Receive a weekly summary and discussion of the top papers of the week by leading researchers in the field.

General General

The Three Terms Task - an open benchmark to compare human and artificial semantic representations.

In Scientific data

Word processing entails retrieval of a unitary yet multidimensional semantic representation (e.g., a lemon's colour, flavour, possible use) and has been investigated in both cognitive neuroscience and artificial intelligence. To enable the direct comparison of human and artificial semantic representations, and to support the use of natural language processing (NLP) for computational modelling of human understanding, a critical challenge is the development of benchmarks of appropriate size and complexity. Here we present a dataset probing semantic knowledge with a three-terms semantic associative task: which of two target words is more closely associated with a given anchor (e.g., is lemon closer to squeezer or sour?). The dataset includes both abstract and concrete nouns for a total of 10,107 triplets. For the 2,255 triplets with varying levels of agreement among NLP word embeddings, we additionally collected behavioural similarity judgments from 1,322 human raters. We hope that this openly available, large-scale dataset will be a useful benchmark for both computational and neuroscientific investigations of semantic knowledge.

Borghesani V, Armoza J, Hebart M N, Bellec P, Brambati S M

2023-Mar-02

Pathology Pathology

Deep learning-enabled virtual histological staining of biological samples.

In Light, science & applications

Histological staining is the gold standard for tissue examination in clinical pathology and life-science research, which visualizes the tissue and cellular structures using chromatic dyes or fluorescence labels to aid the microscopic assessment of tissue. However, the current histological staining workflow requires tedious sample preparation steps, specialized laboratory infrastructure, and trained histotechnologists, making it expensive, time-consuming, and not accessible in resource-limited settings. Deep learning techniques created new opportunities to revolutionize staining methods by digitally generating histological stains using trained neural networks, providing rapid, cost-effective, and accurate alternatives to standard chemical staining methods. These techniques, broadly referred to as virtual staining, were extensively explored by multiple research groups and demonstrated to be successful in generating various types of histological stains from label-free microscopic images of unstained samples; similar approaches were also used for transforming images of an already stained tissue sample into another type of stain, performing virtual stain-to-stain transformations. In this Review, we provide a comprehensive overview of the recent research advances in deep learning-enabled virtual histological staining techniques. The basic concepts and the typical workflow of virtual staining are introduced, followed by a discussion of representative works and their technical innovations. We also share our perspectives on the future of this emerging field, aiming to inspire readers from diverse scientific fields to further expand the scope of deep learning-enabled virtual histological staining techniques and their applications.

Bai Bijie, Yang Xilin, Li Yuzhu, Zhang Yijie, Pillar Nir, Ozcan Aydogan

2023-Mar-03

General General

Machine learning methods to predict outcomes of pharmacological treatment in psychosis.

In Translational psychiatry ; h5-index 60.0

In recent years, machine learning (ML) has been a promising approach in the research of treatment outcome prediction in psychosis. In this study, we reviewed ML studies using different neuroimaging, neurophysiological, genetic, and clinical features to predict antipsychotic treatment outcomes in patients at different stages of schizophrenia. Literature available on PubMed until March 2022 was reviewed. Overall, 28 studies were included, among them 23 using a single-modality approach and 5 combining data from multiple modalities. The majority of included studies considered structural and functional neuroimaging biomarkers as predictive features used in ML models. Specifically, functional magnetic resonance imaging (fMRI) features contributed to antipsychotic treatment response prediction of psychosis with good accuracies. Additionally, several studies found that ML models based on clinical features might present adequate predictive ability. Importantly, by examining the additive effects of combining features, the predictive value might be improved by applying multimodal ML approaches. However, most of the included studies presented several limitations, such as small sample sizes and a lack of replication tests. Moreover, considerable clinical and analytical heterogeneity among included studies posed a challenge in synthesizing findings and generating robust overall conclusions. Despite the complexity and heterogeneity of methodology, prognostic features, clinical presentation, and treatment approaches, studies included in this review suggest that ML tools may have the potential to predict treatment outcomes of psychosis accurately. Future studies need to focus on refining feature characterization, validating prediction models, and evaluate their translation in real-world clinical practice.

Del Fabro Lorenzo, Bondi Elena, Serio Francesca, Maggioni Eleonora, D’Agostino Armando, Brambilla Paolo

2023-Mar-02

Pathology Pathology

Low gamma-butyrobetaine dioxygenase (BBOX1) expression as a prognostic biomarker in patients with clear cell renal cell carcinoma: a machine learning approach.

In The journal of pathology. Clinical research

Gamma-butyrobetaine dioxygenase (BBOX1) is a catalyst for the conversion of gamma-butyrobetaine to l-carnitine, which is detected in normal renal tubules. The purpose of this study was to analyze the prognosis, immune response, and genetic alterations associated with low BBOX1 expression in patients with clear cell renal cell carcinoma (RCC). We analyzed the relative influence of BBOX1 on survival using machine learning and investigated drugs that can inhibit renal cancer cells with low BBOX1 expression. We analyzed clinicopathologic factors, survival rates, immune profiles, and gene sets according to BBOX1 expression in a total of 857 patients with kidney cancer from the Hanyang University Hospital cohort (247 cases) and The Cancer Genome Atlas (610 cases). We employed immunohistochemical staining, gene set enrichment analysis, in silico cytometry, pathway network analyses, in vitro drug screening, and gradient boosting machines. BBOX1 expression in RCC was decreased compared with that in normal tissues. Low BBOX1 expression was associated with poor prognosis, decreased CD8+ T cells, and increased neutrophils. In gene set enrichment analyses, low BBOX1 expression was related to gene sets with oncogenic activity and a weak immune response. In pathway network analysis, BBOX1 was linked to regulation of various T cells and programmed death-ligand 1. In vitro drug screening showed that midostaurin, BAY-61-3606, GSK690693, and linifanib inhibited the growth of RCC cells with low BBOX1 expression. Low BBOX1 expression in patients with RCC is related to short survival time and reduced CD8+ T cells; midostaurin, among other drugs, may have enhanced therapeutic effects in this context.

Kim Kyu-Shik, Moon Kyoung Min, Min Kyueng-Whan, Jung Woon Yong, Shin Su-Jin, Lee Seung Wook, Kwon Mi Jung, Kim Dong-Hoon, Oh Sukjoong, Noh Yung-Kyun

2023-Mar-02

CD8+ T cells, low gamma-butyrobetaine dioxygenase, machine learning, renal cell carcinoma

General General

The Role of ChatGPT, Generative Language Models and Artificial Intelligence in Medical Education: A Conversation with ChatGPT - and a Call for Papers.

In JMIR medical education

ChatGPT is a generative language model tool launched by Open-AI on November 20, 2022, enabling the public to converse with a machine on a broad range of topics. In January 2023, ChatGPT reached over 100 million users, making it the fastest growing consumer application to date. This interview with ChatGPT (Feb 13, 2023 version) is part 2 of a larger interview with ChatGPT. It provides a snapshot of the current capabilities of ChatGPT and illustrates the vast potential for medical education, research and practice, but also hints at current problems and limitations. In this conversation with JMIR publisher Gunther Eysenbach, ChatGPT generates some ideas on how to use chatbots in medical education, and during the interview illustrates its' capabilities to generate a virtual patient simulation, generates quizzes for medical students, critiques a simulated doctor-patient communication, critiques research articles, comments on methods to detect machine-generated text to ensure academic integrity, generates a curriculum for health professionals to learn about AI, and helps with drafting a call for papers for a new theme issue to be launched in JMIR Medical Education on ChatGPT. The conversation also highlights the importance of proper "prompting". While the language generator does make occasional mistakes, it admits these when challenged. The interview provides a fascinating glimpse into the capabilities of ChatGPT and the future of AI-supported medical education. Due to the impact of this new technology on medical education, JMIR Medical Education is launching a call for papers for a new e-collection and theme issue. We are soliciting papers that for example cover the following topics: 1) The potential of generative language models and AI for medical education, including their use in teaching and learning, clinical decision-making, and patient care, 2) The role of generative language models and AI in enhancing the quality of medical education, including the use of simulations, virtual patients, and other forms of digital learning resources. 3) Use of generative language models for automated essay grading and feedback in medical education 4) The development and evaluation of virtual patients generated by generative language models 5) Measuring the quality of information and simulations generated by generative language models, and strategies for improving the quality through proper prompting and other approaches. 6) Training medical students and healthcare professionals on AI and specifically on generative language models, including the development of curricula and instructional materials. 7) Ethical and legal issues related to the use of generative language models and AI in medical education, including issues related to data privacy, bias, and transparency. 8) Academic integrity issues and policies describing how medical schools allow or disallow use of generative language models. The initial call for papers was entirely machine-generated by ChatGPT, but will be edited by the human guest editors of the theme issue.

Eysenbach Gunther

2023-Mar-02

Radiology Radiology

A deep convolutional neural network ensemble for composite identification of pulmonary nodules and incidental findings on routine PET/CT.

In Clinical radiology

AIM : To evaluate primary and secondary pathologies of interest using an artificial intelligence (AI) platform, AI-Rad Companion, on low-dose computed tomography (CT) series from integrated positron-emission tomography (PET)/CT to detect CT findings that might be overlooked.

MATERIALS AND METHODS : One hundred and eighty-nine sequential patients who had undergone PET/CT were included. Images were evaluated using an ensemble of convolutional neural networks (AI-Rad Companion, Siemens Healthineers, Erlangen, Germany). The primary outcome was detection of pulmonary nodules for which the accuracy, identity, and intra-rater reliability was calculated. For secondary outcomes (binary detection of coronary artery calcium, aortic ectasia, vertebral height loss), accuracy and diagnostic performance were calculated.

RESULTS : The overall per-nodule accuracy for detection of lung nodules was 0.847. The overall sensitivity and specificity for detection of lung nodules was 0.915 and 0.781. The overall per-patient accuracy for AI detection of coronary artery calcium, aortic ectasia, and vertebral height loss was 0.979, 0.966, and 0.840, respectively. The sensitivity and specificity for coronary artery calcium was 0.989 and 0.969. The sensitivity and specificity for aortic ectasia was 0.806 and 1.

CONCLUSION : The neural network ensemble accurately assessed the number of pulmonary nodules and presence of coronary artery calcium and aortic ectasia on low-dose CT series of PET/CT. The neural network was highly specific for the diagnosis of vertebral height loss, but not sensitive. The use of the AI ensemble can help radiologists and nuclear medicine physicians to catch CT findings that might be overlooked.

Chamberlin J H, Smith C, Schoepf U J, Nance S, Elojeimy S, O’Doherty J, Baruah D, Burt J R, Varga-Szemes A, Kabakus I M

2023-Feb-16