Sensitivity of an international notification system for wildlife diseases: A case study using the OIE-WAHIS data on tularemia.

The World Organization for Animal Health (OIE) has recently developed a Wildlife Health Framework to respond to the need of members to manage the risk from emerging diseases at the animal-human-ecosystem interface. One of its objectives is to improve surveillance systems, early detection and notification of wildlife diseases. Members share information on disease occurrence by reporting through the OIE World Animal Health Information System (OIE-WAHIS-formerly known as 'WAHIS'). To evaluate the capacity of a surveillance system to detect disease events, it is important to quantify the gap between all known events and those officially notified to the OIE. This study used capture-recapture analysis to estimate the sensitivity of the OIE-WAHIS system for a OIE-listed wildlife disease by comparing information from publicly available sources to identify undetected events. This article presents a case study of the occurrence of tularemia in lagomorphs among selected North American and European countries during the period 2014-2019. First, an analysis using three data sources (OIE-WAHIS, ProMED, WHO-EIOS [Epidemic Intelligence from Open Sources]) was conducted. Subsequent analysis then explored the model integrating information from a fourth source (scientific literature collected in PubMed). Two models were built to evaluate both the sensitivity of the OIE-WAHIS using media reports (ProMED and WHO-EIOS), which is likely to represent current closer to real-time events, and published scientific data, which is more useful for retrospective analysis. Using the three-source approach, the predicted number of tularemia events was 93 (95% CI: 75-114), with an OIE-WAHIS sensitivity of 90%. In the four-source approach, the number of predicted events increased to 120 (95% CI: 99-143), dropping the sensitivity of the OIE-WAHIS to 70%. The results indicate a good sensitivity of the OIE-WAHIS system using the three-source approach, but lower sensitivity when including information from the scientific literature. Further analysis should be undertaken to identify diseases and regions for which international reporting presents a low sensitivity. This will enable evaluation and prioritization of underreported OIE-listed wildlife diseases and identify areas of focus as part of the Wildlife Health Framework. This study also highlights the need for stronger collaborations between academia and National Veterinary Services to enhance surveillance systems for notifiable diseases.

Towards radiologist-level cancer risk assessment in CT lung screening using deep learning.

PURPOSE: Lung cancer is the leading cause of cancer mortality in the US, responsible for more deaths than breast, prostate, colon and pancreas cancer combined and large population studies have indicated that low-dose computed tomography (CT) screening of the chest can significantly reduce this death rate. Recently, the usefulness of Deep Learning (DL) models for lung cancer risk assessment has been demonstrated. However, in many cases model performances are evaluated on small/medium size test sets, thus not providing strong model generalization and stability guarantees which are necessary for clinical adoption. In this work, our goal is to contribute towards clinical adoption by investigating a deep learning framework on larger and heterogeneous datasets while also comparing to state-of-the-art models. METHODS: Three low-dose CT lung cancer screening datasets were used: National Lung Screening Trial (NLST, n = 3410), Lahey Hospital and Medical Center (LHMC, n = 3154) data, Kaggle competition data (from both stages, n = 1397 + 505) and the University of Chicago data (UCM, a subset of NLST, annotated by radiologists, n = 132). At the first stage, our framework employs a nodule detector; while in the second stage, we use both the image context around the nodules and nodule features as inputs to a neural network that estimates the malignancy risk for the entire CT scan. We trained our algorithm on a part of the NLST dataset, and validated it on the other datasets. Special care was taken to ensure there was no patient overlap between the train and validation sets. RESULTS AND CONCLUSIONS: The proposed deep learning model is shown to: (a) generalize well across all three data sets, achieving AUC between 86% to 94%, with our external test-set (LHMC) being at least twice as large compared to other works; (b) have better performance than the widely accepted PanCan Risk Model, achieving 6 and 9% better AUC score in our two test sets; (c) have improved performance compared to the state-of-the-art represented by the winners of the Kaggle Data Science Bowl 2017 competition on lung cancer screening; (d) have comparable performance to radiologists in estimating cancer risk at a patient level.

Precision medicine integrating whole-genome sequencing, comprehensive metabolomics, and advanced imaging.

Genome sequencing has established clinical utility for rare disease diagnosis. While increasing numbers of individuals have undergone elective genome sequencing, a comprehensive study surveying genome-wide disease-associated genes in adults with deep phenotyping has not been reported. Here we report the results of a 3-y precision medicine study with a goal to integrate whole-genome sequencing with deep phenotyping. A cohort of 1,190 adult participants (402 female [33.8%]; mean age, 54 y [range 20 to 89+]; 70.6% European) had whole-genome sequencing, and were deeply phenotyped using metabolomics, advanced imaging, and clinical laboratory tests in addition to family/medical history. Of 1,190 adults, 206 (17.3%) had at least 1 genetic variant with pathogenic (P) or likely pathogenic (LP) assessment that suggests a predisposition of genetic risk. A multidisciplinary clinical team reviewed all reportable findings for the assessment of genotype and phenotype associations, and 137 (11.5%) had genotype and phenotype associations. A high percentage of genotype and phenotype associations (>75%) was observed for dyslipidemia (n = 24), cardiomyopathy, arrhythmia, and other cardiac diseases (n = 42), and diabetes and endocrine diseases (n = 17). A lack of genotype and phenotype associations, a potential burden for patient care, was observed in 69 (5.8%) individuals with P/LP variants. Genomics and metabolomics associations identified 61 (5.1%) heterozygotes with phenotype manifestations affecting serum metabolite levels in amino acid, lipid and cofactor, and vitamin pathways. Our descriptive analysis provides results on the integration of whole-genome sequencing and deep phenotyping for clinical assessments in adults.

An unsupervised learning approach to identify novel signatures of health and disease from multimodal data.

BACKGROUND: Modern medicine is rapidly moving towards a data-driven paradigm based on comprehensive multimodal health assessments. Integrated analysis of data from different modalities has the potential of uncovering novel biomarkers and disease signatures. METHODS: We collected 1385 data features from diverse modalities, including metabolome, microbiome, genetics, and advanced imaging, from 1253 individuals and from a longitudinal validation cohort of 1083 individuals. We utilized a combination of unsupervised machine learning methods to identify multimodal biomarker signatures of health and disease risk. RESULTS: Our method identified a set of cardiometabolic biomarkers that goes beyond standard clinical biomarkers. Stratification of individuals based on the signatures of these biomarkers identified distinct subsets of individuals with similar health statuses. Subset membership was a better predictor for diabetes than established clinical biomarkers such as glucose, insulin resistance, and body mass index. The novel biomarkers in the diabetes signature included 1-stearoyl-2-dihomo-linolenoyl-GPC and 1-(1-enyl-palmitoyl)-2-oleoyl-GPC. Another metabolite, cinnamoylglycine, was identified as a potential biomarker for both gut microbiome health and lean mass percentage. We identified potential early signatures for hypertension and a poor metabolic health outcome. Additionally, we found novel associations between a uremic toxin, p-cresol sulfate, and the abundance of the microbiome genera Intestinimonas and an unclassified genus in the Erysipelotrichaceae family. CONCLUSIONS: Our methodology and results demonstrate the potential of multimodal data integration, from the identification of novel biomarker signatures to a data-driven stratification of individuals into disease subtypes and stages-an essential step towards personalized, preventative health risk assessment.

Multi-Label Active Learning Algorithms for Image Classification: Overview and Future Promise.

Image classification is a key task in image understanding, and multi-label image classification has become a popular topic in recent years. However, the success of multi-label image classification is closely related to the way of constructing a training set. As active learning aims to construct an effective training set through iteratively selecting the most informative examples to query labels from annotators, it was introduced into multi-label image classification. Accordingly, multi-label active learning is becoming an important research direction. In this work, we first review existing multi-label active learning algorithms for image classification. These algorithms can be categorized into two top groups from two aspects respectively: sampling and annotation. The most important component of multi-label active learning is to design an effective sampling strategy that actively selects the examples with the highest informativeness from an unlabeled data pool, according to various information measures. Thus, different informativeness measures are emphasized in this survey. Furthermore, this work also makes a deep investigation on existing challenging issues and future promises in multi-label active learning with a focus on four core aspects: example dimension, label dimension, annotation, and application extension.

Biomonitoring for traditional herbal medicinal products using DNA metabarcoding and single molecule, real-time sequencing

Global concerns have been paid to the potential hazard of traditional herbal medicinal products (THMPs). Substandard and counterfeit THMPs, including traditional Chinese patent medicine, health foods, dietary supplements, etc. are potential threats to public health. Recent marketplace studies using DNA barcoding have determined that the current quality control methods are not sufficient for ensuring the presence of authentic herbal ingredients and detection of contaminants/adulterants. An efficient biomonitoring method for THMPs is of great needed. Herein, metabarcoding and single-molecule, real-time (SMRT) sequencing were used to detect the multiple ingredients in Jiuwei Qianghuo Wan (JWQHW), a classical herbal prescription widely used in China for the last 800 years. Reference experimental mixtures and commercial JWQHW products from the marketplace were used to confirm the method. Successful SMRT sequencing results recovered 5416 and 4342 circular-consensus sequencing (CCS) reads belonging to the ITS2 and regions. The results suggest that with the combination of metabarcoding and SMRT sequencing, it is repeatable, reliable, and sensitive enough to detect species in the THMPs, and the error in SMRT sequencing did not affect the ability to identify multiple prescribed species and several adulterants/contaminants. It has the potential for becoming a valuable tool for the biomonitoring of multi-ingredient THMPs.

Directly detected (55)Mn MRI: application to phantoms for human hyperpolarized (13)C MRI development.

In this work we demonstrate for the first time directly detected manganese-55 ((55)Mn) magnetic resonance imaging (MRI) using a clinical 3T MRI scanner designed for human hyperpolarized (13)C clinical studies with no additional hardware modifications. Due to the similar frequency of the (55)Mn and (13)C resonances, the use of aqueous permanganate for large, signal-dense, and cost-effective "(13)C" MRI phantoms was investigated, addressing the clear need for new phantoms for these studies. Due to 100% natural abundance, higher intrinsic sensitivity, and favorable relaxation properties, (55)Mn MRI of aqueous permanganate demonstrates dramatically increased sensitivity over typical (13)C phantom MRI, at greatly reduced cost as compared with large (13)C-enriched phantoms. A large sensitivity advantage (22-fold) was demonstrated. A cylindrical phantom (d=8 cm) containing concentrated aqueous sodium permanganate (2.7 M) was scanned rapidly by (55)Mn MRI in a human head coil tuned for (13)C, using a balanced steady state free precession acquisition. The requisite penetration of radiofrequency magnetic fields into concentrated permanganate was investigated by experiments and high frequency electromagnetic simulations, and found to be sufficient for (55)Mn MRI with reasonably sized phantoms. A sub-second slice-selective acquisition yielded mean image signal-to-noise ratio of ~60 at 0.5 cm(3) spatial resolution, distributed with minimum central signal ~40% of the maximum edge signal. We anticipate that permanganate phantoms will be very useful for testing HP (13)C coils and methods designed for human studies.

Kinetic and perfusion modeling of hyperpolarized (13)C pyruvate and urea in cancer with arbitrary RF flip angles.

The accurate detection and characterization of cancerous tissue is still a major problem for the clinical management of individual cancer patients and for monitoring their response to therapy. MRI with hyperpolarized agents is a promising technique for cancer characterization because it can non-invasively provide a local assessment of the tissue metabolic profile. In this work, we measured the kinetics of hyperpolarized [1-(13)C] pyruvate and (13)C-urea in prostate and liver tumor models using a compressed sensing dynamic MRSI method. A kinetic model fitting method was developed that incorporated arbitrary RF flip angle excitation and measured a pyruvate to lactate conversion rate, Kpl, of 0.050 and 0.052 (1/s) in prostate and liver tumors, respectively, which was significantly higher than Kpl in healthy tissues [Kpl =0.028 (1/s), P<0.001]. Kpl was highly correlated to the total lactate to total pyruvate signal ratio (correlation coefficient =0.95). We additionally characterized the total pyruvate and urea perfusion, as in cancerous tissue there is both existing vasculature and neovascularization as different kinds of lesions surpass the normal blood supply, including small circulation disturbance in some of the abnormal vessels. A significantly higher perfusion of pyruvate (accounting for conversion to lactate and alanine) relative to urea perfusion was seen in cancerous tissues (liver cancer and prostate cancer) compared to healthy tissues (P<0.001), presumably due to high pyruvate uptake in tumors.

Quantitative measurement of cancer metabolism using stimulated echo hyperpolarized carbon-13 MRS.

PURPOSE: Magnetic resonance spectroscopy of hyperpolarized substrates allows for the observation of label exchange catalyzed by enzymes providing a powerful tool to investigate tissue metabolism and potentially kinetics in vivo. However, the accuracy of current methods to calculate kinetic parameters has been limited by T1 relaxation effects, extracellular signal contributions, and reduced precision at lower signal-to-noise ratio. THEORY AND METHODS: To address these challenges, we investigated a new modeling technique using metabolic activity decomposition-stimulated echo acquisition mode. The metabolic activity decomposition-stimulated echo acquisition mode technique separates exchanging from nonexchanging metabolites providing twice the information as conventional techniques. RESULTS: This allowed for accurate measurements of rates of conversion and of multiple T1 values simultaneously using a single acquisition. CONCLUSION: The additional measurement of T1 values for the reaction metabolites provides further biological information about the cellular environment of the metabolites. The new technique was investigated through simulations and in vivo studies of transgenic mouse models of cancer demonstrating improved assessments of kinetic rate constants and new T1 relaxation value measurements for hyperpolarized (13) C-pyruvate, (13) C-lactate, and (13) C-alanine.

A rapid method for direct detection of metabolic conversion and magnetization exchange with application to hyperpolarized substrates.

In this work, we present a new MR spectroscopy approach for directly observing nuclear spins that undergo exchange, metabolic conversion, or, generally, any frequency shift during a mixing time. Unlike conventional approaches to observe these processes, such as exchange spectroscopy (EXSY), this rapid approach requires only a single encoding step and thus is readily applicable to hyperpolarized MR in which the magnetization is not replenished after T(1) decay and RF excitations. This method is based on stimulated-echoes and uses phase-sensitive detection in conjunction with precisely chosen echo times in order to separate spins generated during the mixing time from those present prior to mixing. We are calling the method Metabolic Activity Decomposition Stimulated-echo Acquisition Mode or MAD-STEAM. We have validated this approach as well as applied it in vivo to normal mice and a transgenic prostate cancer mouse model for observing pyruvate-lactate conversion, which has been shown to be elevated in numerous tumor types. In this application, it provides an improved measure of cellular metabolism by separating [1-(13)C]-lactate produced in tissue by metabolic conversion from [1-(13)C]-lactate that has flowed into the tissue or is in the blood. Generally, MAD-STEAM can be applied to any system in which spins undergo a frequency shift.