Review
Urology & Nephrology
Giorgio Cazzaniga, Mattia Rossi, Albino Eccher, Ilaria Girolami, Vincenzo L'Imperio, Hien Van Nguyen, Jan Ulrich Becker, Maria Gloria Bueno Garcia, Marta Sbaraglia, Angelo Paolo Dei Tos, Giovanni Gambaro, Fabio Pagni
Summary: The integration of AI in nephropathology is a rapidly growing field, but it faces challenges such as the use of various histological techniques, low occurrence of certain diseases, and the need for data sharing. Most of the current research focuses on relatively easy tasks, but there is a trend towards more complex tasks. Deep learning has shown promise in identifying patterns in histopathology data and can be used for comprehensive assessment of renal biopsy. Collaboration among experts from different disciplines is crucial for the development of effective AI tools in nephropathology.
JOURNAL OF NEPHROLOGY
(2023)
Article
Computer Science, Artificial Intelligence
Kerstin N. Vokinger, Urs Gasser
Summary: Regulatory frameworks for artificial intelligence are being developed on both sides of the Atlantic, eagerly anticipated by the scientific and industrial community. Commonalities and differences in approaches to AI in medicine are beginning to emerge.
NATURE MACHINE INTELLIGENCE
(2021)
Article
Computer Science, Artificial Intelligence
Carlo Combi, Beatrice Amico, Riccardo Bellazzi, Andreas Holzinger, Jason H. Moore, Marinka Zitnik, John H. Holmes
Summary: This paper focuses on the importance of explainable artificial intelligence (XAI) in the field of biomedicine. By bringing together researchers with different roles and perspectives, it explores XAI in depth and presents a series of requirements for achieving explainability in AI.
ARTIFICIAL INTELLIGENCE IN MEDICINE
(2022)
Article
Environmental Sciences
Francesca Larosa, Sergio Hoyas, Javier Garcia-Martinez, J. Alberto Conejero, Francesco Fuso Nerini, Ricardo Vinuesa
Summary: Large language models provide an opportunity to advance climate and sustainability research. We believe that regulating and validating generative artificial intelligence models would benefit society more than stopping development.
NATURE CLIMATE CHANGE
(2023)
Article
Multidisciplinary Sciences
Laetitia Coassolo, Tianyun Liu, Yunshi Jung, Nikki P. Taylor, Meng Zhao, Gregory W. Charville, Silas Boye Nissen, Hannele Yki-Jarvinen, Russ B. Altman, Katrin J. Svensson
Summary: Non-alcoholic fatty liver disease (NAFLD) is a complex disease with unclear molecular mechanisms. By using single-cell RNA sequencing, researchers identified distinct clusters of hepatocytes with different expression of the lipid synthesis driver Srebp1. Interestingly, Srebp1 was not a reliable predictor of hepatic lipid accumulation, suggesting the involvement of other factors in lipid metabolism. Computational network analyses revealed a strong association between NAFLD and high constitutive androstane receptor (CAR) expression, which interacted with multiple functional modules related to lipid metabolism. These findings provide insights into the cellular differences in lipid signatures and identify important functional networks involved in hepatic steatosis in both mice and humans.
Editorial Material
Health Care Sciences & Services
Mirja Mittermaier, Marium M. Raza, Joseph C. Kvedar
Summary: Artificial intelligence is increasingly used in healthcare, particularly in surgery. While it holds promise in predicting outcomes and guiding surgeons, AI systems can also be biased, exacerbating existing inequalities. This impacts disadvantaged populations, who may receive less accurate algorithmic predictions or underestimate their need for care. Detecting and mitigating bias is crucial for creating fair and generalizable AI technology. This article discusses a recent study that developed a new strategy to address bias in surgical AI systems.
NPJ DIGITAL MEDICINE
(2023)
Review
Urology & Nephrology
Ingeborg M. Bajema
Summary: Studies have shown that artificial intelligence can effectively identify kidney structures and perform classification tasks in nephropathology, comparable to experienced pathologists. Further development of digitalized pathology, combined with the opportunities of AI and machine learning, will significantly transform the work of clinical pathologists.
NEPHROLOGIE & THERAPEUTIQUE
(2021)
Editorial Material
Biochemistry & Molecular Biology
Eric Wu, Kevin Wu, Roxana Daneshjou, David Ouyang, Daniel E. Ho, James Zou
Summary: A comprehensive overview of medical AI devices approved by the US Food and Drug Administration sheds light on limitations of the evaluation process that may mask vulnerabilities of devices when deployed on patients.
Editorial Material
Biochemistry & Molecular Biology
Zachi I. Attia, Paul A. Friedman
Summary: By applying artificial intelligence to electrocardiograms recorded by patients using Apple watches, we conducted a prospective, digital, remote study to enable large-scale screening for left ventricular dysfunction, a serious and under-detected cardiac disease. The study found that patients engaged with the system and that the watch electrocardiograms effectively screened for the disease.
Article
Computer Science, Artificial Intelligence
Luc Rocher, Arnaud J. Tournier, Yves-Alexandre de Montjoye
Summary: Algorithms are crucial in digital marketplaces, but their deployment has raised concerns about anti-competitive behavior. Previous research focused on cases where firms use the same algorithm, neglecting the possibility of adversarial collusion. In this study, we propose a network-based framework to model pricing algorithms and discover that firms can manipulate competitors' algorithms for increased profits. This calls for policymakers and regulatory agencies to consider adversarial manipulations in algorithmic pricing.
NATURE MACHINE INTELLIGENCE
(2023)
Article
Optics
Bowei Dong, Samarth Aggarwal, Wen Zhou, Utku Emre Ali, Nikolaos Farmakidis, June Sang Lee, Yuhan He, Xuan Li, Dim-Lee Kwong, C. D. Wright, Wolfram H. P. Pernice, H. Bhaskaran
Summary: This paper introduces new developments in hardware-based accelerators, including electronic tensor cores and photonic implementations. By modulating photonic signals and combining them with distributed memories and wavelength multiplexing, we configure the system to be compatible with edge computing frameworks. Through processing electrocardiogram data and constructing a convolutional neural network, we demonstrate that this method can accurately identify patients at risk of sudden cardiac death.
Article
Multidisciplinary Sciences
Lingling Fan, Kai Wang, Heming Wang, Avik Dutt, Shanhui Fan
Summary: Photonic convolution, a crucial operation in signal and image processing, can overcome computational bottlenecks and outperform electronic implementations. This study demonstrates the realization of convolution operations in the synthetic frequency dimension using a modulated ring resonator. By synthesizing arbitrary convolution kernels with high accuracy, we showcase the computation between input frequency combs and synthesized kernels. Our work highlights the efficient data encoding and computation capabilities of the synthetic frequency dimension, paving the way for compact and scalable photonic computation architecture.
Article
Computer Science, Artificial Intelligence
Fernando Martinez-Plumed, Pablo Barredo, Sean O. Heigeartaigh, Jose Hernandez-Orallo
Summary: Experimental benchmarks like ImageNet and Atari games are crucial for advancing AI research. An analysis of results and papers linked to 25 popular benchmarks reveals that competition and collaboration dynamics in AI research are still not well understood. The study provides an innovative methodology to explore the behavior of different entrants in challenges, from academia to tech giants, in response to achievements.
NATURE MACHINE INTELLIGENCE
(2021)
Article
Meteorology & Atmospheric Sciences
Massimo Bonavita, Rochelle Schneider, Rossella Arcucci, Matthew Chantry, Marcin Chrust, Alan Geer, Bertrand Le Saux, Claudia Vitolo
Summary: This report summarizes the main outcomes of the 3(rd) edition of the workshop on Machine Learning (ML) for Earth System Observation and Prediction (ESOP/ML4ESOP) co-organized by ECMWF and ESA. The workshop, which took place in hybrid format, attracted a record number of submissions and over 700 registrations. It aimed to document the current state-of-the-art and challenges in integrating ML technologies in ESOP.
NPJ CLIMATE AND ATMOSPHERIC SCIENCE
(2023)
Article
Multidisciplinary Sciences
Yoshiko Bamba, Shimpei Ogawa, Michio Itabashi, Shingo Kameoka, Takahiro Okamoto, Masakazu Yamamoto
Summary: This study utilized convolutional neural networks to recognize and evaluate the accuracy of forceps types in surgical videos obtained during colorectal surgeries, demonstrating the potential for achieving high accuracy in forceps recognition.
SCIENTIFIC REPORTS
(2021)