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

Radiology Radiology

Visual-Linguistic Causal Intervention for Radiology Report Generation

ArXiv Preprint

Automatic radiology report generation is essential for computer-aided diagnosis and medication guidance. Importantly, automatic radiology report generation (RRG) can relieve the heavy burden of radiologists by generating medical reports automatically from visual-linguistic data relations. However, due to the spurious correlations within image-text data induced by visual and linguistic biases, it is challenging to generate accurate reports that reliably describe abnormalities. Besides, the cross-modal confounder is usually unobservable and difficult to be eliminated explicitly. In this paper, we mitigate the cross-modal data bias for RRG from a new perspective, i.e., visual-linguistic causal intervention, and propose a novel Visual-Linguistic Causal Intervention (VLCI) framework for RRG, which consists of a visual deconfounding module (VDM) and a linguistic deconfounding module (LDM), to implicitly deconfound the visual-linguistic confounder by causal front-door intervention. Specifically, the VDM explores and disentangles the visual confounder from the patch-based local and global features without object detection due to the absence of universal clinic semantic extraction. Simultaneously, the LDM eliminates the linguistic confounder caused by salient visual features and high-frequency context without constructing specific dictionaries. Extensive experiments on IU-Xray and MIMIC-CXR datasets show that our VLCI outperforms the state-of-the-art RRG methods significantly. Source code and models are available at https://github.com/WissingChen/VLCI.

Weixing Chen, Yang Liu, Ce Wang, Guanbin Li, Jiarui Zhu, Liang Lin

2023-03-16

General General

[Artificial intelligence and psychiatry: questions from psychiatrists to ChatGPT].

In Revue medicale suisse

Psychiatrists and psychotherapists specialising in the fields of addiction, personality disorders, ADHD and suicidal crisis, we questioned the ChatGPT artificial intelligence program in order to form an opinion on the quality of its answers to questions on these subjects. Our aim is to satisfy our curiosity about these emerging tools. On the other hand, we want to assess the relevance of the answers in order to know whether relatives and patients can use them safely. In this article, we comment on the question-and-answer dialogue with the artificial intelligence program in the light of the literature.

Prada Paco, Perroud Nader, Thorens Gabriel

2023-Mar-15

General General

Diode Characteristics in Magnetic Domain Wall Devices via Geometrical Pinning for Neuromorphic Computing.

In ACS applied materials & interfaces ; h5-index 147.0

Neuromorphic computing (NC) is considered a potential vehicle for implementing energy-efficient artificial intelligence. To realize NC, several technologies are being investigated. Among them, the spin-orbit torque (SOT)-driven domain wall (DW) devices are one of the potential candidates. Researchers have proposed different device designs to achieve neurons and synapses, the building blocks of NC. However, the experimental realization of DW device-based NC is only at the primeval stage. Here, we have studied pine-tree DW devices, based on the Laplace pressure on the elastic DWs, for achieving synaptic functionalities and diode-like characteristics. We demonstrate an asymmetric pinning strength for DW motion in two opposite directions to show the potential of these devices as DW diodes. We have used micromagnetic simulations to understand the experimental findings and to estimate the Laplace pressure for various design parameters. The study provides a strategy to fabricate a multifunctional DW device, exhibiting synaptic properties and diode characteristics.

Rahaman Hasibur, Kumar Durgesh, Chung Hong Jing, Maddu Ramu, Lim Sze Ter, Jin Tianli, Piramanayagam S N

2023-Mar-15

asymmetric pinning potential, domain wall devices, domain wall diodes, neuromorphic computing, pine-tree devices, spināˆ’orbit torque, synapses

General General

A methylation clock model of mild SARS-CoV-2 infection provides insight into immune dysregulation.

In Molecular systems biology

DNA methylation comprises a cumulative record of lifetime exposures superimposed on genetically determined markers. Little is known about methylation dynamics in humans following an acute perturbation, such as infection. We characterized the temporal trajectory of blood epigenetic remodeling in 133 participants in a prospective study of young adults before, during, and after asymptomatic and mildly symptomatic SARS-CoV-2 infection. The differential methylation caused by asymptomatic or mildly symptomatic infections was indistinguishable. While differential gene expression largely returned to baseline levels after the virus became undetectable, some differentially methylated sites persisted for months of follow-up, with a pattern resembling autoimmune or inflammatory disease. We leveraged these responses to construct methylation-based machine learning models that distinguished samples from pre-, during-, and postinfection time periods, and quantitatively predicted the time since infection. The clinical trajectory in the young adults and in a diverse cohort with more severe outcomes was predicted by the similarity of methylation before or early after SARS-CoV-2 infection to the model-defined postinfection state. Unlike the phenomenon of trained immunity, the postacute SARS-CoV-2 epigenetic landscape we identify is antiprotective.

Mao Weiguang, Miller Clare M, Nair Venugopalan D, Ge Yongchao, Amper Mary Anne S, Cappuccio Antonio, George Mary-Catherine, Goforth Carl W, Guevara Kristy, Marjanovic Nada, Nudelman German, Pincas Hanna, Ramos Irene, Sealfon Rachel S G, Soares-Schanoski Alessandra, Vangeti Sindhu, Vasoya Mital, Weir Dawn L, Zaslavsky Elena, Kim-Schulze Seunghee, Gnjatic Sacha, Merad Miriam, Letizia Andrew G, Troyanskaya Olga G, Sealfon Stuart C, Chikina Maria

2023-Mar-15

DNA methylation, SARS-CoV-2, machine learning model, temporal dynamics, trained immunity

Public Health Public Health

Prediction of Future Caries in 1-Year-Old Children via the Salivary Microbiome.

In Journal of dental research ; h5-index 65.0

Dental caries is the most common chronic disease in children that causes negative effects on their health, development, and well-being. Early preventive interventions are key to reduce early childhood caries prevalence. An efficient strategy is to provide risk-based targeted prevention; however, this requires an accurate caries risk predictor, which is still lacking for infants before caries onset. We aimed to develop a caries prediction model based on the salivary microbiome of caries-free 1-y-old children. Using a nested case-control design within a prospective cohort study, we selected 30 children based on their caries status at 1-y follow-up (at 2 y old): 10 children who remained caries-free, 10 who developed noncavitated caries, and 10 who developed cavitated caries. Saliva samples collected at baseline before caries onset were analyzed through 16S rRNA gene sequencing. The results of β diversity analysis showed a significant difference in salivary microbiome composition between children who remained caries-free and those who developed cavitated caries at 2 y old (analysis of similarities, Benjamini-Hochberg corrected, P = 0.042). The relative abundance of Prevotella nanceiensis, Leptotrichia sp. HMT 215, Prevotella melaninogenica, and Campylobacter concisus in children who remained caries-free was significantly higher than in children who developed cavitated caries (Wilcoxon rank sum test, P = 0.024, 0.040, 0.049, and 0.049, respectively). These taxa were also identified as biomarkers for children who remained caries-free (linear discriminant analysis effect size, linear discriminant analysis score = 3.69, 3.74, 3.53, and 3.46). A machine learning model based on these 4 species distinguished between 1-y-old children who did and did not develop cavitated caries at 2 y old, with an accuracy of 80%, sensitivity of 80%, and specificity of 80% (area under the curve, 0.8; 95% CI, 44.4 to 97.5). Our findings suggest that these salivary microbial biomarkers could assist in predicting future caries in caries-free 1-y-old children and, upon validation, are promising for development into an adjunctive tool for caries risk prediction for prevention and monitoring.

Raksakmanut R, Thanyasrisung P, Sritangsirikul S, Kitsahawong K, Seminario A L, Pitiphat W, Matangkasombut O

2023-Mar-15

biomarkers, dental caries, infant, machine learning, microbiota, saliva

General General

Developing a machine learning model to detect diagnostic uncertainty in clinical documentation.

In Journal of hospital medicine

BACKGROUND AND OBJECTIVE : Diagnostic uncertainty, when unrecognized or poorly communicated, can result in diagnostic error. However, diagnostic uncertainty is challenging to study due to a lack of validated identification methods. This study aims to identify distinct linguistic patterns associated with diagnostic uncertainty in clinical documentation.

DESIGN, SETTING AND PARTICIPANTS : This case-control study compares the clinical documentation of hospitalized children who received a novel uncertain diagnosis (UD) diagnosis label during their admission to a set of matched controls. Linguistic analyses identified potential linguistic indicators (i.e., words or phrases) of diagnostic uncertainty that were then manually reviewed by a linguist and clinical experts to identify those most relevant to diagnostic uncertainty. A natural language processing program categorized medical terminology into semantic types (i.e., sign or symptom), from which we identified a subset of these semantic types that both categorized reliably and were relevant to diagnostic uncertainty. Finally, a competitive machine learning modeling strategy utilizing the linguistic indicators and semantic types compared different predictive models for identifying diagnostic uncertainty.

RESULTS : Our cohort included 242 UD-labeled patients and 932 matched controls with a combination of 3070 clinical notes. The best-performing model was a random forest, utilizing a combination of linguistic indicators and semantic types, yielding a sensitivity of 89.4% and a positive predictive value of 96.7%.

CONCLUSION : Expert labeling, natural language processing, and machine learning methods combined with human validation resulted in highly predictive models to detect diagnostic uncertainty in clinical documentation and represent a promising approach to detecting, studying, and ultimately mitigating diagnostic uncertainty in clinical practice.

Marshall Trisha L, Nickels Lindsay C, Brady Patrick W, Edgerton Ezra J, Lee James J, Hagedorn Philip A

2023-Mar-15