Given the large quantity of diverse, heterogeneous data in a typical patient record, users spend much of their time and effort finding relevant information to help accomplish their tasks. One of the greatest problems in today’s healthcare environment is matching the increased capability of gathering patient data with a comparable ability to understand, analyze, and act rationally upon this information. This dissertation attempts to bridge this gap by presenting methods for creating context-sensitive visualizations using graphical disease models. Building upon past efforts in Bayesian belief networks and explanation generation, this work explores how the model’s variables, defined relationships, and probabilities are used to identify which data elements in the patient record are important and how the information should be presented for a particular context. A data model, called the visual dictionary, is used to integrate contextual information from graphical disease models and other knowledge sources (e.g., ontologies and user/task models) to generate instructions for laying out patient data in a graphical user interface. These concepts are implemented in two separate applications that demonstrate how context-sensitive visualizations can: 1) be applied towards helping users query large biomedical repositories; and 2) generate an integrated, longitudinal view of a multimedia patient record. The applications were used to evaluate the feasibility of using graphical disease models to retrieve relevant documents and to obtain feedback on the adaptive interfaces through pilot usability studies. Initial results from the pilot studies were positive overall. Developing context-sensitive visualizations that facilitate users with querying these models and understanding the results is a significant step towards using collected data to improve patient care at the bedside.

Objective: To summarize significant contributions to sensor, signal, and imaging informatics published in 2016. Methods: We conducted an extensive search using PubMed® and Web of Science® to identify the scientific contributions published in 2016 that addressed sensors, signals, and imaging in medical informatics. The three section editors selected 15 candidate best papers by consensus. Each candidate article was reviewed by the section editors and at least two other external reviewers. The final selection of the six best papers was conducted by the editorial board of the Yearbook. Results: The selected papers of 2016 demonstrate the important scientific advances in management and analysis of sensor, signal, and imaging information. Conclusion: The growing volume of signal and imaging data provides exciting new challenges and opportunities for research in medical informatics. Evolving technologies provide faster and more effective approaches for pattern recognition and diagnostic evaluation. The papers selected here offer a small glimpse of the high-quality scientific work published in 2016 in the domain of sensor, signal, and imaging informatics.

PURPOSE: While Lung Cancer Screening CT is being performed at low doses, the purpose of this study was to investigate the effects of further reducing dose on the performance of a CAD nodule-detection algorithm. METHODS: We selected 50 cases from our local database of National Lung Screening Trial (NLST) patients for which we had both the image series and the raw CT data from the original scans. All scans were acquired with fixed mAs (25 for standard-sized patients, 40 for large patients) on a 64-slice scanner (Sensation 64, Siemens Healthcare). All images were reconstructed with 1-mm slice thickness, B50 kernel. 10 of the cases had at least one nodule reported on the NLST reader forms. Based on a previously-published technique, we added noise to the raw data to simulate reduced-dose versions of each case at 50% and 25% of the original NLST dose (i.e. approximately 1.0 and 0.5 mGy CTDIvol). For each case at each dose level, the CAD detection algorithm was run and nodules greater than 4 mm in diameter were reported. These CAD results were compared to "truth", defined as the approximate nodule centroids from the NLST reports. Subject-level mean sensitivities and false-positive rates were calculated for each dose level. RESULTS: The mean sensitivities of the CAD algorithm were 35% at the original dose, 20% at 50% dose, and 42.5% at 25% dose. The false-positive rates, in decreasing-dose order, were 3.7, 2.9, and 10 per case. In certain cases, particularly in larger patients, there were severe photon-starvation artifacts, especially in the apical region due to the high-attenuating shoulders. CONCLUSION: The detection task was challenging for the CAD algorithm at all dose levels, including the original NLST dose. However, the false-positive rate at 25% dose approximately tripled, suggesting a loss of CAD robustness somewhere between 0.5 and 1.0 mGy. NCI grant U01 CA181156 (Quantitative Imaging Network); Tobacco Related Disease Research Project grant 22RT-0131.

The unique characteristic of head and neck squamous cell carcinoma (HNSCC) is that local invasion rather than distant metastasis is the major route for dissemination. Therefore, targeting the locally invasive cancer cells is more important than preventing systemic metastasis in HNSCC and other invasive-predominant cancers. We previously demonstrate a specific mechanism for HNSCC local invasion: the epithelial-mesenchymal transition (EMT) regulator Twist1 represses microRNA let-7i expression, leading to the activation of the small GTPase Rac1 and engendering the mesenchymal-mode movement in three-dimensional (3D) culture. However, targeting the EMT regulator is relatively difficult because of its transcription factor nature and the strategy for confining HNSCC invasion to facilitate local treatment is limited. Imipramine blue (IB) is a newly identified anti-invasive compound that effectively inhibits glioma invasion. Here we demonstrate that in HNSCC cells, a noncytotoxic dose of IB represses mesenchymal-mode migration in two-and-a-half-dimensional/3D culture system. IB suppresses EMT and stemness of HNSCC cells through inhibition of Twist1-mediated let-7i downregulation and Rac1 activation and the EMT signalling. Mechanistically, IB inhibits reactive oxygen species-induced nuclear factor-κB pathway activation. Importantly, IB promotes degradation of the EMT inducer Twist1 by enhancing F-box and leucine-rich repeat protein 14 (FBXL14)-mediated polyubiquitination of Twist1. Together, this study demonstrates the potent anti-invasion and EMT-inhibition effect of IB, suggesting the potential of IB in treating local invasion-predominant cancers.

Purpose

We examined the relationship between glycemic control and urinary tract infections in women with type 1 diabetes mellitus.

Materials and methods

Women enrolled in the Epidemiology of Diabetes Interventions and Complications study, the observational followup of the Diabetes Control and Complications Trial, were surveyed to assess the rate of physician diagnosed urinary tract infections in the preceding 12 months. The relationship between glycated hemoglobin levels and number of urinary tract infections in the previous 12 months was assessed using a multivariable Poisson regression model.

Results

A total of 572 women were evaluated at year 17. Mean age was 50.7 ± 7.2 years, mean body mass index was 28.6 ± 5.9 kg/m(2), mean type 1 diabetes duration was 29.8 ± 5.0 years and mean glycated hemoglobin was 8.0% ± 0.9%. Of these women 86 (15.0%) reported at least 1 physician diagnosed urinary tract infection during the last 12 months. Higher glycated hemoglobin levels were significantly associated with number of urinary tract infections such that for every unit increase (1%) in recent glycated hemoglobin level, there was a 21% (p=0.02) increase in urinary tract infection frequency in the previous 12 months after adjusting for race, hysterectomy status, urinary incontinence, sexual activity in the last 12 months, peripheral and autonomic neuropathy, and nephropathy.

Conclusions

The frequency of urinary tract infections increases with poor glycemic control in women with type 1 diabetes. This relationship is independent of other well described predictors of urinary tract infections and suggests that factors directly related to glycemic control may influence the risk of lower urinary tract infections.

We examined whether persistence of epigenetic DNA methylation (DNA-me) alterations at specific loci over two different time points in people with diabetes are associated with metabolic memory, the prolonged beneficial effects of intensive vs. conventional therapy during the Diabetes Control and Complications Trial (DCCT) on the progression of microvascular outcomes in the long-term follow-up Epidemiology of Diabetes Interventions and Complications (EDIC) Study. We compared DNA-me profiles in genomic DNA of whole blood (WB) isolated at EDIC Study baseline from 32 cases (DCCT conventional therapy group subjects showing retinopathy or albuminuria progression by EDIC Study year 10) vs. 31 controls (DCCT intensive therapy group subjects without complication progression by EDIC year 10). DNA-me was also profiled in blood monocytes (Monos) of the same patients obtained during EDIC Study years 16-17. In WB, 153 loci depicted hypomethylation, and 225 depicted hypermethylation, whereas in Monos, 155 hypomethylated loci and 247 hypermethylated loci were found (fold change ≥1.3; P < 0.005; cases vs. controls). Twelve annotated differentially methylated loci were common in both WB and Monos, including thioredoxin-interacting protein (TXNIP), known to be associated with hyperglycemia and related complications. A set of differentially methylated loci depicted similar trends of associations with prior HbA1c in both WB and Monos. In vitro, high glucose induced similar persistent hypomethylation at TXNIP in cultured THP1 Monos. These results show that DNA-me differences during the DCCT persist at certain loci associated with glycemia for several years during the EDIC Study and support an epigenetic explanation for metabolic memory.