Physical Sciences and Mathematics Commons^™

Finer Details Of Language Modeling: Text Segmentation, Working Within Resource Limits, And Watermarking, Evan Gordon Lucas

Dissertations, Master's Theses and Master's Reports

Language modeling is a vast sub-field of natural language processing and this work focuses on solving some specific problems within that field. Technically, the work falls into a number of sub-categories within natural language processing; how to segment texts, improving sparse transformer performance for summarization tasks, character level models for dialect determination, watermarking of large language models, and a general method of incorporating minimal human feedback for continual or online learning. Despite touching on many small areas, they all connect as being related to the very general problem of handling sequential data. Language and text can be thought of as …

Tracing The Most Powerful Galactic Cosmic-Ray Accelerators With The Hawc Observatory, Dezhi Huang

Dissertations, Master's Theses and Master's Reports

Since Victor Hess's groundbreaking detection of cosmic rays in the Earth's atmosphere in 1912, the origins of these charged particles have remained an enduring mystery. Recent studies suggest that these cosmic rays are accelerated beyond Peta electronvolts by powerful astrophysical sources within our own galaxy. While the cosmic rays themselves are being deflected in all directions by magnetic fields, the gamma rays produced by them, being electrically neutral, travel to the observer in a straight line. They carry crucial information, allowing us to trace cosmic-ray accelerators within our galaxy. The High Altitude Water Chrenkov (HAWC) Observatory, located on the slopes …

An Exploration Of Cloud Droplet Growth By Condensation And Collision-Coalescence In A Convection-Cloud Chamber, Jacob T. Kuntzleman

Dissertations, Master's Theses and Master's Reports

Atmospheric clouds are crucial to weather and climate, and the rate at which droplets collide and coalesce to form precipitation is one of the fundamental controlling processes. The convection-cloud chamber allows the interactions between aerosols and cloud droplets produced by condensation to be investigated within a turbulent environment. Studying the full range of microphysical conditions in atmospheric clouds is not possible, however, unless conditions for droplet growth by collision and coalescence are also achieved. In this study, we explore the conditions favorable to collision-coalescence growth in convection-cloud chambers, extending previous work on steady-state droplet size distributions due to condensation alone. …

Neuromorphic Computing Applications In Robotics, Noah Zins

Dissertations, Master's Theses and Master's Reports

Deep learning achieves remarkable success through training using massively labeled datasets. However, the high demands on the datasets impede the feasibility of deep learning in edge computing scenarios and suffer from the data scarcity issue. Rather than relying on labeled data, animals learn by interacting with their surroundings and memorizing the relationships between events and objects. This learning paradigm is referred to as associative learning. The successful implementation of associative learning imitates self-learning schemes analogous to animals which resolve the challenges of deep learning. Current state-of-the-art implementations of associative memory are limited to simulations with small-scale and offline paradigms. Thus, …

Machine Learning Methods For Prediction Of Human Infectious Virus And Imputation Of Hla Alleles, Xiaoqing Gao

Dissertations, Master's Theses and Master's Reports

This dissertation contains three Chapters. The following is a concise description of each Chapters.

In Chapter 1, we introduced the Random Forest, a machine learning method, to foresee whether a virus is capable of infecting humans. The Covid pandemic informs us the importance of predicting the ability of a zoonotic virus that can infect humans from its genomic sequence. We used the -mer with and as features of a virus to predict if it can affect humans. We further employed the Boruta algorithm to select the important features, then fed those important features into the Random Forest method to train …

Remote Sensing Approach For Terramechanics Applications Utilizing Machine And Deep Learning, Jordan J. Ewing

Dissertations, Master's Theses and Master's Reports

Terrain traversability is critical for developing Go/No Go maps, significantly impacting a mission's success. To predict the mobility of a vehicle over a terrain, one must understand the soil characteristics. In situ measurements performed by soldiers in the field are the current method of collecting this information, which is time-consuming, are only point measurements, and can put soldiers in harm's way. Therefore, this study investigates using remote sensing as an alternative approach to characterize terrain properties.

This approach will explore the relationships between electromagnetic radiation and soil types with varying properties. Optical, thermal, and hyperspectral sensors will be used to …

Types Of Questions Teachers Ask To Engage Students In Making Sense Of A Student Contribution, Nishat B. Alam

Dissertations, Master's Theses and Master's Reports

In the student-centered classroom, a teacher’s interpretation and response to student mathematical contributions plays an important role to shape and direct students’ opportunities for sense-making. This research used a scenario-based survey questionnaire to examine what types of questions middle and high school mathematics teachers indicate they would ask to engage students in making sense of a high-leverage student mathematical contribution and their reasoning about why particular questions are or are not productive. From the results, it could be concluded that teachers asked more productive questions after seeing a set of possible questions. Their beliefs about the productivity of the questions …

Explicit Rule Learning: A Cognitive Tutorial Method To Train Users Of Artificial Intelligence/Machine Learning Systems, Anne Linja

Dissertations, Master's Theses and Master's Reports

Today’s intelligent software systems, such as Artificial Intelligence/Machine Learning systems, are sophisticated, complicated, sometimes complex systems. In order to effectively interact with these systems, novice users need to have a certain level of understanding. An awareness of a system’s underlying principles, rationale, logic, and goals can enhance the synergistic human-machine interaction. It also benefits the user to know when they can trust the systems’ output, and to discern boundary conditions that might change the output. The purpose of this research is to empirically test the viability of a Cognitive Tutorial approach, called Explicit Rule Learning. Several approaches have been used …

Comprehensive Analysis Of Seismic Signals From Pacaya Volcano Using Deep Learning Event Detection, Jessica L. Devlieg

Dissertations, Master's Theses and Master's Reports

Pacaya volcano located 30 km SW of Guatemala City, Guatemala, has been erupting intermittently since 1961. Monitoring of seismicity is crucial to understanding current activity levels within Pacaya. Traditional methods of picking these small earthquakes in this noisy environment are imprecise. Pacaya produces many small events that can easily blend in with the background noise. A possible solution for this problem is a machine learning program to pick first arrivals for these earthquakes. We tested a deep learning algorithm (Mousavi et al., 2020) for fast and reliable seismic signal detection within a volcanic system. Data from multiple deployments were used, …

Design And Implementation Of A Graphql Mesh Gateway: Federating Api Endpoints Based On A Defined Data Model, Marcus D. Scese

Dissertations, Master's Theses and Master's Reports

This paper introduces the GraphQL Mesh federated API (Application Programming Interface) gateway project, a comprehensive initiative implemented using GraphQL Mesh to solve data related issues within the USW-DSS (Undersea Warfare - Decision Support System). The project contributes to the evolving discourse on the pivotal role of Data Fabrics and Data Meshes in dismantling the barriers imposed by digital data silos. The project is a collaboration between researchers at Michigan Technological University, and engineers at ARiA (Applied Research in Acoustics LLC). The aim of the project is to resolve difficulties in understanding a large collection of API endpoints. By navigating the …

Fabrication And Optical Properties Of Two-Dimensional Transition Metal Dichalcogenides, Manpreet Boora

Dissertations, Master's Theses and Master's Reports

Two-dimensional layered materials such as graphene and transition metal dichalcogenides have gained a lot of attention because of their distinctive chemical, optical, and electronic properties. Transition metal dichalcogenides can have a tunable bandgap in the range of 1 − 3 eV (visible), which enables their applications in ultrathin field effect transistors, photovoltaics, sensors, and optoelectronic devices. They exhibit an indirect-to-direct band gap transition when thinned down from bulk to a single layer. They show strong photoluminescence, electron–photon interaction, valley pseudospin, nonlinear optical response, and lack of dangling bonds. Many exotic phenomena appear when materials are thinned to nanoscale size because …

An Investigation Of Microseismicity During The 2018 Kilauea Caldera Collapse, Isabella S. Metts

Dissertations, Master's Theses and Master's Reports

The 2018 Kīlauea volcano eruption and incremental caldera collapse was accompanied by more than 60,000 seismic events cataloged by the Hawaiian Volcano Observatory as well as 62 caldera collapse events that occurred roughly once per day. The majority of seismicity occurred on the eastern side of the caldera between daily collapses. However, the majority of caldera subsidence occurred to the west. To understand the collapse mechanics behind this variance in subsidence and seismicity across the caldera region, repetitive waveforms and source properties can be studied.

Repeating seismic events suggest a common source that is not moving or destroyed. At Kīlauea, …

The Impact Of Pre-Experiment Walking On Distance Perception In Vr, Soheil Sepahyar

Dissertations, Master's Theses and Master's Reports

While individuals can accurately estimate distances in the real world, this ability is often diminished in virtual reality (VR) simulations, hampering performance across training, entertainment, prototyping, and education domains. To assess distance judgments, the direct blind walking method—having participants walk blindfolded to targets—is frequently used. Typically, direct blind walking measurements are performed after an initial practice phase, where people become comfortable with walking while blindfolded. Surprisingly, little research has explored how such pre-experiment walking impacts subsequent VR distance judgments. Our initial investigation revealed increased pre-experiment blind walking reduced distance underestimations, underscoring the importance of detailing these preparatory procedures in research—details …

A Search For Compact Object Dark Matter In The Universe Utilizing Gravitational Millilensing Of Gamma-Ray Bursts, Oindabi Mukherjee

Dissertations, Master's Theses and Master's Reports

In this dissertation, I explore the existence of Compact Object (CO) Dark Matter (DM), examining its predicted millilensing impact on the light curves of Gamma-Ray Bursts (GRBs). The hypothesis under investigation is that if CO DM exists, it will act as a gravitational lens, influencing light emanating from GRBs and creating a measurable gravitational echo. The detection of this echo is conditional on several factors, including the spatial alignment of the observer, lens, and source and the mass of the CO DM.

Recent studies have reported the potential detection of millilensing in several GRBs, including GRB 950830, GRB 081122A, GRB …

Multilevel Computational Investigation Into The Catalytic Mechanisms Of Matrix Metalloproteinase-1 And Fat Mass And Obesity-Associated Enzyme, Ann Varghese

Dissertations, Master's Theses and Master's Reports

Enzymes are biological macromolecules, typically proteins, that efficiently accelerate the rate of chemical reactions. Their remarkable catalytic power plays a vital role in essential processes across all kingdoms of life. Nowadays, computational chemistry methods provide valuable insights into enzymatic functions aiding our understanding of biological processes and providing advancements in fields such as biomedical sciences, biomimetic catalysis, drug design and biotechnology. This dissertation employs multilevel computational chemistry methods to investigate the structure-function relationships and the catalytic mechanisms of two metalloenzymes -Zn(II)-dependent matrix metalloproteinase-1 (MMP-1) and non-heme Fe(II)/2-oxoglutarate (2OG) dependent fat-mass and obesity-associated (FTO) enzyme. Chapter 2 explores the role of …

An Ambiguous Technique For Nonvisual Text Entry, Dylan C. Gaines

Dissertations, Master's Theses and Master's Reports

Text entry is a common daily task for many people, but it can be a challenge for people with visual impairments when using virtual touchscreen keyboards that lack physical key boundaries. In this thesis, we investigate using a small number of gestures to select from groups of characters to remove most or all dependence on touch locations. We leverage a predictive language model to select the most likely characters from the selected groups once a user completes each word.

Using a preliminary interface with six groups of characters based on a Qwerty keyboard, we find that users are able to …

Investigating Collaborative Explainable Ai (Cxai)/Social Forum As An Explainable Ai (Xai) Method In Autonomous Driving (Ad), Tauseef Ibne Mamun

Dissertations, Master's Theses and Master's Reports

Explainable AI (XAI) systems primarily focus on algorithms, integrating additional information into AI decisions and classifications to enhance user or developer comprehension of the system's behavior. These systems often incorporate untested concepts of explainability, lacking grounding in the cognitive and educational psychology literature (S. T. Mueller et al., 2021). Consequently, their effectiveness may be limited, as they may address problems that real users don't encounter or provide information that users do not seek.

In contrast, an alternative approach called Collaborative XAI (CXAI), as proposed by S. Mueller et al (2021), emphasizes generating explanations without relying solely on algorithms. CXAI centers …

Novel Bayesian Neural Networks And Uncertainty Quantification Of Computational Mechanics Models, Ponkrshnan Thiagarajan

Dissertations, Master's Theses and Master's Reports

Computational and data-driven models suffer from a wide range of uncertainties that impact the reliability of such models. Given the exponential proliferation of machine learning models in real-world systems, establishing a degree of confidence in their predictions becomes paramount. Reliability in predictions takes on utmost significance in domains such as autonomous driving, medical image analysis, etc., where human lives are involved, and inaccuracies in predictions could lead to disastrous outcomes. For these reasons, comprehending and quantifying uncertainties in computational and data-driven models is of utmost importance. A number of techniques have been developed to quantify uncertainties in machine learning models. …

Physical, Optical, And Chemical Properties Of Light Absorbing Aerosols And Their Climatic Impacts, Susan Mathai

Dissertations, Master's Theses and Master's Reports

Aerosols are particles suspended in the atmosphere; they are emitted during natural phenomena such as dust storms, wildfires, and volcanic eruptions, and during anthropogenic activities like household wood burning, vehicles operations, and industrial productions, or they can form in the atmosphere from gas to particle partition. Aerosols impact earth’s weather and climate by absorbing and scattering the incoming solar and the outgoing earth thermal radiation and interacting with clouds. The optical properties of aerosols evolve as the chemical and physical properties vary during their residence in the atmosphere. In addition, the aerosols’ properties strongly depend on the vertical distribution in …

Large Cloud Droplets And The Initiation Of Ice By Pressure Fluctuations: Molecular Simulations And Airborne In-Situ Observations, Elise Rosky

Dissertations, Master's Theses and Master's Reports

From the molecular dynamics of water molecules to the global processes that control atmospheric circulation: describing the evolution of atmospheric clouds necessitates physics that spans vast spatial scales. In this work, steps are taken towards connecting the molecular physics of ice-nucleation to the growth and subsequent freezing of cloud droplets in convective cumulus clouds. Motivated by experimental evidence of ice nucleation triggered by agitation and distortion of a water droplet surface, the first study in this dissertation uses molecular dynamics simulations to demonstrate that negative Laplace pressure from a curved water surface leads to heterogeneous ice nucleation at higher temperatures. …

Deep Learning For Medical Image Segmentation Using Prior Knowledge And Topology, Chen Zhao

Dissertations, Master's Theses and Master's Reports

Image segmentation refers to the division of a digital image into distinct segments or groups of pixels/voxels. However, most of the existing deep learning approaches lack the utilization of prior knowledge, such as shape information, which could improve segmentation accuracy. In addition, conventional image segmentation frequently falls short in preserving intricate spatial details, motivating the innovation of strategies for multi-scaled feature integration. Furthermore, traditional image segmentation methods primarily concentrate on pixel-level or region-level analysis. However, given the inherent morphological similarities among various image objects, the significance of topology information surpasses that of pixel-level data in the realm of medical image …

Surface Reconstruction In Iron Garnets, Sushree Dash

Dissertations, Master's Theses and Master's Reports

This dissertation presents the results of a study investigating the physical mechanisms underlying an unexpectedly large increase in magneto-optic efficiency observed in iron garnet. Such materials are technologically important for telecommunications due to their nonreciprocal optical action. In the past, our group had found evidence of an enhanced Faraday rotation in bismuth-substituted iron garnet films less than 50 nm thick. Subsequent investigation revealed that this enhancement could be traced to surface effects. This is significant because understanding these phenomena could be used to formulate engineering solutions for device miniaturization. In this dissertation, we present the result of a research project …

Predicting The Reactivities And Reaction Mechanisms Of Photochemically Produced Reactive Intermediates, Benjamin Barrios Cerda

Dissertations, Master's Theses and Master's Reports

Photochemically produced reactive intermediates (PPRIs) such as the hydroxyl radical, carbonate radical (CO₃^•-) singlet oxygen (¹O₂) and triplet state of chromophoric dissolved organic matter (³CDOM^*) are formed in sunlit natural waters upon photoexcitation of chromophoric dissolved organic matter (CDOM). PPRIs react with the organic compounds involved in key environmental processes, resulting in transformation products of smaller molecular weight than their parent compounds. Photochemical transformation of these key water constituents due to their reactions with PPRIs may pose potential effects on human and aquatic ecosystems. Consequently, there is a need …

Recurrent Event Data Analysis With Mismeasured Covariates, Ravinath Alahakoon Mudiyanselage

Doctoral Dissertations

"Consider a study with n units wherein every unit is monitored for the occurrence of an event that can recur with random end of monitoring. At each recurrence, p concomitant variables associated to the event recurrence are recorded with q (q ≤ p) collected with errors. Of interest in this dissertation is the estimation of the regression parameters of event time regression models accounting for the covariates. To circumvent the problem of bias and consistency associated with model's parameter estimation in the presence of measurement errors, we propose inference for corrected estimating functions with well-behaved roots under additive measurement errors …

Sc-Matrl: Semi-Centralized Multi-Agent Transfer Reinforcement Learning, Ayesha Siddika Nipu

Graduate Theses/Dissertations

Distributed decision-making in multi-agent systems (MAS) poses significant challenges for interactive behavior learning in both cooperative and competitive environments. While reinforcement learning (RL) has shown great success in single-agent domains like Checkers, Chess and Go, researchers are motivated to extend RL to MAS. However, as the number of agents increases, effectively dealing with each agent becomes increasingly complex. To mitigate the resulting complexity, a semi-centralized Multi-Agent Influence Dense Reinforcement Learning (MAIDRL) algorithm was previously developed, enhancing agent influence maps to facilitate effective multi-agent control in StarCraft Multi-Agent Challenge (SMAC) scenarios. While MAIDRL shows improved performance in homogeneous multi-agent scenarios, it …

An Explainable Deep Learning Prediction Model For Severity Of Alzheimer's Disease From Brain Images, Godwin O. Ekuma

Graduate Theses/Dissertations

Deep Convolutional Neural Networks (CNNs) have become the go-to method for medical imaging classification on various imaging modalities for binary and multiclass problems. Deep CNNs extract spatial features from image data hierarchically, with deeper layers learning more relevant features for the classification application. The effectiveness of deep learning models are hampered by limited data sets, skewed class distributions, and the undesirable "black box" of neural networks, which decreases their understandability and usability in precision medicine applications. This thesis addresses the challenge of building an explainable deep learning model for a clinical application: predicting the severity of Alzheimer's disease (AD). AD …

Magmatic Evolution Of The Chasca Orkho Lava Series And Its Magmatic Enclaves, Volcán Ollagüe, Central Andes, Nathaniel W. Lenhard

Graduate Theses/Dissertations

Magma mixing is a common factor in the creation of intermediate composition magmas and a potential instigator of a volcanic eruption. Magmatic enclaves, physical evidence of magma mixing within a volcanic system, are a phenomenon whose mechanisms remain unclear and debated. Common hypotheses explaining the occurrence of magmatic enclaves within a host lava range from the repeated injection of a new magma into a shallow reservoir to the disruption of equilibrium within a stratified magma chamber. Within the Andean Central Volcanic Zone (CVZ), the occurrence of magmatic enclaves yields similar geochemical compositions to their respective host rocks, bringing more difficulty …

Recent Climate Change Influence On Flood Magnitude And Frequency, Channel Widening, And Bar Deposition In Big River, Missouri Ozarks, Patrick Saulys

Graduate Theses/Dissertations

Climate and land use can control the flood regime of a river and regulate channel form and size over periods of decades to centuries. Recent climate change has increased rainfall intensity and flood magnitude/frequency in many watersheds in the midwestern United States. Thus, river channels affected by more frequent and larger floods are expected to respond by increasing width (or depth) by fluvial erosion to accommodate larger peak discharges. This hypothesis was evaluated along 186 km of the Big River in southeast Missouri by analysis of historical aerial photography, precipitation studies, and United States Geological Survey (USGS) flow gage records. …

เครื่องสร้างปริภูมิสถานะสำหรับโมเดลเช็กกิงแบบแอลทีแอลโดยใช้แอลเอสทีเอ็มที่มีการจำแนกแบบหลายป้าย, ชลิกา ศักดิ์สุภาวัฒนกุล

Chulalongkorn University Theses and Dissertations (Chula ETD)

กระบวนการทวนสอบเชิงรูปนัยเป็นสิ่งที่สำคัญในการตรวจสอบความปลอดภัย ความดำเนินชีวิตและความสัมพันธ์ของระบบซอฟต์แวร์ตั้งแต่เริ่มต้นออกแบบซอฟต์แวร์ ซึ่งสามารถช่วยให้ลดความพยายามในการพัฒนาอย่างเห็นได้ชัด โมเดลเช็กกิงเป็นวิธีการเชิงรูปนัยอย่างหนึ่งที่สามารถทวนสอบระบบซอฟต์แวร์ตามคุณสมบัติเชิงพฤติกรรมที่เขียนเป็นสูตรตรรกศาสตร์เชิงเวลาแบบลำดับหรือเรียกว่าสูตรแอลทีแอล โดยใช้วิธีการสำรวจเส้นทางดำเนินงานที่เป็นไปได้ทั้งหมดอย่างละเอียดถี่ถ้วนในปริภูมิสถานะของระบบซอฟต์แวร์และกำหนดว่าเส้นทางดำเนินงานทั้งหมดของระบบเป็นไปตามคุณสมบัติที่เขียนโดยใช้สูตรแอลทีแอลหรือไม่ อย่างไรก็ตาม ระบบที่มีขนาดใหญ่และมีความซับซ้อนอาจทำให้เกิดการระเบิดของปริภูมิสถานะในการทำโมเดลเช็กกิงได้ ซึ่งมีการเสนอวิธีการลดพื้นที่ของปริภูมิสถานะหลายวิธีและแม้แต่วิธีการของแมชชีนเลิร์นนิงก็ถูกนำมาใช้เพื่อทำนายผลลัพธ์ความพึงพอใจของคุณสมบัติแอลทีแอลของระบบซอฟต์แวร์ เมื่อพิจารณาถึงความสำเร็จในการฝึกอบรมโครงข่ายความจำระยะสั้นแบบยาวหรือแอลเอสทีเอ็มสำหรับการทำนายลำดับข้อมูลอนุกรมเวลาแบบยาว ในวิทยานิพนธ์นี้ ได้เสนอแนวคิดแอลเอสทีเอ็มที่ขยายการจำแนกแบบหลายป้ายและกลยุทธ์การหาขีดแบ่งเพื่อบรรเทาปัญหาการระเบิดของปริภูมิสถานะและสร้างเป็นเครื่องสร้างปริภูมิสถานะสำหรับสร้างเส้นทางดำเนินงานของระบบซอฟต์แวร์ได้ตามที่ต้องการ โดย ได้สร้างเครื่องมือในการทวนสอบคุณสมบัติพื้นฐานของระบบซอฟต์แวร์ ได้แก่ คุณสมบัติด้านความปลอดภัย ด้านความดำเนินชีวิตและด้านความสัมพันธ์ ด้วยเทคนิคออน-เดอะ-ฟลาย

การเรียนรู้เชิงลึกสำหรับเข้ารหัสบัญชีการจำแนกโรคระหว่างประเทศ (ไอซีดี) จากบันทึกเวชระเบียนโดยใช้คลังข้อมูลภาษาไทยและภาษาอังกฤษ, ณัฐชา ซาซุม

Chulalongkorn University Theses and Dissertations (Chula ETD)

งานวิจัยนี้เสนอแบบจำลองสำหรับจำแนกรหัสไอซีดีแบบหลายฉลากเพื่อเป็นตัวช่วยจำแนกรหัสไอซีดี การให้รหัสไอซีดีไม่ครบส่งผลให้โรงพยาบาลไม่ได้รับเงินค่าชดเชยอย่างเหมาะสม เราจึงมุ่งเน้นที่จะช่วยโรงพยาบาลให้รหัสไอซีดีอย่างครบถ้วนในขั้นตอนการเบิกจ่ายเงินค่ารักษาพยาบาลซึ่งผลลัพธ์ที่ได้คือการสนับสนุนทางการเงินแก่โรงพยาบาล โดยปกติแล้วปัญหาการให้รหัสไอซีดีแบบหลายฉลากเป็นชุดข้อมูลบันทึกเวชระเบียนที่มีการกระจายตัวแบบหางยาวซึ่งไม่ควรละทิ้งข้อมูลใด ๆ ดังนั้น การจำแนกแบบหลายฉลากของเราจึงเป็นการรวมกันของวิธีการรวมแบบจำลองการเรียนรู้เชิงลึกทั้งสาม (Bidirectional Long-short term memory, Convolutional neural network and Transformer encoder) ด้วยการเลือกค่าทำนายสูงสุดและแบบจำลองทางสถิติคือ Multinomial Naïve Bayes ร่วมกับวิธี Binary Relevance ในส่วนของแบบจำลองการเรียนรู้เชิงลึกจะรับผิดชอบกลุ่มรหัสไอซีดีทั่วไป (โรคที่พบบ่อย) ในขณะที่แบบจำลองทางสถิติจะจัดการกับกลุ่มรหัสไอซีดีพบยาก (โรคที่ผู้ป่วยไม่ค่อยเป็น) ผลการทำนายรหัสไอซีดี-10 จากบันทึกการรักษาด้วยแบบจำลองนี้ให้ค่า Jaccard index ที่ 0.792 สำหรับกลุ่มรหัสไอซีดีทั่วไปและที่ 0.205 สำหรับกลุ่มรหัสไอซีดีพบยาก ส่วนของผลการทำนายรหัสไอซีดี-9 จากบันทึกการให้ยา (หัตถการรักษา) ให้ค่า Jaccard index ที่ 0.963 สำหรับกลุ่มรหัสไอซีดีทั่วไปและที่ 0.201 สำหรับกลุ่มรหัสไอซีดีพบยาก ซึ่งแบบจำลองนี้มีความสามารถในการแนะนำรหัสไอซีดีที่ขั้นตอนการเบิกจ่ายเงินค่ารักษาพยาบาล