Health Sciences

Prediction and Assessment of Acute Respiratory Distress Syndrome: Effects of Assumptions in Imputation Methods

Acute respiratory distress syndrome (ARDS) is a life-threatening lung condition that is under-diagnosed in the clinical setting. With a 43% hospitalized mortality rate, it is critical that diagnosis is made in a timely manner for ventilation strategies to be instituted. The Biomedical and Clinical Informatics lab at U-M has been developing a real-time clinical decision support system to detect ARDS through machine learning methods, specifically a modified support vector machine. A clear distinction from other ARDS-related support systems is the use of “privileged information” “” data accessible at the time of training themachine learning model but which is not available in deployment. In the context of this project, we define privileged information to consist of CT scans, which are required for ARDS diagnosis. Such information is not available to the clinician during the early period of a patient’s stay. Therefore, by incorporating such privileged information, we expect the model to lead to faster ARDS diagnosis in clinical practice. This research investigates the electronic health record (EHR) pipeline of the model. With inconsistencies and missing values, some EHR features are absent from one patient while present in another. We aim to improve the model’s accuracy in ARDS detection by testing a unique data imputation method against different machine learning models and settings. Specifically, missing values will be imputed under the assumption that data is not missing at random but as a result of the clinician’s decision that the patient had optimal health in such cases. This method is implemented by imputing data taken from the Michigan Medicine’s established reference ranges when possible. The proposed method will be compared with different machine learning models and feature reduction techniques.

Prediction and Assessment of Acute Respiratory Distress Syndrome: Effects of Assumptions in Imputation Methods Read More »

Promoting Flu Vaccination among Chinese Americans; Providing Flu Education through Trusted Messengers

kellyyan

Objective: The flu is a severe respiratory disease that can cause severe complications, including hospitalization, and is responsible for 290,000-650,000 [6] global deaths per year. The flu vaccine is safe and the best way to prevent the spread of the virus. However, there remains a lot of hesitancy and misinformation regarding the vaccine and the virus itself especially within the Asian community [2]. This study sought to examine reasons for flu vaccination hesitancy as well as to increase influenza knowledge, specifically among the Chinese American community. Methods: Eastern Michigan University Center for Health Disparities Innovations & Studies held 11 CDC funded mobile flu clinics in Michigan in Fall 2020 and provided vaccinations to 337 individuals; during the clinic, bilingual trusted messengers provided flu education. This project reported the results from pre and post flu vaccination surveys with 11 questions about influenza knowledge and demographic information. The results were based on data from 39 Chinese participants. Statistical analyses were conducted in order to measure the efficacy of the flu informational sessions for overall and Chinese participants. Results: The survey data showed that the main reason why the Chinese sub-population had not gotten vaccinated in the past was due to lack of access to the vaccine. Paired t-tests showed that the flu education resulted in significant improvements in overall participants’ knowledge on various aspects about flu, for example, timing of the flu season, vaccination timeline, ways to prevent flu, and others. Conclusion: This study indicated that education on flu and flu vaccination is necessary in order to promote increased vaccination, specifically among Asian communities. Increased education and vaccination will ultimately slow the spread of the virus and improve health outcomes of underserved communities. However, due to the small sample size of the Chinese participants, further research with a larger sample will provide more information about this specific Asian American population. Future intervention efforts can be focused on addressing flu vaccination hesitancy which will provide additional insights for effective strategies to promote COVID-19 vaccination in this population.

Promoting Flu Vaccination among Chinese Americans; Providing Flu Education through Trusted Messengers Read More »

Coronary artery segmentation for automatic stenosis detection

cxinyu

One type of coronary artery disease (CAD), stenosis, is characterized by the narrowing of the coronary arteries, which is the leading cause of death in the United States. One of the most popular ways to diagnose stenosis is the coronary angiogram, from which a doctor can diagnose stenosis by finding places of narrowing in the arteries with the video or the pictures acquired. A lot of studies are currently focusing on automatic coronary artery segmentation, a critical step of a computer-aided system that assists doctors in detecting coronary stenosis. Here we propose a deep learning pipeline using DenseNet-backbone U-Net for coronary artery segmentation in angiogram images, which could be combined with pre-processing and post-processing steps for stenosis detection.

Coronary artery segmentation for automatic stenosis detection Read More »

Development and Assessment of an In-Vehicle Cardiac Monitoring and Severe Event Prediction System

singhaa

Cardiac issues are serious medical issues that affect a large part of the population. Many studies have looked into using programs to automate the identification process and help discover symptoms early without the need for a doctor’s visit. The project focuses on developing machine-learning algorithms that can rapidly detect cardiac issues, specifically arrhythmias, to be implemented directly into motor vehicles. This project involves conducting training and testing of non-parametric machine learning algorithms on the publicly available MIT-BIH Arrhythmia Database. Models are created on MATLAB and tested based on statistics such as accuracy, specificity, and sensitivity through a confusion matrix. So far, decision tree based models have shown accuracies around 90% on smaller datasets after tuning various hyperparameters. Testing is to be continued with different models such as Support Vector Machines (SVM) and Convoluted Neural Networks (CNN) on larger datasets that will ideally result in significant improvements to the final model. These models will be better suited to accurately predict cardiac issues even with the extra signal noise when built into embedded systems within motor vehicles. These systems will add an extra layer of security to vehicles and help identify symptoms earlier than traditionally possible.

Development and Assessment of an In-Vehicle Cardiac Monitoring and Severe Event Prediction System Read More »

Bioinformatics study on protein structures

avapardo

The majority of proteins are composed of foldable, stable subunits called domains. The structures of these proteins can be made up of a single domain or multiple domains. Determining structures of multidomain proteins is a crucial step in elucidating their functions and designing new drugs to regulate these functions. However, it has been largely ignored by the mainstream of computational biology due to the difficulty in modeling inter-domain interactions. Therefore, almost all of the advanced protein structure prediction methods are optimized for modeling single domain proteins. In this study, we presented a method to construct a multidomain protein structure library with known full-length structures to assist the multidomain protein structure prediction. We collect all multidomain proteins from the Protein Data Bank based on the DomainParser, and multidomain proteins defined in CATH and SCOPe databases are also included in the library. This resulted in a total of 15,293 multidomain proteins in the library. The completeness of the library is examined by structurally matching a set of non-redundant multidomain proteins through the library using TM-align. The results show that most of the cases can obtain at least 1 template with correct global fold (TM-score >0.5) from the library, which indicates that the constructed multidomain protein library can likely be used to guide the multidomain protein structure modeling.

Bioinformatics study on protein structures Read More »

Role of the MYRF transcription factor in retinal development

Athera Yakoo

Vision is critical to quality of life. Diseases of the retina, which senses light and sends signals to the brain, can lead to significant vision impairment. One such disease called nanophthalmos affects the overall growth of the eye and can be associated with genetic mutations in the myelin regulatory factor (MYRF) gene. In this project, we use a mouse model to examine the role of the MYRF as a regulator of eye development and disease in the RPE, and to define new genes that could be implicated in retinal disease. Myrf is a transcription factor expressed in the retinal pigment epithelium (RPE), a supporting cell in the retina. We utilized single cell RNA sequencing, scRNA, from mouse eye cups, at varying stages of development to compare gene expression profiles with wild-type eye cups to those loss of Myrf. We used Seurat software for analysis to identify genes that are differently expressed in these two genotypes among different cell types to provide clues into how MYRF impairs eve development. We hypothesize that genes involved in pigmentation and structure of the RPE will be most affected by loss of MYRF, given that mice lacking MYRF have pigmentary defects and develop retina degeneration. To identify which of the genes regulated by MYRF could be implicated in retinal disease, we have also used Clinvar, an online database to search for diseases associated with genes and their allele variants. We have found approximately seventy-seven genes that are likely regulated by MYRF, but not yet linked to human disease, which will now be screened for variants in disease populations. This project will further our understanding of eye disease caused by MYRF mutations, discover genes regulated by mouse Myrf in eye development, and may lead to new targets for treatments.

Role of the MYRF transcription factor in retinal development Read More »

Measurement of Coronary Artery Diameter using Image Processing and Geometric Modeling

dpaylan

While doctors use their eyes to understand and interpret coronary artery data provided by Magnetic Resonance Angiography (MRA) or Computed Tomography Angiography (CTA), researchers have been looking for ways to use technology to automatically track coronary arteries. This document looks to combine two methods of interpreting angiograms, Kalman filtering and a geometric vessel model in order to glean a fuller estimate of a coronary artery’s radius. So far, I have created an algorithm in MATLAB to detect radii across different artery connections. The next step is to compare these radii results to formulae referenced in other research papers and apply the two aforementioned methods to improve the accuracy of my radii measurements. A better estimate of a coronary artery’s radius will prove vital to doctors, who, in relying on their eyes to interpret coronary artery data, might struggle to differentiate noise from weak stenosis requiring treatment. Further, these doctors, though they differ from each other on how they treat varying levels of stenosis, would love to have a standardized way of understanding how severe the stenosis is, and having a more reliable radius will provide exactly that.

Measurement of Coronary Artery Diameter using Image Processing and Geometric Modeling Read More »

Identifying Brain Edema in CT Scans Using Machine Learning

leelydia

Brain edema is the swelling of the brain as a result of traumatic brain injuries, strokes, tumors, and infections. This affects the patients cognitive and motor function and can lead to lasting adverse health risks and death. Early and accurate identification of edema can prevent these hazards. Studies have found that brain edema is difficult for clinicians to accurately identify, as it often blends in with other brain matter. Additionally finding a link between the volume of edema and the effect on the patient is considered valuable, but there is currently no standard software in place for this end. Even when clinicians are able to identify edema, they are not able to quantify the volume present. Convolutional neural networks were used for training the model to segment the edema region. Images from the PROTECT III collection at the University of Michigan hospital were used for this research. Some images were previously annotated by clinicians and these images were subsequently used in the process of training the machine learning model. The performance of the model was evaluated using quantitative techniques such as dice, sensitivity, specificity, accuracy, and AUC. The goal of this software is to decrease adverse effects and death related to brain edema by creating a system to quantitatively measure edema and make informed decisions on how to treat the patient based on the information collected.

Identifying Brain Edema in CT Scans Using Machine Learning Read More »