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

Radiology Radiology

Robust framework for COVID-19 identication from a multicenter dataset of chest CT scans.

In PloS one ; h5-index 176.0

The main objective of this study is to develop a robust deep learning-based framework to distinguish COVID-19, Community-Acquired Pneumonia (CAP), and Normal cases based on volumetric chest CT scans, which are acquired in different imaging centers using different scanners and technical settings. We demonstrated that while our proposed model is trained on a relatively small dataset acquired from only one imaging center using a specific scanning protocol, it performs well on heterogeneous test sets obtained by multiple scanners using different technical parameters. We also showed that the model can be updated via an unsupervised approach to cope with the data shift between the train and test sets and enhance the robustness of the model upon receiving a new external dataset from a different center. More specifically, we extracted the subset of the test images for which the model generated a confident prediction and used the extracted subset along with the training set to retrain and update the benchmark model (the model trained on the initial train set). Finally, we adopted an ensemble architecture to aggregate the predictions from multiple versions of the model. For initial training and development purposes, an in-house dataset of 171 COVID-19, 60 CAP, and 76 Normal cases was used, which contained volumetric CT scans acquired from one imaging center using a single scanning protocol and standard radiation dose. To evaluate the model, we collected four different test sets retrospectively to investigate the effects of the shifts in the data characteristics on the model's performance. Among the test cases, there were CT scans with similar characteristics as the train set as well as noisy low-dose and ultra-low-dose CT scans. In addition, some test CT scans were obtained from patients with a history of cardiovascular diseases or surgeries. This dataset is referred to as the "SPGC-COVID" dataset. The entire test dataset used in this study contains 51 COVID-19, 28 CAP, and 51 Normal cases. Experimental results indicate that our proposed framework performs well on all test sets achieving total accuracy of 96.15% (95%CI: [91.25-98.74]), COVID-19 sensitivity of 96.08% (95%CI: [86.54-99.5]), CAP sensitivity of 92.86% (95%CI: [76.50-99.19]), Normal sensitivity of 98.04% (95%CI: [89.55-99.95]) while the confidence intervals are obtained using the significance level of 0.05. The obtained AUC values (One class vs Others) are 0.993 (95%CI: [0.977-1]), 0.989 (95%CI: [0.962-1]), and 0.990 (95%CI: [0.971-1]) for COVID-19, CAP, and Normal classes, respectively. The experimental results also demonstrate the capability of the proposed unsupervised enhancement approach in improving the performance and robustness of the model when being evaluated on varied external test sets.

Khademi Sadaf, Heidarian Shahin, Afshar Parnian, Enshaei Nastaran, Naderkhani Farnoosh, Rafiee Moezedin Javad, Oikonomou Anastasia, Shafiee Akbar, Babaki Fard Faranak, Plataniotis Konstantinos N, Mohammadi Arash

2023

oncology Oncology

The early neutrophil-committed progenitors aberrantly differentiate into immunoregulatory monocytes during emergency myelopoiesis.

In Cell reports ; h5-index 119.0

Inflammatory stimuli cause a state of emergency myelopoiesis leading to neutrophil-like monocyte expansion. However, their function, the committed precursors, or growth factors remain elusive. In this study we find that Ym1+Ly6Chi monocytes, an immunoregulatory entity of neutrophil-like monocytes, arise from progenitors of neutrophil 1 (proNeu1). Granulocyte-colony stimulating factor (G-CSF) favors the production of neutrophil-like monocytes through previously unknown CD81+CX3CR1lo monocyte precursors. GFI1 promotes the differentiation of proNeu2 from proNeu1 at the cost of producing neutrophil-like monocytes. The human counterpart of neutrophil-like monocytes that also expands in response to G-CSF is found in CD14+CD16- monocyte fraction. The human neutrophil-like monocytes are discriminated from CD14+CD16- classical monocytes by CXCR1 expression and the capacity to suppress T cell proliferation. Collectively, our findings suggest that the aberrant expansion of neutrophil-like monocytes under inflammatory conditions is a process conserved between mouse and human, which may be beneficial for the resolution of inflammation.

Ikeda Naoki, Kubota Hiroaki, Suzuki Risa, Morita Mitsuki, Yoshimura Ayana, Osada Yuya, Kishida Keigo, Kitamura Daiki, Iwata Ayaka, Yotsumoto Satoshi, Kurotaki Daisuke, Nishimura Koutarou, Nishiyama Akira, Tamura Tomohiko, Kamatani Takashi, Tsunoda Tatsuhiko, Murakawa Miyako, Asahina Yasuhiro, Hayashi Yoshihiro, Harada Hironori, Harada Yuka, Yokota Asumi, Hirai Hideyo, Seki Takao, Kuwahara Makoto, Yamashita Masakatsu, Shichino Shigeyuki, Tanaka Masato, Asano Kenichi

2023-Feb-28

CP: Immunology, CXCR1, G-CSF, Ym1, demand-adapted myelopoiesis, emergency myelopoiesis, machine learning, monocyte, neutrophil-like monocyte, ontogeny

Public Health Public Health

Unassisted Clinicians Versus Deep Learning-Assisted Clinicians in Image-Based Cancer Diagnostics: Systematic Review With Meta-analysis.

In Journal of medical Internet research ; h5-index 88.0

BACKGROUND : A number of publications have demonstrated that deep learning (DL) algorithms matched or outperformed clinicians in image-based cancer diagnostics, but these algorithms are frequently considered as opponents rather than partners. Despite the clinicians-in-the-loop DL approach having great potential, no study has systematically quantified the diagnostic accuracy of clinicians with and without the assistance of DL in image-based cancer identification.

OBJECTIVE : We systematically quantified the diagnostic accuracy of clinicians with and without the assistance of DL in image-based cancer identification.

METHODS : PubMed, Embase, IEEEXplore, and the Cochrane Library were searched for studies published between January 1, 2012, and December 7, 2021. Any type of study design was permitted that focused on comparing unassisted clinicians and DL-assisted clinicians in cancer identification using medical imaging. Studies using medical waveform-data graphics material and those investigating image segmentation rather than classification were excluded. Studies providing binary diagnostic accuracy data and contingency tables were included for further meta-analysis. Two subgroups were defined and analyzed, including cancer type and imaging modality.

RESULTS : In total, 9796 studies were identified, of which 48 were deemed eligible for systematic review. Twenty-five of these studies made comparisons between unassisted clinicians and DL-assisted clinicians and provided sufficient data for statistical synthesis. We found a pooled sensitivity of 83% (95% CI 80%-86%) for unassisted clinicians and 88% (95% CI 86%-90%) for DL-assisted clinicians. Pooled specificity was 86% (95% CI 83%-88%) for unassisted clinicians and 88% (95% CI 85%-90%) for DL-assisted clinicians. The pooled sensitivity and specificity values for DL-assisted clinicians were higher than for unassisted clinicians, at ratios of 1.07 (95% CI 1.05-1.09) and 1.03 (95% CI 1.02-1.05), respectively. Similar diagnostic performance by DL-assisted clinicians was also observed across the predefined subgroups.

CONCLUSIONS : The diagnostic performance of DL-assisted clinicians appears better than unassisted clinicians in image-based cancer identification. However, caution should be exercised, because the evidence provided in the reviewed studies does not cover all the minutiae involved in real-world clinical practice. Combining qualitative insights from clinical practice with data-science approaches may improve DL-assisted practice, although further research is required.

TRIAL REGISTRATION : PROSPERO CRD42021281372; https://www.crd.york.ac.uk/prospero/display_record.php?RecordID=281372.

Xue Peng, Si Mingyu, Qin Dongxu, Wei Bingrui, Seery Samuel, Ye Zichen, Chen Mingyang, Wang Sumeng, Song Cheng, Zhang Bo, Ding Ming, Zhang Wenling, Bai Anying, Yan Huijiao, Dang Le, Zhao Yuqian, Rezhake Remila, Zhang Shaokai, Qiao Youlin, Qu Yimin, Jiang Yu

2023-Mar-02

cancer diagnosis, deep learning, meta-analysis, systematic review

General General

Smartphone Global Positioning System-Based System to Assess Mobility in Health Research: Development, Accuracy, and Usability Study.

In JMIR rehabilitation and assistive technologies

BACKGROUND : As global positioning system (GPS) measurement is getting more precise and affordable, health researchers can now objectively measure mobility using GPS sensors. Available systems, however, often lack data security and means of adaptation and often rely on a permanent internet connection.

OBJECTIVE : To overcome these issues, we aimed to develop and test an easy-to-use, easy-to-adapt, and offline working app using smartphone sensors (GPS and accelerometry) for the quantification of mobility parameters.

METHODS : An Android app, a server backend, and a specialized analysis pipeline have been developed (development substudy). Parameters of mobility by the study team members were extracted from the recorded GPS data using existing and newly developed algorithms. Test measurements were performed with participants to complete accuracy and reliability tests (accuracy substudy). Usability was examined by interviewing community-dwelling older adults after 1 week of device use, followed by an iterative app design process (usability substudy).

RESULTS : The study protocol and the software toolchain worked reliably and accurately, even under suboptimal conditions, such as narrow streets and rural areas. The developed algorithms had high accuracy (97.4% correctness, F1-score=0.975) in distinguishing dwelling periods from moving intervals. The accuracy of the stop/trip classification is fundamental to second-order analyses such as the time out of home, as they rely on a precise discrimination between the 2 classes. The usability of the app and the study protocol was piloted with older adults, which showed low barriers and easy implementation into daily routines.

CONCLUSIONS : Based on accuracy analyses and users' experience with the proposed system for GPS assessments, the developed algorithm showed great potential for app-based estimation of mobility in diverse health research contexts, including mobility patterns of community-dwelling older adults living in rural areas.

INTERNATIONAL REGISTERED REPORT IDENTIFIER (IRRID) : RR2-10.1186/s12877-021-02739-0.

Spang Robert P, Haeger Christine, Mümken Sandra A, Brauer Max, Voigt-Antons Jan-Niklas, Gellert Paul

2023-Mar-02

geographic information system, geoinformatics, out-of-home mobility, prevention medicine, rehabilitation

Public Health Public Health

The Health Care Utilization and Medical Costs in Long-Term Follow-Up of Children Diagnosed With Leukemia, Solid Tumor, or Brain Tumor: Population-Based Study Using the National Health Insurance Claims Data.

In JMIR public health and surveillance

BACKGROUND : Childhood cancer survivors are at a high risk of medical consequences of their disease and treatment. There is growing information about the long-term health issues of childhood cancer survivors; however, there are very few studies describing the health care utilization and costs for this unique population. Understanding their utilization of health care services and costs will provide the basis for developing strategies to better serve these individuals and potentially reduce the cost.

OBJECTIVE : This study aims to determine the utilization of health services and costs for long-term survivors of childhood cancer in Taiwan.

METHODS : This is a nationwide, population-based, retrospective case-control study. We analyzed the claims data of the National Health Insurance that covers 99% of the Taiwanese population of 25.68 million. A total of 33,105 children had survived for at least 5 years after the first appearance of a diagnostic code of cancer or a benign brain tumor before the age of 18 years from 2000 to 2010 with follow-up to 2015. An age- and gender-matched control group of 64,754 individuals with no cancer was randomly selected for comparison. Utilization was compared between the cancer and no cancer groups by χ2 test. The annual medical expense was compared by the Mann-Whitney U test and Kruskal-Wallis rank-sum test.

RESULTS : At a median follow-up of 7 years, childhood cancer survivors utilized a significantly higher proportion of medical center, regional hospital, inpatient, and emergency services in contrast to no cancer individuals: 57.92% (19,174/33,105) versus 44.51% (28,825/64,754), 90.66% (30,014/33,105) versus 85.70% (55,493/64,754), 27.19% (9000/33,105) versus 20.31% (13,152/64,754), and 65.26% (21,604/33,105) versus 59.36% (38,441/64,754), respectively (all P<.001). The annual total expense (median, interquartile range) of childhood cancer survivors was significantly higher than that of the comparison group (US $285.56, US $161.78-US $535.80 per year vs US $203.90, US $118.98-US $347.55 per year; P<.001). Survivors with female gender, diagnosis before the age of 3 years, and diagnosis of brain cancer or a benign brain tumor had significantly higher annual outpatient expenses (all P<.001). Moreover, the analysis of outpatient medication costs showed that hormonal and neurological medications comprised the 2 largest costs in brain cancer and benign brain tumor survivors.

CONCLUSIONS : Survivors of childhood cancer and a benign brain tumor had higher utilization of advanced health resources and higher costs of care. The design of the initial treatment plan minimizing long-term consequences, early intervention strategies, and survivorship programs have the potential to mitigate costs of late effects due to childhood cancer and its treatment.

Miser James S, Shia Ben-Chang, Kao Yi-Wei, Liu Yen-Lin, Chen Shih-Yen, Ho Wan-Ling

2023-Mar-02

brain tumor, cancer survivor, children, cost of care, health care, health resource, leukemia, long-term follow-up, population-based study, solid tumor

General General

Assessment of Natural Language Processing of Electronic Health Records to Measure Goals-of-Care Discussions as a Clinical Trial Outcome.

In JAMA network open

IMPORTANCE : Many clinical trial outcomes are documented in free-text electronic health records (EHRs), making manual data collection costly and infeasible at scale. Natural language processing (NLP) is a promising approach for measuring such outcomes efficiently, but ignoring NLP-related misclassification may lead to underpowered studies.

OBJECTIVE : To evaluate the performance, feasibility, and power implications of using NLP to measure the primary outcome of EHR-documented goals-of-care discussions in a pragmatic randomized clinical trial of a communication intervention.

DESIGN, SETTING, AND PARTICIPANTS : This diagnostic study compared the performance, feasibility, and power implications of measuring EHR-documented goals-of-care discussions using 3 approaches: (1) deep-learning NLP, (2) NLP-screened human abstraction (manual verification of NLP-positive records), and (3) conventional manual abstraction. The study included hospitalized patients aged 55 years or older with serious illness enrolled between April 23, 2020, and March 26, 2021, in a pragmatic randomized clinical trial of a communication intervention in a multihospital US academic health system.

MAIN OUTCOMES AND MEASURES : Main outcomes were natural language processing performance characteristics, human abstractor-hours, and misclassification-adjusted statistical power of methods of measuring clinician-documented goals-of-care discussions. Performance of NLP was evaluated with receiver operating characteristic (ROC) curves and precision-recall (PR) analyses and examined the effects of misclassification on power using mathematical substitution and Monte Carlo simulation.

RESULTS : A total of 2512 trial participants (mean [SD] age, 71.7 [10.8] years; 1456 [58%] female) amassed 44 324 clinical notes during 30-day follow-up. In a validation sample of 159 participants, deep-learning NLP trained on a separate training data set from identified patients with documented goals-of-care discussions with moderate accuracy (maximal F1 score, 0.82; area under the ROC curve, 0.924; area under the PR curve, 0.879). Manual abstraction of the outcome from the trial data set would require an estimated 2000 abstractor-hours and would power the trial to detect a risk difference of 5.4% (assuming 33.5% control-arm prevalence, 80% power, and 2-sided α = .05). Measuring the outcome by NLP alone would power the trial to detect a risk difference of 7.6%. Measuring the outcome by NLP-screened human abstraction would require 34.3 abstractor-hours to achieve estimated sensitivity of 92.6% and would power the trial to detect a risk difference of 5.7%. Monte Carlo simulations corroborated misclassification-adjusted power calculations.

CONCLUSIONS AND RELEVANCE : In this diagnostic study, deep-learning NLP and NLP-screened human abstraction had favorable characteristics for measuring an EHR outcome at scale. Adjusted power calculations accurately quantified power loss from NLP-related misclassification, suggesting that incorporation of this approach into the design of studies using NLP would be beneficial.

Lee Robert Y, Kross Erin K, Torrence Janaki, Li Kevin S, Sibley James, Cohen Trevor, Lober William B, Engelberg Ruth A, Curtis J Randall

2023-Mar-01