Article
Radiology, Nuclear Medicine & Medical Imaging
Heejong Kim, Daniel J. A. Margolis, Himanshu Nagar, Mert R. Sabuncu
Summary: By utilizing multiparametric magnetic resonance imaging (mpMRI) combined with deep learning models, early detection and localization of prostate cancer can be achieved. The mpMRI model with T2-ADC-DWI sequence achieved a high AUC score of 0.90 in the test set, slightly outperforming the model using Ktrans instead of DWI. The study demonstrates that convolutional neural networks incorporating multiple pulse sequences show high performance for detecting clinically significant prostate cancer.
ACADEMIC RADIOLOGY
(2023)
Article
Environmental Sciences
Glenn R. Moncrieff
Summary: Existing efforts in monitoring land cover change have focused mostly on forested ecosystems, but this study demonstrates the potential of using neural networks to accurately detect transformation in a critically endangered shrubland ecosystem within a few days of its occurrence. The trained models achieved a high accuracy in detecting land cover change events, surpassing the performance of methods designed for forested ecosystems. Continuous monitoring of habitat loss in non-forest ecosystems can be facilitated with accurately dated datasets and machine learning classifiers.
Article
Computer Science, Information Systems
Miles Q. Li, Benjamin C. M. Fung, Philippe Charland, Steven H. H. Ding
Summary: Malware poses serious threats to computer users, and current signature-based detection methods have limitations in detecting new malware. Machine learning methods like I-MAD show promising results in accuracy and interpretability.
COMPUTERS & SECURITY
(2021)
Article
Engineering, Biomedical
Xuyang Zhao, Noboru Yoshida, Tetsuya Ueda, Hidenori Sugano, Toshihisa Tanaka
Summary: This study applies commonly used models such as LeNet, VGG, ResNet, and ViT to the EEG image classification task, and solves the problems of data imbalance and model interpretation through data augmentation and model explanation methods. The models achieve good performance in seizure detection and provide visual and quantitative information for clinical experts in diagnosis.
JOURNAL OF NEURAL ENGINEERING
(2023)
Article
Biochemical Research Methods
Elham Khalili, Shahin Ramazi, Faezeh Ghanati, Samaneh Kouchaki
Summary: Phosphorylation of proteins is a significant post-translational modification that plays a crucial role in plant functionality. Accurate prediction of plant phosphorylation sites is vital, and this study develops machine learning-based techniques to improve the prediction of protein phosphorylation sites in soybean. The proposed technique achieves high accuracy and specificity, and can be used to automatically analyze data and predict potential protein phosphorylation sites in plants.
BRIEFINGS IN BIOINFORMATICS
(2022)
Article
Engineering, Biomedical
Morgan Stuart, Srdjan Lesaja, Jerry J. Shih, Tanja Schultz, Milos Manic, Dean J. Krusienski
Summary: This article presents a deep learning architecture that learns input bandpass filters capturing task-relevant spectral features directly from data, furthering the goal of end-to-end architectures and achieving good performance in speech tasks.
IEEE TRANSACTIONS ON NEURAL SYSTEMS AND REHABILITATION ENGINEERING
(2022)
Article
Environmental Sciences
Hang Zheng, Yueyi Liu, Wenhua Wan, Jianshi Zhao, Guanti Xie
Summary: Deep learning methods are increasingly used in water quality prediction due to their ability to map nonlinear relationships quickly. However, the lack of physical explanation limits their practicality. In this study, an interpretable deep learning framework was established to predict water quality variations. The model achieved satisfactory prediction performance, with coefficients of determination above 0.80 for COD, TP, and TN. The SHapley Additive exPlanations method was effective in interpreting the results and identifying important variables affecting water quality variations.
JOURNAL OF ENVIRONMENTAL MANAGEMENT
(2023)
Article
Environmental Sciences
Qi Liu, Dongwei Gui, Lei Zhang, Jie Niu, Heng Dai, Guanghui Wei, Bill X. Hu
Summary: This study used machine learning and deep learning approaches to forecast groundwater levels in arid regions, successfully predicting the spatiotemporal variations in the lower Tarim River basin. The findings identified the critical impact of flow volume, distance to the river channel, and reservoir on groundwater changes. These results have significant implications for sustainable water resources management in arid regions.
SCIENCE OF THE TOTAL ENVIRONMENT
(2022)
Article
Geosciences, Multidisciplinary
Zijing Luo, Renguang Zuo, Yihui Xiong
Summary: Deep learning algorithms have outperformed traditional methods in multivariate geochemical anomaly recognition due to their ability to extract features from nonlinear data. However, most DLAs are black-box approaches due to the high nonlinearity of the hidden layer. This study improves the interpretability of the model by visualizing features and integrating geological domain knowledge into the loss function. The addition of penalty terms and the integration of spatiotemporal and genetic relationships enhance the geological interpretability of the network. A case study demonstrates strong spatial correlation between the results of the geologically constrained AAE and the geological features in the study area.
NATURAL RESOURCES RESEARCH
(2022)
Article
Construction & Building Technology
Yuan Gao, Shohei Miyata, Yuki Matsunami, Yasunori Akashi
Summary: This study investigates the use of a transformer model to improve solar radiation prediction, achieving better accuracy by considering factors such as humidity and temperature. Detailed case studies and one-step analysis results confirm the importance of these factors in the transformer model's performance.
ENERGY AND BUILDINGS
(2023)
Article
Genetics & Heredity
V. V. Kuznetsov, V. A. Moskalenko, D. V. Gribanov, Nikolai Yu. Zolotykh
Summary: The method proposed in this study uses a variational autoencoder to generate an ECG signal for one cardiac cycle, extracting a vector of 25 new features. The generated ECG has a natural appearance and high quality, with the new features helping improve automatic diagnostics of cardiovascular diseases and addressing the lack of labeled ECG for supervised learning.
FRONTIERS IN GENETICS
(2021)
Article
Computer Science, Interdisciplinary Applications
Xiao Luo, Priyanka Gandhi, Zuoyi Zhang, Wei Shao, Zhi Han, Vasu Chandrasekaran, Vladimir Turzhitsky, Vishal Bali, Anna R. Roberts, Megan Metzger, Jarod Baker, Carmen La Rosa, Jessica Weaver, Paul Dexter, Kun Huang
Summary: The study effectively predicted chronic cough patients using deep learning algorithms with structured and unstructured EHR data, achieving high sensitivity and specificity in patient identification.
COMPUTER METHODS AND PROGRAMS IN BIOMEDICINE
(2021)
Article
Radiology, Nuclear Medicine & Medical Imaging
Ho Young Park, Woo Hyun Shim, Chong Hyun Suh, Hwon Heo, Hyun Woo Oh, Jinyoung Kim, Jinkyeong Sung, Jae-Sung Lim, Jae-Hong Lee, Ho Sung Kim, Sang Joon Kim
Summary: This study developed an automatic classification algorithm for diagnosing Alzheimer's disease (AD) or mild cognitive impairment (MCI) and compared its performance with the widely used classifier XGBoost. The results showed that TabNet achieved high performance in AD classification and provided detailed interpretation of the selected regions.
EUROPEAN RADIOLOGY
(2023)
Article
Radiology, Nuclear Medicine & Medical Imaging
Yohan Jun, Yae Won Park, Hyungseob Shin, Yejee Shin, Jeong Ryong Lee, Kyunghwa Han, Sung Soo Ahn, Soo Mee Lim, Dosik Hwang, Seung-Koo Lee
Summary: The study aimed to establish an interpretable multiparametric deep learning model for automatic noninvasive grading and segmentation of meningiomas. The model was trained and validated using brain MRI data from 257 patients. The results showed that the model combining T1C and T2 achieved the highest performance in segmentation and grading, accurately identifying the tumor margin.
EUROPEAN RADIOLOGY
(2023)
Article
Automation & Control Systems
Blake VanBerlo, Matthew A. S. Ross, Jonathan Rivard, Ryan Booker
Summary: This study introduces a machine learning approach to predict chronic homelessness using de-identified client shelter records from a Canadian homelessness management information system. The training method was fine-tuned to achieve a high level of performance, balancing recall and precision.
ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE
(2021)
Article
Surgery
Tiffany N. Anderson, Eric M. Williams, Tyler J. Loftus, Crystal N. Johnson-Mann, Jessica E. Taylor
Summary: Trauma patients with obesity experience disparities in outcomes, but a verified bariatric surgery center of excellence may improve care for these patients.
Article
Urology & Nephrology
Benjamin Shickel, Tyler J. Loftus, Yuanfang Ren, Parisa Rashidi, Azra Bihorac, Tezcan Ozrazgat-Baslanti
CLINICAL JOURNAL OF THE AMERICAN SOCIETY OF NEPHROLOGY
(2023)
Article
Surgery
Tyler J. Loftus, Matthew M. Ruppert, Benjamin Shickel, Tezcan Ozrazgat-Baslanti, Jeremy A. Balch, Die Hu, Adnan Javed, Firas Madbak, David J. Skarupa, Faheem Guirgis, Philip A. Efron, Patrick J. Tighe, William R. Hogan, Parisa Rashidi, Gilbert R. Upchurch, Azra Bihorac
Summary: A single-institution study found that overtriaging low-risk postoperative patients to ICUs is associated with low-value care, while undertriaging high-risk patients to general wards increases mortality and morbidity. This study tested the reproducibility of an automated postoperative triage classification system and its potential for generating an actionable and explainable decision support system.
JOURNAL OF THE AMERICAN COLLEGE OF SURGEONS
(2023)
Article
Biophysics
Jeremy A. Balch, Matthew M. Rupert, Benjamin Shickel, Tezcan Ozrazgat-Baslanti, Patrick J. Tighe, Philip A. Efron, Gilbert R. Upchurch, Parisa Rashidi, Azra Bihorac, Tyler J. Loftus
Summary: This study introduces the MySurgeryRisk algorithm developed by the University of Florida College of Medicine, which predicts eight major post-operative complications using automatically extracted data from electronic health records. The study highlights the development of an efficient and accurate model for data processing and predictive analytics, in parallel with the Intelligent Critical Care Center, and discusses the future directions of the model.
PHYSIOLOGICAL MEASUREMENT
(2023)
Article
Multidisciplinary Sciences
Benjamin Shickel, Tyler J. Loftus, Matthew Ruppert, Gilbert R. Upchurch, Tezcan Ozrazgat-Baslanti, Parisa Rashidi, Azra Bihorac
Summary: Accurate prediction of postoperative complications is important for shared decision-making regarding prognosis, risk-reduction, and resource allocation. This study compared the performance of deep learning models with traditional machine learning models in predicting postoperative complications using patient data. The results showed that deep learning models were more efficient and provided more precise representations of patient health. The interpretability mechanisms identified modifiable risk factors, and the uncertainty metrics enhanced clinical trust. These findings suggest the potential of deep learning for effective clinical implementation.
SCIENTIFIC REPORTS
(2023)
Article
Cardiac & Cardiovascular Systems
Mohammad A. Al-Ani, Chen Bai, Maisara Bledsoe, Mustafa M. Ahmed, Juan R. Vilaro, Alex M. Parker, Juan M. Aranda Jr, Eric Jeng, Benjamin Shickel, Azra Bihorac, Giles J. Peek, Mark S. Bleiweis, Jeffrey P. Jacobs, Mamoun T. Mardini
Summary: This study aimed to explore the impact of device selection on heart transplantation outcomes, taking into account regional practice variation. The results showed that patients using Impella devices had higher medical acuity and lower success rate of transplantation as status 2 compared to those using IABP. The utilization ratio of IABP to Impella varied widely between regions, but this difference was not justified by medical acuity or other factors.
JOURNAL OF HEART AND LUNG TRANSPLANTATION
(2023)
Article
Surgery
Amanda C. Filiberto, Molly Q. Nyren, Patrick W. Underwood, Jeremy A. Balch, Kenneth L. Abbott, Philip A. Efron, George A. Sarosi Jr, Azra Bihorac, Gilbert R. Upchurch Jr, Tyler J. Loftus
Summary: This study examined the impact of intraoperative cholangiography on resource use in laparoscopic cholecystectomy. The results showed that the use of intraoperative cholangiography was associated with a lower incidence of postoperative endoscopic retrograde cholangiography, shorter interval, shorter length of stay, and lower total costs.
Article
Surgery
Jeremy A. Balch, Tyler J. Loftus
Summary: Clinical prediction models based on artificial intelligence algorithms have the potential to improve patient care, reduce errors, and add value to the health care system. However, their adoption is hindered by legitimate concerns in terms of economy, practicality, professionalism, and intellectual property. This article explores these barriers and suggests well-studied instruments that can help overcome them. Adapting actionable predictive models requires incorporating patient, clinical, technical, and administrative perspectives, and ensuring explainability, low error frequency and severity, safety, and fairness. Continuous validation and monitoring are necessary to address healthcare setting variations and comply with evolving regulations. By following these principles, surgeons and healthcare providers can effectively leverage artificial intelligence to optimize patient care.
Review
Surgery
Jeremy A. Balch, Jonathan R. Krebs, Amanda C. Filiberto, William G. Montgomery, Lauren C. Berkow, Gilbert R. Upchurch Jr, Tyler J. Loftus
Summary: This scoping review examines waste reduction strategies in operating rooms and highlights the importance of changing surgical instruments and optimizing processes. The study also identifies barriers to implementation, such as lack of policies and funding. Few studies discuss the sustainability of waste reduction initiatives.
Review
Surgery
Evan L. Barrios, Valerie E. Polcz, Sara E. Hensley, George A. Sarosi, Alicia M. Mohr, Tyler J. Loftus, Gilbert R. Upchurch, Jill M. Sumfest, Philip A. Efron, Kim Dunleavy, Letitia Bible, Krista P. Terracina, Mazen R. Al-Mansour, Nicole Gravina
Summary: Surgical ergonomic development and awareness are crucial for the long-term health and well-being of surgeons. Work-related musculoskeletal disorders have a significant impact on surgeons, and different surgical modalities affect the musculoskeletal system differently. This study aims to synthesize ergonomic analysis by surgical modality and discuss future directions based on current perioperative interventions.
Article
Surgery
Amanda C. Filiberto, Shunshun Miao, Yuanfang Ren, Tezcan Ozrazgat-Baslanti, Sara E. Hensley, Christopher R. Jacobs, M. Libby Weaver, Gilbert R. Upchurch Jr, Azra Bihorac, Michol Cooper
Summary: This study found that bilateral atherosclerotic renal artery stenosis (RAS) is associated with increased incidence of postoperative AKI and mortality. It suggests that bilateral RAS may serve as a marker of poor outcomes and should be considered in preoperative risk stratification.
SURGERY OPEN SCIENCE
(2023)
Article
Multidisciplinary Sciences
Ruben D. Zapata, Shu Huang, Earl Morris, Chang Wang, Christopher Harle, Tanja Magoc, Mamoun Mardini, Tyler Loftus, Francois Modave
Summary: This study developed and validated predictive models using EHR data to determine if hospitalized COVID-19-positive patients would be admitted to alternative medical care or discharged home.
Review
Medical Informatics
Jeremy A. Balch, Matthew M. Ruppert, Tyler J. Loftus, Ziyuan Guan, Yuanfang Ren, Gilbert R. Upchurch, Tezcan Ozrazgat-Baslanti, Parisa Rashidi, Azra Bihorac
Summary: This study evaluates and compares the functionalities, strengths, and weaknesses of existing ML-CISs and proposes guidelines for future work optimization. The articles on FHIR-based ML-CISs were categorized into clinical decision support systems, data management and analytic platforms, or auxiliary modules and APIs. Shortcomings in current ML-CISs include lack of EHR interoperability and clinical efficacy validation.
JMIR MEDICAL INFORMATICS
(2023)
Review
Urology & Nephrology
Nabihah Amatullah, Britney A. Stottlemyer, Isabelle Zerfas, Cole Stevens, Tezcan Ozrazgat-Baslanti, Azra Bihorac, Sandra L. Kane-Gill
Summary: This study reviewed the variability in criteria for drug-associated acute kidney injury (D-AKI) and emphasized the importance of establishing minimum reporting expectations and criteria for standardized pharmacovigilance strategies.