routine and a deeper appreciation for the beauty of their natural curls
A framework for evaluating the security and privacy of smart-home devices, and its application to common platforms
Get PDFIn this article, we outline the challenges associated with the widespread adoption of smart devices in homes. These challenges are primarily driven by scale and device heterogeneity: a home may soon include dozens or hundreds of devices, across many device types, and may include multiple residents and other stakeholders. We develop a framework for reasoning about these challenges based on the deployment, operation, and decommissioning life cycle stages of smart devices within a smart home. We evaluate the challenges in each stage using the well-known CIA triad—Confidentiality, Integrity, and Availability. In addition, we highlight open research questions at each stage. Further, we evaluate solutions from Apple and Google using our framework and find notable shortcomings in these products. Finally, we sketch some preliminary thoughts on a solution for the smart home of the near future
Generative AI Text to Query Tool
No full textOur sponsor, Equitus, an artificial intelligence company based in Clearwater, Florida, is committed to making vast amounts of information and data easily accessible to its customers. Equitus enables users to input various document types into its platform to receive relevant data in visualized graphic form. To enhance their platform further, Equitus aims to incorporate data from Wikidata, a structured database with extensive information. However, accessing Wikidata efficiently requires using SPARQL, a complex and uncommon coding language, posing a barrier for usage. In light of this obstacle, our team was tasked with developing a solution: a generative AI tool capable of transforming natural language questions into valid SPARQL queries. Our proposed machine learning model performs with a BLEU score of 25.21, signifying its potential in accurately translating natural language into SPARQL queries. The finalized model has the ability to process natural language input, ultimately eliminating the need for users to possess prior knowledge of the SPARQL coding language to access the data available from Wikidata. The solution is designed to seamlessly integrate into Equitus\u27 platform to enable users to effortlessly access and utilize the data in Wikidata for informed decision making. This tool not only lowers the barrier to entry but also democratizes access to data, driving innovation and informed decision making.With our machine learning model, Equitus customers can confidently navigate and extract valuable insights from Wikidata\u27s collection of structured data
