SCALAR ON NETWORK REGRESSION VIA BOOSTING.

Neuroimaging studies have a growing interest in learning the association between the individual brain connectivity networks and their clinical characteristics. It is also of great interest to identify the sub brain networks as biomarkers to predict the clinical symptoms, such as disease status, potentially providing insight on neuropathology. This motivates the need for developing a new type of regression model where the response variable is scalar, and predictors are networks that are typically represented as adjacent matrices or weighted adjacent matrices, to which we refer as scalar-on-network regression. In this work, we develop a new boosting method for model fitting with sub-network markers selection. Our approach, as opposed to group lasso or other existing regularization methods, is essentially a gradient descent algorithm leveraging known network structure. We demonstrate the utility of our methods via simulation studies and analysis of the resting-state fMRI data in a cognitive developmental cohort study.

On predictability of individual functional connectivity networks from clinical characteristics.

In recent years, understanding functional brain connectivity has become increasingly important as a scientific tool with potential clinical implications. Statistical methods, such as graphical models and network analysis, have been adopted to construct functional connectivity networks for single subjects. Here we focus on studying the association between functional connectivity networks and clinical characteristics such as psychiatric symptoms and diagnoses. Utilizing machine learning algorithms, we propose a method to examine predictability of functional connectivity networks from clinical characteristics. Our methods can identify salient clinical characteristics predictive of the whole brain network or specific subnetworks. We illustrate our methods on the analysis of fMRI data in the Philadelphia Neurodevelopmental Cohort study, demonstrating clinically meaningful results.

Cardiac Magnetic Resonance Imaging and Blood Biomarkers for Evaluation of Radiation-Induced Cardiotoxicity in Patients With Breast Cancer: Results of a Phase 2 Clinical Trial.

PURPOSE: Radiation therapy (RT) can increase the risk of cardiac events in patients with breast cancer (BC), but biomarkers predicting risk for developing RT-induced cardiac disease are currently lacking. We report results from a prospective clinical trial evaluating early magnetic resonance imaging (MRI) and serum biomarker changes as predictors of cardiac injury and risk of subsequent cardiac events after RT for left-sided disease. METHODS: Women with node-negative and node-positive (N-/+) left-sided BC were enrolled on 2 institutional review board (IRB)-approved protocols at 2 institutions. MRI was conducted pretreatment (within 1 week of starting radiation), at the end of treatment (last day of treatment ±1 week), and 3 months after the last day of treatment (±2 weeks) to quantify left and right ventricular volumes and function, myocardial fibrosis, and edema. Perfusion changes during regadenoson stress perfusion were also assessed on a subset of patients (n = 28). Serum was collected at the same time points. Whole heart and cardiac substructures were contoured using CT and MRI. Models were constructed using baseline cardiac and clinical risk factors. Associations between MRI-measured changes and dose were evaluated. RESULTS: Among 51 women enrolled, mean heart dose ranged from 0.80 to 4.7 Gy and mean left ventricular (LV) dose from 1.1 to 8.2 Gy, with mean heart dose 2.0 Gy. T1 time, a marker of fibrosis, and right ventricular (RV) ejection fraction (EF) significantly changed with treatment; these were not dose dependent. T2 (marker of edema) and LV EF did not significantly change. No risk factors were associated with baseline global perfusion. Prior receipt of doxorubicin was marginally associated with decreased myocardial perfusion after RT (P = .059), and mean MHD was not associated with perfusion changes. A significant correlation between baseline IL-6 and mean heart dose (MHD) at the end of RT (ρ 0.44, P = .007) and a strong trend between troponin I and MHD at 3 months post-treatment (ρ 0.33, P = .07) were observed. No other significant correlations were identified. CONCLUSIONS: In this prospective study of women with left-sided breast cancer treated with contemporary treatment planning, cardiac radiation doses were very low relative to historical doses reported by Darby et al. Although we observed significant changes in T1 and RV EF shortly after RT, these changes were not correlated with whole heart or substructure doses. Serum biomarker analysis of cardiac injury demonstrates an interesting trend between markers and MHD that warrants further investigation.

Comprehensive evaluation of COVID-19 patient short- and long-term outcomes: Disparities in healthcare utilization and post-hospitalization outcomes.

BACKGROUND: Understanding risk factors for short- and long-term COVID-19 outcomes have implications for current guidelines and practice. We study whether early identified risk factors for COVID-19 persist one year later and through varying disease progression trajectories. METHODS: This was a retrospective study of 6,731 COVID-19 patients presenting to Michigan Medicine between March 10, 2020 and March 10, 2021. We describe disease progression trajectories from diagnosis to potential hospital admission, discharge, readmission, or death. Outcomes pertained to all patients: rate of medical encounters, hospitalization-free survival, and overall survival, and hospitalized patients: discharge versus in-hospital death and readmission. Risk factors included patient age, sex, race, body mass index, and 29 comorbidity conditions. RESULTS: Younger, non-Black patients utilized healthcare resources at higher rates, while older, male, and Black patients had higher rates of hospitalization and mortality. Diabetes with complications, coagulopathy, fluid and electrolyte disorders, and blood loss anemia were risk factors for these outcomes. Diabetes with complications, coagulopathy, fluid and electrolyte disorders, and blood loss were associated with lower discharge and higher inpatient mortality rates. CONCLUSIONS: This study found differences in healthcare utilization and adverse COVID-19 outcomes, as well as differing risk factors for short- and long-term outcomes throughout disease progression. These findings may inform providers in emergency departments or critical care settings of treatment priorities, empower healthcare stakeholders with effective disease management strategies, and aid health policy makers in optimizing allocations of medical resources.

Stream Nitrogen, Phosphorus, and Sediment Concentrations Show Contrasting Long-term Trends Associated with Agricultural Change.

Stream water quality can be greatly influenced by changes in agricultural practices, but studies of long-term dynamics are scarce. Here we describe trends over 21 yr (1994-2014) in nutrients and suspended sediments in three streams in a Midwestern US agricultural watershed. During this time, the watershed experienced substantial changes in agricultural practices, most importantly a pronounced shift from conventional to conservation tillage. In the 1990s and early 2000s, NH, soluble reactive P, and suspended sediment concentrations (standardized for discharge and season) each declined significantly (>4-12% per year) in at least two of the three streams ( < 0.01), whereas NO changed relatively little. However, since the early 2000s, declines in NH and sediment concentrations have slowed, soluble reactive P concentrations have not declined and may actually have increased, and NO concentrations have declined sharply. The more recent lack of decline in soluble reactive P coincides with a plateau in the prevalence of conservation tillage and may be because of increased soil P stratification due to long-term reduced tillage. The more recent decline in NO may be due to improved efficiency of N fertilizer use, increased soil denitrification, and/or declines in atmospheric N deposition. Our study shows that stream concentrations of N, P, and sediment can respond in contrasting ways to changes in agriculture, and that temporal trends can moderate, accelerate, or reverse over decadal timescales. Management strategies must consider contrasting temporal responses of water quality indicators and may need to be adaptively adjusted at scales of years to decades.