A5019245881- Advanced Vision and Imaging
- Genetic Associations and Epidemiology
- Suicide and Self-Harm Studies
- Mental Health Treatment and Access
- Substance Abuse Treatment and Outcomes
- Optical measurement and interference techniques
- Alcohol Consumption and Health Effects
- Machine Learning in Healthcare
- Mental Health Research Topics
- Image Enhancement Techniques
- Robotics and Sensor-Based Localization
- Nutrition, Genetics, and Disease
- Advanced Image and Video Retrieval Techniques
- Opioid Use Disorder Treatment
- Face recognition and analysis
- Health disparities and outcomes
- HIV, Drug Use, Sexual Risk
- Cardiac Health and Mental Health
- Generative Adversarial Networks and Image Synthesis
- Bipolar Disorder and Treatment
- Healthcare and Venom Research
- Chronic Disease Management Strategies
- Health Systems, Economic Evaluations, Quality of Life
- Anomaly Detection Techniques and Applications
- Image and Signal Denoising Methods
Massachusetts General Hospital
2021-2025
Harvard University
2023-2025
Vanderbilt University Medical Center
2024-2025
Liechtenstein Institute
2023
Hudson Institute
2023
John Wiley & Sons (United States)
2023
Broad Institute
2022
Kao Corporation (Japan)
2021
Brain (Germany)
2021
Brown University
2020
This paper presents a novel structure-preserving image decomposition operator called bilateral texture filter . As simple modification of the original [Tomasi and Manduchi 1998], it performs local patch-based analysis features incorporates its results into range kernel. The central idea to ensure proper texture/structure separation is based on patch shift that captures information from most representative clear prominent structure edges. Our method outperforms in removing while preserving...
Abstract Problematic alcohol use (PAU), a trait that combines disorder and alcohol-related problems assessed with questionnaire, is leading cause of death morbidity worldwide. Here we conducted large cross-ancestry meta-analysis PAU in 1,079,947 individuals (European, N = 903,147; African, 122,571; Latin American, 38,962; East Asian, 13,551; South 1,716 ancestries). We observed high degree cross-ancestral similarity the genetic architecture identified 110 independent risk variants within-...
Abstract This paper presents a novel method to enhance the performance of structure‐preserving image and texture filtering. With conventional edge‐aware filters, it is often challenging handle images high complexity where features multiple scales coexist. In particular, not always easy find right balance between removing unimportant details protecting important when they come in sizes, shapes, contrasts. Unlike previous approaches, we address this issue from perspective adaptive kernel...
Abstract This paper presents a novel filtering‐based method for decomposing an image into structures and textures. Unlike previous filtering algorithms, our adaptively smooths gradients to filter out textures from images. A new gradient operator, the interval gradient, is proposed adaptive smoothing. Using gradients, can be distinguished structure edges smoothly varying shadings. We also propose effective gradient‐guided algorithm produce high‐quality results filtered gradients. Our avoids...
Abstract Bipolar disorder is a leading contributor to disability, premature mortality, and suicide. Early identification of risk for bipolar using generalizable predictive models trained on diverse cohorts around the United States could improve targeted assessment high individuals, reduce misdiagnosis, allocation limited mental health resources. This observational case-control study intended develop validate as part multisite, multinational PsycheMERGE Network across large biobanks with...
Growing evidence has shown that applying machine learning models to large clinical data sources may exceed clinician performance in suicide risk stratification. However, many existing prediction either suffer from "temporal bias" (a bias stems using case-control sampling) or require training on all available patient visit data. Here, we adopt a "landmark model" framework aligns with practice for of suicide-related behaviors (SRBs) electronic health record database. Using the landmark...
Man-made structures often appear to be distorted in photos captured by casual photographers, as the scene layout conflicts with how it is expected human perception. In this paper, we propose an automatic approach for straightening up slanted man-made input image improve its perceptual quality. We call type of correction upright adjustment. a set criteria adjustment based on perception studies, and develop optimization framework which yields optimal homography also new optimization-based...
Current clinician-based and automated risk assessment methods treat the of suicide-related behaviors (SRBs) as static, while in actual clinical practice, SRB fluctuates over time. Here, we develop two closely related model classes, Event-GRU-ODE Event-GRU-Discretized, that can predict dynamic events a continuous trajectory across future time points, even without new observations, updating these estimates data become available. Models were trained validated for prediction using large...
Schizophrenia patients have substantial comorbidity contributing to reduced life expectancy of 10-20 years. Identifying modifiable comorbidities could improve rates premature mortality. Conditions that frequently co-occur but lack shared genetic risk with schizophrenia are more likely be products treatment, behavior, or environmental factors and therefore enriched for potentially associations. Phenome-wide was calculated from electronic health records (EHR) 250,000 across two independent...
Background Early identification of bipolar disorder (BD) provides an important opportunity for timely intervention. In this study, we aimed to develop machine learning models using large‐scale electronic health record (EHR) data including clinical notes predicting early‐onset BD. Methods Structured and unstructured were extracted from the longitudinal EHR Mass General Brigham system. We defined three cohorts aged 10–25 years: (1) full youth cohort ( N = 300,398); (2) a subcohort by having...
Importance Clinical practice guidelines recommend suicide risk screening and assessment across behavioral health settings. The predictive accuracy of real-world clinician assessments for stratifying patients by future suicidal behavior, however, remains understudied. Objective To evaluate routine clinical prospectively predicting attempt. Design, Setting, Participants This electronic record–based, prognostic study included 89 957 (≥5 years age) with a structured (based on the Suicide...
Single image camera calibration is the task of estimating parameters from a single input image, such as vanishing points, focal length, and horizon line. In this work, we propose Camera TRansformer with Line-Classification (CTRL-C), an end-to-end neural network-based approach to calibration, which directly estimates set line segments. Our network adopts transformer architecture capture global structure multi-modal inputs in manner. We also auxiliary classification train extract geometric...
Interest in developing machine learning models that use electronic health record data to predict patients' risk of suicidal behavior has recently proliferated. However, whether and how such might be implemented useful clinical practice remain unknown. To ultimately make automated suicide risk-prediction practice, thus better prevent patient suicides, it is critical partner with key stakeholders, including the frontline providers who will using tools, at each stage implementation process.The...
Problematic alcohol use (PAU) is a leading cause of death and disability worldwide. To improve our understanding the genetics PAU, we conducted large cross-ancestry meta-analysis PAU in 1,079,947 individuals. We observed high degree cross-ancestral similarity genetic architecture identified 110 independent risk variants within- analyses. Cross-ancestry fine-mapping improved identification likely causal variants. Prioritizing genes through gene expression and/or chromatin interaction brain...
Multiple areas in the United States of America (USA) are experiencing high rates overdose and outbreaks bloodborne infections, including HIV hepatitis C virus (HCV), due to non-sterile injection drug use. We aimed identify neighbourhoods at increased vulnerability for infectious disease Rhode Island, USA. The primary aim was pilot machine learning methods which neighbourhood-level factors were important creating "vulnerability assessment scores" across state. secondary engage stakeholders an...
How did mental healthcare utilization change during the COVID-19 pandemic period among individuals with pre-existing disorder? Understanding patterns of these at-risk and identifying those most likely to exhibit increased could improve patient stratification efficient delivery health services. This study leveraged large-scale electronic record (EHR) data describe disorder before identify correlates high utilization. Using EHR from a large system in Massachusetts, we identified three...
Man-made structures often appear to be distorted in photos captured by casual photographers, as the scene layout conflicts with how it is expected human perception. In this paper we propose an automatic approach for straightening up slanted man-made input image improve its perceptual quality. We call type of correction upright adjustment. a set criteria adjustment based on perception studies, and develop optimization framework which yields optimal homography also new optimization-based...
Opioid use disorders (OUDs) constitute a major public health issue, and we urgently need alternative methods for characterizing risk OUD. Electronic records (EHRs) are useful tools understanding complex medical phenotypes but have been underutilized OUD because of challenges related to underdiagnosis, binary diagnostic frameworks, minimally characterized reference groups. As first step in addressing these challenges, new paradigm is warranted that characterizes opioid prescription misuse on...
Tobacco use disorder (
Abstract Objective Early identification of bipolar disorder (BD) provides an important opportunity for timely intervention. In this study, we aimed to develop machine learning models using large-scale electronic health record (EHR) data including clinical notes predicting early-onset BD. Method Structured and unstructured were extracted from the longitudinal EHR Mass General Brigham system. We defined three cohorts aged 10 – 25 years: (1) full youth cohort (N=300,398); (2) a sub-cohort by...
Throughout the COVID-19 pandemic, graduate students have faced increased risk of mental health challenges. Research suggests that experiencing adversity may induce positive psychological changes, called post-traumatic growth (PTG). These changes can include improved relationships with others, perceptions oneself, and enjoyment life. Few existing studies explored this phenomenon among students. This secondary data analysis a survey conducted in November 2020 at private R1 University northeast...
Prior suicide attempts are a relatively strong risk factor for future attempts. There is growing interest in using longitudinal electronic health record (EHR) data to derive statistical prediction models and other suicidal behavior outcomes. However, model performance may be inflated by largely unrecognized form of "data leakage" during training: diagnostic codes attempt outcomes refer prior that also included the as predictors.