In condition-based maintenance, preventive replacement threshold and inspection scheme play important roles in maintenance performance. Major research considers cost as the main objective for measuring maintenance performance; here the average cost per unit time is used as the only objective in a single-unit system. The intention of this study was to investigate that how the average cost per unit time varies through changing replacement threshold and inspection scheme, the aim is to simultaneously determine an optimal replacement threshold and inspection scheme. The heterogeneity of hazard rate over stages of the equipment entails less frequent inspections when the equipment is at a healthy condition and more frequent inspections when the equipment is at a progressively deteriorated condition. Therefore the proposed condition-based inspection in the present work is non-periodic. A method based on dynamic programming is developed in order to implement a condition-based inspection scheme; furthermore the threshold for preventive replacement and the inspection scheme are simultaneously determined and then the optimal average maintenance cost is obtained.

Most of the current problems can be solved by referring to the solutions of the previous problems. Case Based reasoning (CBR) is one of the methods that solves a problem by retrieving the similar problems from the past and adapting the solutions of the past problems to solve the new problem. Recent studies that apply CBR include time as a parameter to retrieve most effective solutions that vary with time. This approach is more helpful in healthcare area in which one needs to look at historical evidence to find an accurate diagnostic or treatment regime. Hence, in this study, a time-based CBR is applied to track the outcomes of the drug therapy on hypertensive patients and find the most effective drug as a prescription. Initially, episodes in each patient’s medical records are chronologically ordered such that the oldest episode is placed first in the episode sequence and the latest episode is placed the last. It is assumed that the first episode of each patient is the first instance of diagnose; so when a new patient comes for checkup, his/her state (health condition) is compared with the initial state of the past patients. Therefore, the retrieval process calculates the similarity between the new patient’s current state and the most similar patients at their first episodes in the patient records. Due to the diversity of therapies for matching patients, the best treatment couldn’t be determined without knowing the efficacy of the different treatments. Therefore, the subsequent episodes of matching patients are examined to find the best treatment for the new patient. This might even require using a combination of treatments from all matching patients to find a good treatment for the new patient. After the treatment is defined for the first visit, the record of the new patient is stored in the library for future case retrieval. This method is a novel approach to personalized treatment of patients having chronic disease by tracking the medical records past patients over a long period of time. The current approach for treating the hypertensive patients uses evidence-based guidelines for managing the disease. However, this approach is more general and doesn’t take into account all the patient characteristics such as lab results and physical examination parameters. In the current approach the similarity between patients can’t be leveraged; the change of the treatment regime is based only on the risk parameter. However, in this method several parameters are being checked for efficiency of the medication. In contrast, the proposed CBR-based method personalizes the treatment based on what worked well for similar patients. In this paper, the clinical records of hypertensive patients are provided by a Boston based hospital. The preliminary results confirm that the proposed approach will give good recommendation for hypertension treatment.

In this paper a reinforcement learning algorithm is applied to regulating the blood glucose level of Type I diabetic patients using insulin pump. In this approach the agent learns from its exploration and experiences to selects its actions. In the current reinforcement learning algorithm, body weight, A1C level, and physical activity define the state of a diabetic patient. For the agent, insulin dose levels constitute the actions. There are five alternative actions for the agent: (1) raising the insulin infusion rate during 24 hours, (2) keeping it the same, (3) decreasing insulin infusion rate, (4) adjusting basal rate two times during 24 hours, and (5) adjusting basal rate three times during 24 hours. As a result of a patient’s treatment, after each time step t , the reinforcement learning agent receives a numerical reward depending on the response of the patient’s health condition. At each stage the reward is calculated as a function of the deviation of the A1C from its target value. Since reinforcement learning algorithm can select actions that improve patient condition by taking into account delayed effects it has tremendous potential to control blood glucose level in diabetic patients. This research will utilize ten years of clinical data obtained from a hospital.

Global Learning Charter Public School (GLCPS) is an urban secondary school located in the city of New Bedford, Massachusetts. GLCPS educates students in grades 5–12. It is a Title I school with over 74% of the student population on free and reduced lunch. Historically, only 60% of students graduating from New Bedford move on to postsecondary education. It is the goal of our school to change this and increase the number of students entering post secondary education and more specifically to increase their interest in STEAM (science, technology, engineering, arts, and math) fields. GLCPS provides a unique educational experience where students demonstrate academic excellence and mastery of essential skills. These skills include: technology literacy, public speaking, global citizenship and arts exploration. Incorporation of STEAM (science, technology, engineering, art, and mathematics) is a continued goal for our school. After attending teacher educator training/professional development in engineering-based learning (EBL), we decided to create a robotics course, which fully embedded EBL into the curriculum. The goal of this robotics course is two fold: 1) Combine engineering, math, science, and art/creativity into one course; and 2) engineering-based learning can impact the way students learn STEAM principles, retain STEAM theory, and apply them to real world, relevant applications. The purpose of this paper is to illustrate how engineering-based learning inspired and impacted the development of a robotics course in an urban, financially disadvantaged, secondary charter school. Specifically, we detail how the principles and tools of the engineering-based learning pedagogy affected the development and implementation of this robotics course. Lastly, we will demonstrate how EBL and the robotics course have changed student perceptions of science, engineering, and math.

Standard medical practice is known to have a history of varying definition of “standard”. As with any industry with multiple entities, each entity defines their standards and expectations according to what they believe is their customers’ (i.e. patients) needs and preferences. Recently, our research on developing a predictive wound care assessment methodology and system has extended our study into analyzing evidence-based best practices in wound care clinics. Our research on wound healing predictive model systems, revealed key differences in operational practice between the clinics that were visited in different institutional settings. The scope of this study evaluates our observed wound care practice and wound care treatment to determine if there is a common set of effective practice that can be developed to better standardize care. The purpose of this paper is to compare and contrast the operational practice and procedures at various community and teaching hospitals to determine if there is an ideal combination of tools and standard techniques that would be most beneficial to patient wound care. This paper will focus on methods of patient wound care. We will then present a model of “Evidence-Based Best Practices of Wound Care Assessment” that is based on the observation and interactions with various hospitals.

Chronic, non-healing ulcers are expensive to treat and heal. Approximately 2% of the United States’ general population suffers from non-healing wounds. Conservatively, the cost of treating these chronic wounds is estimated to exceed $50 billion per year. This is approximately ten times more than the annual budget of the World Health Organization. The prevalence of wound healing is similar to that of heart failure and cardiac diseases. However, unlike cardiac diseases little is known regarding the comparative treatments of wounds and their respective outcomes. Additionally, the field of monitoring and tracking wound care lacks the awareness compared with cardiac diseases.

Chronic wound assessment and analysis has long been a major healthcare issue. Chronic wound management and treatment cost billions of dollars each year. The research to alleviate the burden of non-healing wounds and predicting when they will heal is progressing at incremental pace. Characteristics of a chronic wound are unique to both the patient and wound itself. Like a fingerprint, each wound has a unique set of properties that tell a story about its health and viability. Although each person’s wound is individual, there are a few underlying pathologies that are common amongst all wounds. For example, all wounds have a definite surface area, depth, and temperature at any given time. By knowing these common characteristics across all wounds, we can use both historical data and collected data to determine wound healing patterns and wound healing rates. The purpose of this study is to develop an algorithm that uses photography and statistical modeling to predict an approximate wound healing rate for lower appendage wounds. We focus on lower appendage wounds with a depth of 1–2 mm because lower appendage wounds account for approximately 70% of wounds seen at wound clinics.

Wounds are as unique as a fingerprint. A wound can be a tear, scrape or cut — simply anything that damages the protective layer of skin. Wounds occur anywhere on the body from a small paper cut to a large gash. Because the topology of a wound varies so significantly, current tracing and modeling methods fail to adapt to the changing environment. Similarly, traumatic wounds are a greater challenge to heal without infection due to their nature, size, depth and moisture. Presently, tracing and modeling methods are inaccurate and unreliable for wounds that exceed a certain depth. With any wound, non-invasive, full field methods are necessary to effectively measure and assess the severity of the wound. Due to lack of diagnosis options, wounds are expensive to treat and heal. There are few methodologies that exist to help affect the decision for clinicians regarding wound care, how wounds are evaluated, and how wounds are created. In the U.S., there are an estimated 2.5 million pressure ulcers treated each year in acute care facilities. The epidemiology of pressure wounds result in particular physiological principles that we can use to quantify and determine a baseline wound condition. At the moment, the physician and his experience rather than a standard diagnostic system determines the wound condition. This paper presents methods of quantifying wound progression and their impact thus far on patient care. It also provides open-ended research issues that support the need for a structured, systematic methodology for the modeling and analysis of wound progression. The purpose of this paper is also to disseminate a preliminary epistemology on wound progression and modeling in a quantifiable manner.

For many years, literature has documented the benefits of project-based learning (PBL) and its impact on student learning especially at the high school level. More often than not however, students are still losing interest in STEM (Science, Technology, Engineering, and Mathematics) education because current educational teaching pedagogies have become antiquated and are not impacting student learning, as it should. With that said, our discovery through elicitation of high school educators has cited the main reason for such disinterest is due to the inability of students to connect STEM abstract concepts and theory with STEM application to appreciate the value of learning STEM. With access to information easier than ever, students are forgetting that learning is not about getting the right answer but understanding how to solve a complex problem. In the past, PBL has benefited students in engaging them in hands-on learning however, with a more complex paradigm shift in student learning style, PBL and lecture-based learning are no longer the most effective methods of teaching. Engineering-based learning has the opportunity and potential to modify STEM education and revolutionize STEM teaching pedagogy by changing the one-size-fits-all model to an individual, student-centered learning approach where education is mass customized. This paper discusses a new teaching pedagogy dubbed Engineering-Based Learning (EBL) that is a more systematic approach to high school STEM teaching for open-ended problems. This paper presents the EBL model, the EBL tools, and its impact thus far on high school students. It also presents sample feedback from both teachers and students and how it has influenced their outlook of engineering and STEM in the real world. The purpose of this paper is also to disseminate this new teaching pedagogy to support the notion that STEM education can be successfully taught and provide students with a structured, systematic, hands-on approach, as well as the appropriate tools and resources allowing them to connect complex STEM theory and real-world application.

Magnesium oxide (MgO) is a candidate for interface layer application in multifunctional heterostructures. Molecular beam epitaxy (MBE) of MgO on 6H-SiC is studied for effective integration of functional oxides. In this research we present the neural network assisted design of experiment (DOE) analysis of the crystalline structure of MgO films grown over 6H-SiC substrates. Based on reflection high energy electron diffraction (RHEED) characterization of the MgO (111) films grown we present the DOE analyses of the strain propagation in two different orientations (110 and 112) during growth process with respect to growth time, substrate temperature, magnesium source temperature, plasma intensity and percentage oxygen on the starting surface of 6H-SiC substrates. We conducted constant substrate temperature (150°C) and constant plasma intensity (350mV) analyses to quantify the relation between lattice strains in <110> and <112> directions. The constant plasma intensity analysis reveals that the growth propagates in a tensile strain mode for well reconstructed SiC surface. However, excessive oxygen on the SiC surface causes the lattice of MgO twist due to tensile strain as seen from the <110> direction and compressive strain in <112> direction. In constant substrate temperature analysis we found that the principal tensile strain inducing variable in <110> direction is growth time and the strain continues to increase as the film grows thicker. However, in <112> direction the optimum plasma intensity is critical to maintain the optimum strain in the film structure.

Carbon nanotubes (CNTs) have received much attention from both the scientific and industrial communities due to their structural properties and unique morphology. There has also been growing interest in vertically aligned single walled carbon nanotubes (VA-SWNTs) because of their suitability for building devices such as solar cells and nanomembrane. Various methods including chemical vapor deposition (CVD) have been developed for growing VA-SWNTs. Among them is alcohol catalytic CVD which is well known for its economic viability, comprehensive substrates selectivity and good yield of VA-SWNTs. This work studies the length assurance of VA-SWNTs growth by an experimental design and an artificial neural network (ANN) metamodel. Process analysis shows that the interaction between gas flow rate and growth time are the most significant input factors. In addition, with high probability flow rate less than 150 sccm and a growth time of 20 minutes are suitable for the repeatability of medium length VA-SWNTs.

Many educational models are derived from a specific pedagogical stance ultimately manifesting into teacher personal pedagogy based on previous knowledge and inherent instinct. Past educational models like T 4 E emphasize core concepts such as structure of content for the learner, effective and engaging communication, and promotion and encouragement of hands-on learning. However, many of the educational models fall short of showing teachers how to implement new methods in their classrooms, and in providing them with the needed tools. The CAPSULE professional development was created and designed to fill this gap. Further, CAPSULE is designed to address the existing national problem of lack of interest in STEM subjects with high school students. Nevertheless, the problem is not with students but in the delivery method of STEM education. There is a disconnect between abstract STEM concepts and real-world application, therefore a loss of interest by students. Engineering-based learning (EBL) is a new approach that was designed to mitigate this problem by using existing and established tools and apply them to the high school classroom. EBL utilizes the engineering-design process (EDP), computer-aided design (CAD), and the capstone experience to relate theoretical STEM concepts to real-world applications. By leveraging these tools, we have created a unique methodology to take a recognized university capstone course model and modify it for high school STEM courses. The purpose of this study is to contribute to our understanding of how high school students learn and absorb basic STEM principles. Our goal is to disseminate engineering-based learning through our teacher professional development workshop. This paper presents the CAPSULE model and teacher feedback. It also presents sample implementation plans of teachers in their classroom as well as their student feedback.

Carbon nanotubes (CNTs) have received much attention from both the scientific and industrial communities due to their structural properties and unique morphology. There has also been growing interest in vertically aligned single walled carbon nanotubes (VA-SWNTs) because of their suitability for building devices such as hydrogen storage and super capacitors. Various methods including chemical vapor deposition (CVD) have been developed for growing VA-SWNTs. Among them is alcohol catalytic CVD which is well known for its economic viability, comprehensive substrates selectivity and good yield of VA-SWNTs. In order to fully understand the growth mechanism of those CNTs, an examination of the role of inputs like hydrocarbon flow rate, reaction time, chamber temperature, and pressure is essential. This work studies the controllability of VA-SWNTs growth by a hybrid process model of an experimental design and an artificial neural network (ANN). Process analysis shows that CVD pressure and temperature are the most significant input factors. In addition, interactions and response surface plots confirm these results and add that higher temperature and pressure will yield VA-SWNTs with high probability.

School children in general and high school students, in particular more often than not lose interest in STEM (science, technology, engineering, and math) education. Underrepresented and female students are even more discouraged by STEM courses. Our investigation and interviews with high school teachers cite that the main reason for such disinterest is the disconnect between school and reality. Students cannot relate the abstract concepts they learn in physics, biology, chemistry, or math to their surroundings. This paper discusses a new capstone project-based approach that closes this gap. This work is an outcome of an NSF funded project called CAPSULE (Capstone Unique Learning Experience). We use the top-down pedagogical approach instead of the traditional bottom-up approach. The top-down approach relates the abstract concepts to exciting open-ended capstone projects where students are engaged in designing solutions, like products to solve open-ended problems. This top-down approach is modeled after the college-level capstone design courses. The paper presents the model, its details, and implementation. It also presents the formative and summative evaluation of the model after deploying it in the Boston Public Schools, a system heavily populated by the targeted student groups.

Functional properties of thin film metal oxides depend upon their stoichiometric and structural uniformity. Controlling the film deposition process can help tune the functionality of these films by ensuring the control over chemistry and structure of the films. The high volume manufacturing of functional devices will benefit from the development of reliable control models developed from research efforts in designing robust manufacturing processes. The use of neural networks as computer models to simulate the molecular beam epitaxy (MBE) of iron oxide thin films is presented in this work. Monte Carlo experiments are used to study the sensitivities and significances of process control variables to the stoichiometric performance indicators. Moreover, we also explore the relationship between growth dynamics of iron oxide (Fe 2 O 3 , Fe 3 O 4 , and mixed Fe x O y ) and magnesium oxide (MgO) thin films. The common metal adsorption controlled growth mechanism of two films with different structural and stoichiometric complexities were observed and the similarities among the trends of analogous stoichiometric indicators at comparable metal arrival rates of the two films are presented. The dependence of undesirable bonding states of iron and magnesium metals with the film thicknesses was also observed in both processes. The commonalities suggest the potential to use of neural network assisted Monte Carlo analysis to link common atomic-level mechanisms to processing variables in one nano-scale system and use them to predict some level of behavior in other nanoscale processes with similar atomic-level mechanisms.

Prognostics and health management of computer hard disk drives is beneficial from two different angles: it can help computer users plan for timely replacement of HDDs before they catastrophically fail and cause serious data loss; it can also help product recover facilities reuse hard disks recovered from the end-of-life computers for building refurbished computers. This paper presents a HDD health assessment method using Self-Monitoring, Analysis, and Reporting Technology (SMART) attributes. It also presents the state-of-the art results in monitoring the condition of hard disks and offers future directions for distributed hard disk monitoring.

Every machine or equipment in a manufacturing facility is subject to failure due to deterioration based on cumulative wear, crack growth, erosion, etc. This failure will cause production losses and delays resulting in high costs. As the modern manufacturing systems are getting more and more complex, intelligent maintenance schemes must replace the old labor intensive planned maintenance systems to ensure that equipment continues to function. If the maintenance decision is based on the state of the system rather than its age, this leads to the choice of a Condition Based Maintenance (CBM) policy to prevent catastrophic unexpected machine breakdowns and increase the availability of individual machines, but it also introduces randomness into the manufacturing operation. This paper presents a Q-Learning model to dynamically group maintenance actions on different machines and execute them simultaneously, so that one can reduce maintenance cost and increase the efficiency of the manufacturing system.

In this paper we present a neural network-based approach to predict engine stall events 2 seconds prior to their occurrence. The objective of this effort is to use engine and aircraft data to investigate stall events that have occurred on a modern gas turbine engine and to use that data to develop methods to predict stall events probabilistically. Engine data included pre-stall event data for multiple stall events, take-off trend data for the flights when the stalls occurred, and flight data for dozens of flights. The results indicate that neural network-based approach is able predict 99 out of 100 the stall events with a false alarm rate of 12%. The insights gained through this research can be used for developing an elaborate probabilistic model for predicting stall events.

The hard disk drive (HDD) is a critical component of any computer system. The performance of a computer system largely depends on the performance and health of its HDD. This paper investigates degradation signatures for the estimation of remaining useful life and the assessment of health of a HDD. Most of the mechanical faults in a HDD results in head-disk collision or friction. As a HDD ages, it may experience gradual damage to the head and scratches on the disk. One can expect that changes in the condition of head and disk may result in comparable changes in the characteristics of HDD vibration and acoustic emission signals. Based on this premise, this research conducted experiments on HDDs subject to accelerated deterioration. HDDs are monitored through vibration and acoustic emission sensors. Extracting features from these sensor signals, HDD degradation signatures are created. The results indicate that though degradation signatures exhibit a gradual trend with HDD aging, accurate assessment remaining useful life and health are not possible using these degradation signatures. Poor signal to noise ratio is the main impediment in this approach. The conclusion is that the best vibration and acoustic sensors available for this application are neither sensitive nor selective enough to capture the changes in the head and the disk of an aging HDD.