Favorable efficacy of S-1 treatment for locoregionally advanced cutaneous squamous cell carcinoma in the head and neck region.

Cutaneous squamous cell carcinoma is usually treated with surgery; however, locoregionally advanced cutaneous squamous cell carcinoma can be difficult to resect. Although recent guidelines from Western countries recommend using anti-programmed cell death protein 1 (PD-1) antibodies, including cemiplimab and pembrolizumab, there are no approved anti-PD-1 antibodies for locoregional cutaneous squamous cell carcinoma in Asian countries. S-1 is an oral drug with a low incidence of severe toxicity that can be used for head and neck cancers, including head and neck locoregional cutaneous squamous cell carcinoma, in Japan. We retrospectively evaluated patients with head and neck locoregional cutaneous squamous cell carcinoma treated with S-1 at two Japanese institutions (2008-2022). The initial dosage was determined by the body surface area (<1.25 m2 : 80 mg/day, 1.25-1.5 m2 : 100 mg/day, ≥1.5 m2: 120 mg/day) for 28 consecutive days. The outcome measures were objective response rate (ORR), progression-free survival (PFS), and overall survival (OS). Fourteen patients were included. The ORR was 78%, and the complete response (CR) rate was 64.3%. The median PFS and OS were not reached (NR) (95% confidence interval [CI], 5.9 months-NR) and NR (95% CI, 13.8 months-NR), respectively. The 12-month PFS and OS rates were 51% and 85%, respectively. Six of the nine patients who achieved CR showed no recurrence during the follow-up period (median follow-up, 24.7 months). After CR, three patients experienced recurrence. Among these, two resumed S-1 treatment and subsequently underwent salvage surgery, resulting in a sustained absence of recurrence. One patient developed lung metastasis and died, although S-1 therapy was resumed. Only one patient (7.1%) developed grade 3 anemia. S-1 showed favorable efficacy and low toxicity in patients with head and neck locoregionally advanced cutaneous squamous cell carcinoma. S-1 may be a good alternative to the anti-PD-1 antibody for treating head and neck locoregionally advanced squamous cell carcinoma.

Anesthetic management of fetal pulmonary valvuloplasty: A case report.

Anesthesia management of fetal pulmonary valvuloplasty (FPV) is difficult, requiring careful consideration of both the mother and the fetus. Few reports have been published on specific anesthesia implementation and intraoperative management. We report the case of a pregnant woman who was treated with FPV under combined spinal epidural anesthesia (CSEA) with dexmedetomidine in the second trimester of pregnancy. Meanwhile, the application of fetal anesthesia through the umbilical vein was optimal. During the operation, the vital signs of the pregnant woman were stable with no complications and the fetal bradycardia was corrected by intracardiac injection of epinephrine. Four months postoperatively, a boy was born alive by full-term transvaginal delivery. CSEA may be a suitable anesthesia method for FPV surgery. Nevertheless, maternal hemodynamic stability maintenance, effective fetal anesthesia, and timely fetal resuscitation were necessary.

Innate Immune System Regulated by Stimulator of Interferon Genes, a Cytosolic DNA Sensor, Regulates Endothelial Function.

Background Sterile inflammation caused by metabolic disorders impairs endothelial function; however, the underlying mechanism by which hyperglycemia induces inflammation remains obscure. Recent studies have suggested that stimulator of interferon genes (STING), a key cytosolic DNA sensor in the innate immune system, contributes to the pathogenesis of inflammatory diseases. This study examines the role of the STING in endothelial dysfunction in streptozotocin-induced diabetic mice. Methods and Results Injection of streptozotocin promoted the expression of STING and DNA damage markers in the aorta of wild-type mice. Streptozotocin elevated blood glucose and lipid levels in both wild-type and STING-deficient mice, which showed no statistical differences. Genetic deletion of STING ameliorated endothelial dysfunction as determined by the vascular relaxation in response to acetylcholine (P<0.001) and increased endothelial nitric oxide synthase phosphorylation in the aorta (P<0.05) in STZ-injected mice. Endothelium-independent vascular response to sodium nitroprusside did not differ. Treatment with a direct STING agonist, cyclic GMP-AMP, or mitochondrial DNA increased inflammatory molecule expression (eg, VCAM1 and IFNB) and decreased endothelial nitric oxide synthase phosphorylation in human umbilical vein endothelial cells, partially through the STING pathway. Cyclic GMP-AMP significantly impaired endothelial function of aortic segments obtained from wild-type mice, which was ameliorated in the presence of C-176, a STING inhibitor, or a neutralizing interferon-ß antibody. Furthermore, the administration of C-176 ameliorated endothelial dysfunction in STZ-induced diabetic mice (P<0.01). Conclusions The DNA damage response regulated by STING impairs endothelial function. STING signaling may be a potential therapeutic target of endothelial dysfunction caused by hyperglycemia.

Anesthetic drug concentrations and placental transfer rate in fetus between term and preterm infants, twins, and singletons.

Objective: This study aims to determine the drug concentration of etomidate, remifentanil, and rocuronium bromide for general anesthesia in fetus as well as the placental transport rate between term and preterm delivery, twins, and singleton. Study design: Sixty parturients with 72 fetuses undergoing cesarean section under general anesthesia were included. According to whether the fetus was a twin or premature, parturients were divided into Group I (term singleton), Group II (premature singleton), Group III (term twins), and Group IV (premature twins). The preoperative demographic characteristics and laboratory examination of parturients, hemodynamic indicators, the Apgar score of neonates at 1, 5, and 10 min after delivery and at specific assigned values, umbilical artery blood gas analysis results, neonatal weight, and resuscitative measures were recorded. Anesthetic drug concentrations in maternal arterial (MA), umbilical arterial (UA), and umbilical venous (UV) blood were detected by Ultra Performance Liquid Chromatography Tandem Mass Spectrometry (UPLC-MS/MS). Result: No significant differences were observed in the concentrations of etomidate, remifentanil, and rocuronium bromide in MA, UV, and UA blood, or in the UV/MA and UA/UV ratios between term and preterm infants, twins, and singletons. Moreover, there was no variation in the anesthetic drug concentration among each pair of twins. Additionally, no correlation was found between the neonatal weight and the plasma concentrations of anesthetic drugs in UV and UA blood, except for remifentanil in UA blood. Conclusion: Preterm or twin deliveries do not affect the neonatal concentration of etomidate, remifentanil, and rocuronium bromide used in general anesthesia for cesarean sections. Clinical Trial Registration: www.chictr.org.cn, identifier ChiCTR2100046547.

Remifentanil at a Relatively Elevated Dose in Active Phase is Safe and More Suitable Than Fixed Lower Dose for Intravenous Labor Analgesia.

Background: Intravenous labor analgesia is recommended as an alternative for parturients who have contraindications to epidural analgesia. There are several opioid analgesics and different administering regimens used in the clinic. This study aimed to compare the effectiveness and safety of two intravenous remifentanil dosage regimens in the first labor stage. Patients and Methods: One hundred and fifteen parturients with a contraindication to epidural analgesia but were willing to receive systemic labor analgesia were randomized into group A received a fixed dose of remifentanil throughout the first stage of labor, and group B received an elevated dose of remifentanil during the active phase of the first stage both by patient-controlled analgesia (PCA). Maternal numerical rating scale (NRS) pain score and oxygen desaturation, sedation efficacy, satisfaction, as well as maternal and fetal adverse reactions were recorded and compared. Results: The mean NRS pain scores before analgesia and in the latent phase showed no statistically significant difference between the two groups (P > 0.05). However, during the active phase, group B demonstrated significantly lower mean NRS pain scores and lowest pain score compared to group A (P < 0.05). Furthermore, group B exhibited higher overall sedation scores and satisfaction scores in comparison to group A (P < 0.05). The incidence of adverse reactions between the two groups was similar (P > 0.05). Conclusion: Relatively elevated intravenous dosage of remifentanil with PCA during the active phase in the first stage of labor is safe and more effective than a fixed-dosage regimen for labor analgesia. Trial Registration: This study was registered with ChiCTR on 24/08/2021 with trial identification number: ChiCTR2100050247. First participant was recruited on 31/08/2021. The last patient was recruited on 12/08/2022.

Knockdown of mechanosensitive adaptor Hic-5 ameliorates post-traumatic osteoarthritis in rats through repression of MMP-13.

Osteoarthritis (OA) is the most common joint disease associated with articular cartilage destruction. Matrix metalloproteinase-13 (MMP-13) has an essential role in OA pathogenesis by degradation of collagen II, a major component of articular cartilage. Hydrogen peroxide-inducible clone-5 (Hic-5; TGFB1I1), a transforming growth factor-ß-inducible mechanosensor, has previously been reported to promote OA pathogenesis by upregulating MMP-13 expression in mouse osteoarthritic lesions. In our current study, immunohistochemical analysis showed that Hic-5 protein expression was increased in human OA cartilage compared with normal cartilage. Functional experiments demonstrated that Hic-5 and MMP-13 expression was increased by mechanical stress, and mechanical stress-induced MMP-13 expression was suppressed by Hic-5 siRNA in human chondrocytes. Moreover, intracellular localization of Hic-5 shifted to the nucleus from focal adhesions in human chondrocytes subjected to mechanical stress, and nuclear Hic-5 increased MMP-13 gene expression. In vivo, intra-articular injection of Hic-5 siRNA decreased the Osteoarthritis Research Society International score and MMP-13 protein expression in articular cartilage of OA rats. Our findings suggest that Hic-5 regulates transcription of MMP-13 in human chondrocytes, and Hic-5 may be a novel therapeutic target for OA because OA progression was suppressed by intra-articular injection of Hic-5 siRNA in rats.

Deep Neural Network Augments Performance of Junior Residents in Diagnosing COVID-19 Pneumonia on Chest Radiographs.

Chest X-rays (CXRs) are essential in the preliminary radiographic assessment of patients affected by COVID-19. Junior residents, as the first point-of-contact in the diagnostic process, are expected to interpret these CXRs accurately. We aimed to assess the effectiveness of a deep neural network in distinguishing COVID-19 from other types of pneumonia, and to determine its potential contribution to improving the diagnostic precision of less experienced residents. A total of 5051 CXRs were utilized to develop and assess an artificial intelligence (AI) model capable of performing three-class classification, namely non-pneumonia, non-COVID-19 pneumonia, and COVID-19 pneumonia. Additionally, an external dataset comprising 500 distinct CXRs was examined by three junior residents with differing levels of training. The CXRs were evaluated both with and without AI assistance. The AI model demonstrated impressive performance, with an Area under the ROC Curve (AUC) of 0.9518 on the internal test set and 0.8594 on the external test set, which improves the AUC score of the current state-of-the-art algorithms by 1.25% and 4.26%, respectively. When assisted by the AI model, the performance of the junior residents improved in a manner that was inversely proportional to their level of training. Among the three junior residents, two showed significant improvement with the assistance of AI. This research highlights the novel development of an AI model for three-class CXR classification and its potential to augment junior residents' diagnostic accuracy, with validation on external data to demonstrate real-world applicability. In practical use, the AI model effectively supported junior residents in interpreting CXRs, boosting their confidence in diagnosis. While the AI model improved junior residents' performance, a decline in performance was observed on the external test compared to the internal test set. This suggests a domain shift between the patient dataset and the external dataset, highlighting the need for future research on test-time training domain adaptation to address this issue.

Prolonged Anesthesia Induction to Delivery Time Did Not Influence Plasma Remifentanil Concentration in Neonates.

Objective: Remifentanil, in combination with etomidate and sevoflurane, is commonly used in clinics for general anesthesia induction in cesarean section (CS). This study aimed to evaluate the correlation between the induction to delivery (I-D) time and neonatal plasma drug concentration and anesthesia, as well as its effects on neonates. Methods: Fifty-two parturients in whom general anesthesia was induced for CS were divided into group A (I-D<8 min) and group B (I-D≥8 min). Maternal arterial (MA), umbilical venous (UV), and umbilical arterial (UA) blood samples were collected at delivery to analyze the remifentanil and etomidate concentrations using liquid chromatography-tandem mass spectrometry. Results: There were no statistically significant differences between the two groups in terms of plasma concentrations of remifentanil in the MA, UA, and UV blood (P > 0.05). The plasma concentration of etomidate in MA and UV was higher in group A than that in group B (P<0.05), whereas the UA/UV ratio of etomidate was higher in group B than that in group A (P<0.05). The Spearman rank correlation test showed no correlation between the I-D time and plasma remifentanil concentration in the MA, UA, and UV plasma (P>0.05). The concentrations of etomidate in the MA and UV were negatively correlated with the I-D time (P < 0.05). Conclusion: Prolonged I-D time did not significantly influence the maternal or neonatal plasma concentration of remifentanil. It is safe to administer remifentanil target-controlled infusion in combination with etomidate and sevoflurane for general anesthesia induction during CS.

High Areal Capacity, Long Cycle Life Li-Air Batteries Enabled by Nano/Micro Hierarchical Porous Cathode.

The limited cycle life of Li-air batteries (LABs) with high areal capacity remains the chief challenge that hinders their practical applications. Here, the study proposes a hierarchical porous electrode (HPE) design strategy, in which porous MnO nanoflowers are built into mesopore/macropore electrodes through a combination of chemical dealloying and physical de-templating procedures. The MnO nanoflowers with 10-30 nm pore provides active sites to catalyze the O2 reduction and decomposition of discharged products. The 5-10 µm macroscopic pores in the cathode serve as channels of O2 transportation and facilitate the electrolyte permeation. The proposed HPE exhibits a full discharge capacity of 17.49 mAh cm-2 and stable cycle life >2000 h with a limited capacity of 6 mAh cm-2 . These results suggest that the HPE design strategy for LABs can simultaneously provide large capacity and robust cycle life, which is promising for advanced metal-air batteries.

Development and testing of a multi-lingual Natural Language Processing-based deep learning system in 10 languages for COVID-19 pandemic crisis: A multi-center study.

Purpose: The COVID-19 pandemic has drastically disrupted global healthcare systems. With the higher demand for healthcare and misinformation related to COVID-19, there is a need to explore alternative models to improve communication. Artificial Intelligence (AI) and Natural Language Processing (NLP) have emerged as promising solutions to improve healthcare delivery. Chatbots could fill a pivotal role in the dissemination and easy accessibility of accurate information in a pandemic. In this study, we developed a multi-lingual NLP-based AI chatbot, DR-COVID, which responds accurately to open-ended, COVID-19 related questions. This was used to facilitate pandemic education and healthcare delivery. Methods: First, we developed DR-COVID with an ensemble NLP model on the Telegram platform (https://t.me/drcovid_nlp_chatbot). Second, we evaluated various performance metrics. Third, we evaluated multi-lingual text-to-text translation to Chinese, Malay, Tamil, Filipino, Thai, Japanese, French, Spanish, and Portuguese. We utilized 2,728 training questions and 821 test questions in English. Primary outcome measurements were (A) overall and top 3 accuracies; (B) Area Under the Curve (AUC), precision, recall, and F1 score. Overall accuracy referred to a correct response for the top answer, whereas top 3 accuracy referred to an appropriate response for any one answer amongst the top 3 answers. AUC and its relevant matrices were obtained from the Receiver Operation Characteristics (ROC) curve. Secondary outcomes were (A) multi-lingual accuracy; (B) comparison to enterprise-grade chatbot systems. The sharing of training and testing datasets on an open-source platform will also contribute to existing data. Results: Our NLP model, utilizing the ensemble architecture, achieved overall and top 3 accuracies of 0.838 [95% confidence interval (CI): 0.826-0.851] and 0.922 [95% CI: 0.913-0.932] respectively. For overall and top 3 results, AUC scores of 0.917 [95% CI: 0.911-0.925] and 0.960 [95% CI: 0.955-0.964] were achieved respectively. We achieved multi-linguicism with nine non-English languages, with Portuguese performing the best overall at 0.900. Lastly, DR-COVID generated answers more accurately and quickly than other chatbots, within 1.12-2.15 s across three devices tested. Conclusion: DR-COVID is a clinically effective NLP-based conversational AI chatbot, and a promising solution for healthcare delivery in the pandemic era.

mHealth App to Facilitate Remote Care for Patients With COVID-19: Rapid Development of the DrCovid+ App.

BACKGROUND: The 2019 novel COVID-19 has severely burdened the health care system through its rapid transmission. Mobile health (mHealth) is a viable solution to facilitate remote monitoring and continuity of care for patients with COVID-19 in a home environment. However, the conceptualization and development of mHealth apps are often time and labor-intensive and are laden with concerns relating to data security and privacy. Implementing mHealth apps is also a challenging feat as language-related barriers limit adoption, whereas its perceived lack of benefits affects sustained use. The rapid development of an mHealth app that is cost-effective, secure, and user-friendly will be a timely enabler. OBJECTIVE: This project aimed to develop an mHealth app, DrCovid+, to facilitate remote monitoring and continuity of care for patients with COVID-19 by using the rapid development approach. It also aimed to address the challenges of mHealth app adoption and sustained use. METHODS: The Rapid Application Development approach was adopted. Stakeholders including decision makers, physicians, nurses, health care administrators, and research engineers were engaged. The process began with requirements gathering to define and finalize the project scope, followed by an iterative process of developing a working prototype, conducting User Acceptance Tests, and improving the prototype before implementation. Co-designing principles were applied to ensure equal collaborative efforts and collective agreement among stakeholders. RESULTS: DrCovid+ was developed on Telegram Messenger and hosted on a cloud server. It features a secure patient enrollment and data interface, a multilingual communication channel, and both automatic and personalized push messaging. A back-end dashboard was also developed to collect patients' vital signs for remote monitoring and continuity of care. To date, 400 patients have been enrolled into the system, amounting to 2822 hospital bed-days saved. CONCLUSIONS: The rapid development and implementation of DrCovid+ allowed for timely clinical care management for patients with COVID-19. It facilitated early patient hospital discharge and continuity of care while addressing issues relating to data security and labor-, time-, and cost-effectiveness. The use case for DrCovid+ may be extended to other medical conditions to advance patient care and empowerment within the community, thereby meeting existing and rising population health challenges.

Contrastive domain adaptation with consistency match for automated pneumonia diagnosis.

Pneumonia can be difficult to diagnose since its symptoms are too variable, and the radiographic signs are often very similar to those seen in other illnesses such as a cold or influenza. Deep neural networks have shown promising performance in automated pneumonia diagnosis using chest X-ray radiography, allowing mass screening and early intervention to reduce the severe cases and death toll. However, they usually require many well-labelled chest X-ray images for training to achieve high diagnostic accuracy. To reduce the need for training data and annotation resources, we propose a novel method called Contrastive Domain Adaptation with Consistency Match (CDACM). It transfers the knowledge from different but relevant datasets to the unlabelled small-size target dataset and improves the semantic quality of the learnt representations. Specifically, we design a conditional domain adversarial network to exploit discriminative information conveyed in the predictions to mitigate the domain gap between the source and target datasets. Furthermore, due to the small scale of the target dataset, we construct a feature cloud for each target sample and leverage contrastive learning to extract more discriminative features. Lastly, we propose adaptive feature cloud expansion to push the decision boundary to a low-density area. Unlike most existing transfer learning methods that aim only to mitigate the domain gap, our method instead simultaneously considers the domain gap and the data deficiency problem of the target dataset. The conditional domain adaptation and the feature cloud generation of our method are learning jointly to extract discriminative features in an end-to-end manner. Besides, the adaptive feature cloud expansion improves the model's generalisation ability in the target domain. Extensive experiments on pneumonia and COVID-19 diagnosis tasks demonstrate that our method outperforms several state-of-the-art unsupervised domain adaptation approaches, which verifies the effectiveness of CDACM for automated pneumonia diagnosis using chest X-ray imaging.

Analgesic effect of ultrasound-guided erector spinae plane block (espb) in general anesthesia for cesarean section: a randomized controlled trial.

BACKGROUND: The analgesic effects of erector spinae plane block in general anesthesia for cesarean section and recovery from puerperae remain unclear. METHODS: Sixty patients with contraindications for spinal anesthesia who required general anesthesia for cesarean section were enrolled and randomly divided into the erector spinal plane block (ESPB) combined with the general anesthesia group (group E) and general anesthesia group (group G). Group E received bilateral ESPB (20 ml of 0.25% ropivacaine on each side) under ultrasound guidance 30 min before general anesthesia. The primary outcomes were the number of patient-controlled intravenous analgesia (PCIA) boluses, and Bruggemann comfort scale (BCS) scores at 2 h, 6 h, 12 h, and 24 h after operation. The second outcome was intraoperative anesthesia dosage, fetal delivery time, puerperae emergence time, visual analog scale (VAS) at 2 h, 6 h, 12 h, and 24 h after operation, and incidence of nausea and vomiting. Heart rate (HR) and mean arterial pressure (MAP) were recorded 10 min before the start of anesthesia (T0), at the induction of anesthesia (T1), at skin incision (T2), and fetal delivery (T3), and immediately after surgery (T4). RESULTS: The number of PCIA boluses was lower in group E than in group G (P < 0.001). The BCS score increased at 2 h and 6 h after the operation in group E (P < 0.05), while the VAS score significantly decreased in group E at the same time (P < 0.05). Compared with group G, the doses of propofol and remifentanil were significantly decreased in group E (P < 0.001), the emergence time of puerperae was shortened (P = 0.003), and the incidence of nausea and vomiting was significantly decreased (P = 0.014). CONCLUSION: Ultrasound-guided ESPB applied to general anesthesia for a cesarean section can significantly reduce the required dose of general anesthetic drugs, shorten the recovery time of the puerperae, and improve postoperative analgesia. TRIAL REGISTRATION: www. CLINICALTRIALS: gov under the number ChiCTR2200056337 (04-02-2022).

Case Report: Anesthetic Management of Cesarean Section in a Patient With Paraplegia.

Background: With the advancement of medical science and rehabilitative care, more women with spinal cord injury (SCI) can conceive. However, autonomic hyperreflexia due to SCI complicates anesthesia management during cesarean sections. Case Presentation: This study reports the anesthesia management in a woman with paraplegia with a T6 SCI lesion who underwent a cesarean section. It also reviews the anesthesia strategies used in other studies. Spinal anesthesia with a low concentration of ropivacaine was administered along with dexmedetomidine for sedation. Stable hemodynamics were achieved without complications. Conclusions: Based on the reported case and literature review, we conclude that the intrathecal block is the preferred choice for women with paraplegia who require cesarean section if the lumbar bone structure allows puncture attempts.

Predictive model for the intraoperative unresectability of hilar cholangiocarcinoma: Reducing futile surgical exploration.

INTRODUCTION: Surgical exploration is widely performed in hilar cholangiocarcinoma (HCCA), but the intraoperative resectability rate is only 60%-80%. Exploration substantially increases pain and mental stress, and the costs and length of hospital stay are considerably increased. Identifying preoperative risk factors associated with unresectability could decrease unnecessary exploration. MATERIALS AND METHODS: In total, 440 HCCA patients from multiple centers were enrolled. Those receiving surgical exploration were divided into the resected and unresected groups. Morphological variables including Bismuth classification, lymph node metastasis and vessel invasion were obtained from radiological exams. Logistic regression for the training cohort was used to identify risk factors for unresectability, and a nomogram was constructed to calculate the unresectability rate. A calibration curve assessed the power of the nomogram. RESULTS: Among 311 patients receiving surgical exploration, 45 (14.7%) were unresectable by intraoperative judgment. Compared with the resected group, unresected patients had similar costs (p = 0.359) and lengths of hospital stay (p = 0.439). Multivariable logistic regression of the training cohort (235 patients) revealed that CA125, Bismuth-Corlette type IV, lymph node metastasis and hepatic artery invasion were risk factors for unresectability. Liver atrophy (p = 0.374) and portal vein invasion (p = 0.114) were not risk factors. The nomogram was constructed based on the risk factors. The concordance index (C-index) values of the calibration curve for predicting the unresectability rate of the training and validation (76 patients) cohorts were 0.900 (95% CI, 0.835-0.966) and 0.829 (95% CI, 0.546-0.902), respectively. CONCLUSION: Analysis of preoperative factors could reveal intraoperative unresectability and reduce futile surgical explorations, ultimately benefiting HCCA patients.

Diagnostic Performance of a Deep Learning Model Deployed at a National COVID-19 Screening Facility for Detection of Pneumonia on Frontal Chest Radiographs.

(1) Background: Chest radiographs are the mainstay of initial radiological investigation in this COVID-19 pandemic. A reliable and readily deployable artificial intelligence (AI) algorithm that detects pneumonia in COVID-19 suspects can be useful for screening or triage in a hospital setting. This study has a few objectives: first, to develop a model that accurately detects pneumonia in COVID-19 suspects; second, to assess its performance in a real-world clinical setting; and third, by integrating the model with the daily clinical workflow, to measure its impact on report turn-around time. (2) Methods: The model was developed from the NIH Chest-14 open-source dataset and fine-tuned using an internal dataset comprising more than 4000 CXRs acquired in our institution. Input from two senior radiologists provided the reference standard. The model was integrated into daily clinical workflow, prioritising abnormal CXRs for expedited reporting. Area under the receiver operating characteristic curve (AUC), F1 score, sensitivity, and specificity were calculated to characterise diagnostic performance. The average time taken by radiologists in reporting the CXRs was compared against the mean baseline time taken prior to implementation of the AI model. (3) Results: 9431 unique CXRs were included in the datasets, of which 1232 were ground truth-labelled positive for pneumonia. On the "live" dataset, the model achieved an AUC of 0.95 (95% confidence interval (CI): 0.92, 0.96) corresponding to a specificity of 97% (95% CI: 0.97, 0.98) and sensitivity of 79% (95% CI: 0.72, 0.84). No statistically significant degradation of diagnostic performance was encountered during clinical deployment, and report turn-around time was reduced by 22%. (4) Conclusion: In real-world clinical deployment, our model expedites reporting of pneumonia in COVID-19 suspects while preserving diagnostic performance without significant model drift.

DeepLensNet: Deep Learning Automated Diagnosis and Quantitative Classification of Cataract Type and Severity.

PURPOSE: To develop deep learning models to perform automated diagnosis and quantitative classification of age-related cataract from anterior segment photographs. DESIGN: DeepLensNet was trained by applying deep learning models to the Age-Related Eye Disease Study (AREDS) dataset. PARTICIPANTS: A total of 18 999 photographs (6333 triplets) from longitudinal follow-up of 1137 eyes (576 AREDS participants). METHODS: Deep learning models were trained to detect and quantify nuclear sclerosis (NS; scale 0.9-7.1) from 45-degree slit-lamp photographs and cortical lens opacity (CLO; scale 0%-100%) and posterior subcapsular cataract (PSC; scale 0%-100%) from retroillumination photographs. DeepLensNet performance was compared with that of 14 ophthalmologists and 24 medical students. MAIN OUTCOME MEASURES: Mean squared error (MSE). RESULTS: On the full test set, mean MSE for DeepLensNet was 0.23 (standard deviation [SD], 0.01) for NS, 13.1 (SD, 1.6) for CLO, and 16.6 (SD, 2.4) for PSC. On a subset of the test set (substantially enriched for positive cases of CLO and PSC), for NS, mean MSE for DeepLensNet was 0.23 (SD, 0.02), compared with 0.98 (SD, 0.24; P = 0.000001) for the ophthalmologists and 1.24 (SD, 0.34; P = 0.000005) for the medical students. For CLO, mean MSE was 53.5 (SD, 14.8), compared with 134.9 (SD, 89.9; P = 0.003) for the ophthalmologists and 433.6 (SD, 962.1; P = 0.0007) for the medical students. For PSC, mean MSE was 171.9 (SD, 38.9), compared with 176.8 (SD, 98.0; P = 0.67) for the ophthalmologists and 398.2 (SD, 645.4; P = 0.18) for the medical students. In external validation on the Singapore Malay Eye Study (sampled to reflect the cataract severity distribution in AREDS), the MSE for DeepSeeNet was 1.27 for NS and 25.5 for PSC. CONCLUSIONS: DeepLensNet performed automated and quantitative classification of cataract severity for all 3 types of age-related cataract. For the 2 most common types (NS and CLO), the accuracy was significantly superior to that of ophthalmologists; for the least common type (PSC), it was similar. DeepLensNet may have wide potential applications in both clinical and research domains. In the future, such approaches may increase the accessibility of cataract assessment globally. The code and models are available at https://github.com/ncbi/deeplensnet.