Accurate, Robust, and Scalable Machine Abstraction of Mayo Endoscopic Subscores From Colonoscopy Reports.

BACKGROUND: The Mayo endoscopic subscore (MES) is an important quantitative measure of disease activity in ulcerative colitis. Colonoscopy reports in routine clinical care usually characterize ulcerative colitis disease activity using free text description, limiting their utility for clinical research and quality improvement. We sought to develop algorithms to classify colonoscopy reports according to their MES. METHODS: We annotated 500 colonoscopy reports from 2 health systems. We trained and evaluated 4 classes of algorithms. Our primary outcome was accuracy in identifying scorable reports (binary) and assigning an MES (ordinal). Secondary outcomes included learning efficiency, generalizability, and fairness. RESULTS: Automated machine learning models achieved 98% and 97% accuracy on the binary and ordinal prediction tasks, outperforming other models. Binary models trained on the University of California, San Francisco data alone maintained accuracy (96%) on validation data from Zuckerberg San Francisco General. When using 80% of the training data, models remained accurate for the binary task (97% [n = 320]) but lost accuracy on the ordinal task (67% [n = 194]). We found no evidence of bias by gender (P = .65) or area deprivation index (P = .80). CONCLUSIONS: We derived a highly accurate pair of models capable of classifying reports by their MES and recognizing when to abstain from prediction. Our models were generalizable on outside institution validation. There was no evidence of algorithmic bias. Our methods have the potential to enable retrospective studies of treatment effectiveness, prospective identification of patients meeting study criteria, and quality improvement efforts in inflammatory bowel diseases.

Our accurate pair of models automatically classify colonoscopy reports by Mayo endoscopic subscore and abstain from prediction appropriately. Our methods can enable large-scale electronic health record studies of treatment effectiveness, prospective identification of patients for clinical trials, and quality improvement efforts in ulcerative colitis.

Unique portal venous phase imaging discordance between CEUS and MRI: a valuable predictor of intrahepatic cholangiocarcinoma?

PURPOSE: We have long noted unique portal venous phase (PVP) imaging discordance of focal liver masses between CEUS, showing rapid marked washout, and MRI, showing progressive or sustained enhancement. We postulate association of this unique discordance with intrahepatic cholangiocarcinoma (ICC) and causal relationship to different contrast agent behavior. We investigate this unique discordance, propose its clinical significance for ICC diagnosis, and confirm further histologic associations. METHODS: Cases were collected within our CEUS department and from pathology records over a ten-year interval. This retrospective review includes 99 patients, 73 with confirmed ICC and 26 other diagnoses, showing unique PVP discordance. The CEUS and MRI enhancement characteristics were compared for all patients. RESULTS: Unique discordance is identified in 67/73 (92%) ICC and difference between the PVP appearance on MRI and CEUS is statistically significant (p <  0.0001). Arterial phase enhancement did not show statistically significant difference between CEUS and MRI, p >  0.05. Other diagnoses showing unique discordance include especially lymphoma (n = 7), sclerosed hemangioma (n = 6), HCC (n = 4), metastases (n = 2), and other rare entities. CONCLUSION: ICC shows this discrepant intermodality enhancement pattern in a statistically significant number of cases and should be considered along with other LR-M features in at-risk patients. Discordance is also rarely seen in a number of other liver lesions.

ClotCatcher: a novel natural language model to accurately adjudicate venous thromboembolism from radiology reports.

INTRODUCTION: Accurate identification of venous thromboembolism (VTE) is critical to develop replicable epidemiological studies and rigorous predictions models. Traditionally, VTE studies have relied on international classification of diseases (ICD) codes which are inaccurate - leading to misclassification bias. Here, we developed ClotCatcher, a novel deep learning model that uses natural language processing to detect VTE from radiology reports. METHODS: Radiology reports to detect VTE were obtained from patients admitted to Emory University Hospital (EUH) and Grady Memorial Hospital (GMH). Data augmentation was performed using the Google PEGASUS paraphraser. This data was then used to fine-tune ClotCatcher, a novel deep learning model. ClotCatcher was validated on both the EUH dataset alone and GMH dataset alone. RESULTS: The dataset contained 1358 studies from EUH and 915 studies from GMH (n = 2273). The dataset contained 1506 ultrasound studies with 528 (35.1%) studies positive for VTE, and 767 CT studies with 91 (11.9%) positive for VTE. When validated on the EUH dataset, ClotCatcher performed best (AUC = 0.980) when trained on both EUH and GMH dataset without paraphrasing. When validated on the GMH dataset, ClotCatcher performed best (AUC = 0.995) when trained on both EUH and GMH dataset with paraphrasing. CONCLUSION: ClotCatcher, a novel deep learning model with data augmentation rapidly and accurately adjudicated the presence of VTE from radiology reports. Applying ClotCatcher to large databases would allow for rapid and accurate adjudication of incident VTE. This would reduce misclassification bias and form the foundation for future studies to estimate individual risk for patient to develop incident VTE.

A Novel Technology for Resolution of CEUS Imaging Problems in Patients With High BMI and Fatty Liver.

OBJECTIVES: In high-BMI patients with and without fatty liver, we evaluate performance of a commercially available specially designed ultrasound probe (SDP) for scanning at depth. Greyscale and contrast-enhanced ultrasound (CEUS) capability of SDP for parenchymal assessment and liver mass characterization, emphasizing HCC, is compared with standard curvilinear probes. METHODS: This retrospective study included 60 patients. Fifty-five with measured BMI included 46/55 (84%) overweight or obese, and 9/55(16%) in the normal range with severe fatty liver. Fifty-six patients with focal liver abnormality included 37 with a mass and 19 with post-ablative treatment site. Masses included 23 confirmed malignancies, 15 HCC, 4 ICC, and 4 metastases. SDP followed suboptimal ultrasound using a standard probe. Images with varying fat content were compared for depth of penetration on greyscale and ability of CEUS to diagnose tumors. RESULTS: SDP showed statistically significant improvement P = <.05 in CEUS penetration for all degrees of fatty liver (mild, moderate, and severe). In malignant tumors, SDP improved detection of lesion washout in the portal venous/late phase (PVP/LP) at depth >10 cm, and in all malignant masses (P < .05). Fifteen confirmed deep HCC showed arterial phase hyperenhancement on standard probe in 10/15 (67%) and 15/15 (100%) on SDP. PVP/LP washout on standard probe was shown in 4/15 (26%) and on SDP, 14/15, (93%). Therefore, 93% of LR-5 tumors were diagnosed with SDP. Removing necessity for biopsy. CONCLUSIONS: Metabolic syndrome and obesity challenge ultrasound, especially CEUS. SDP overcame limitations of standard probes for CEUS penetration especially in fatty liver. SDP was optimal for the liver mass characterization by detecting washout.

Fast-charging aluminium-chalcogen batteries resistant to dendritic shorting.

Although batteries fitted with a metal negative electrode are attractive for their higher energy density and lower complexity, the latter making them more easily recyclable, the threat of cell shorting by dendrites has stalled deployment of the technology1,2. Here we disclose a bidirectional, rapidly charging aluminium-chalcogen battery operating with a molten-salt electrolyte composed of NaCl-KCl-AlCl3. Formulated with high levels of AlCl3, these chloroaluminate melts contain catenated AlnCl3n+1- species, for example, Al2Cl7-, Al3Cl10- and Al4Cl13-, which with their Al-Cl-Al linkages confer facile Al3+ desolvation kinetics resulting in high faradaic exchange currents, to form the foundation for high-rate charging of the battery. This chemistry is distinguished from other aluminium batteries in the choice of a positive elemental-chalcogen electrode as opposed to various low-capacity compound formulations3-6, and in the choice of a molten-salt electrolyte as opposed to room-temperature ionic liquids that induce high polarization7-12. We show that the multi-step conversion pathway between aluminium and chalcogen allows rapid charging at up to 200C, and the battery endures hundreds of cycles at very high charging rates without aluminium dendrite formation. Importantly for scalability, the cell-level cost of the aluminium-sulfur battery is projected to be less than one-sixth that of current lithium-ion technologies. Composed of earth-abundant elements that can be ethically sourced and operated at moderately elevated temperatures just above the boiling point of water, this chemistry has all the requisites of a low-cost, rechargeable, fire-resistant, recyclable battery.

Deep Learning Analyses of Brain MRI to Identify Sustained Attention Deficit in Treated Obstructive Sleep Apnea: A Pilot Study.

Purpose: Persistent sustained attention deficit (SAD) after continuous positive airway pressure (CPAP) treatment is a source of quality of life and occupational impairment in obstructive sleep apnea (OSA). However, persistent SAD is difficult to predict in patients initiated on CPAP treatment. We performed secondary analyses of brain magnetic resonance (MR) images in treated OSA participants, using deep learning, to predict SAD. Methods: 26 middle-aged men with CPAP use of more than 6 hours daily and MR imaging were included. SAD was defined by psychomotor vigilance task lapses of more than 2. 17 participants had SAD and 9 were without SAD. A Convolutional Neural Network (CNN) model was used for classifying the MR images into +SAD and -SAD categories. Results: The CNN model achieved an accuracy of 97.02±0.80% in classifying MR images into +SAD and -SAD categories. Assuming a threshold of 90% probability for the MR image being correctly classified, the model provided a participant-level accuracy of 99.11±0.55% and a stable image level accuracy of 97.45±0.63%. Conclusion: Deep learning methods, such as the proposed CNN model, can accurately predict persistent SAD based on MR images. Further replication of these findings will allow early initiation of adjunctive pharmacologic treatment in high-risk patients, along with CPAP, to improve quality of life and occupational fitness. Future augmentation of this approach with explainable artificial intelligence methods may elucidate the neuroanatomical areas underlying persistent SAD to provide mechanistic insights and novel therapeutic targets.

Integrative Network Modeling Highlights the Crucial Roles of Rho-GDI Signaling Pathway in the Progression of non-Small Cell Lung Cancer.

Non-small cell lung cancer (NSCLC) is the most prevalent form of lung cancer and a leading cause of cancer-related deaths worldwide. Using an integrative approach, we analyzed a publicly available merged NSCLC transcriptome dataset using machine learning, protein-protein interaction (PPI) networks and bayesian modeling to pinpoint key cellular factors and pathways likely to be involved with the onset and progression of NSCLC. First, we generated multiple prediction models using various machine learning classifiers to classify NSCLC and healthy cohorts. Our models achieved prediction accuracies ranging from 0.83 to 1.0, with XGBoost emerging as the best performer. Next, using functional enrichment analysis (and gene co-expression network analysis with WGCNA) of the machine learning feature-selected genes, we determined that genes involved in Rho GTPase signaling that modulate actin stability and cytoskeleton were likely to be crucial in NSCLC. We further assembled a PPI network for the feature-selected genes that was partitioned using Markov clustering to detect protein complexes functionally relevant to NSCLC. Finally, we modeled the perturbations in RhoGDI signaling using a bayesian network; our simulations suggest that aberrations in ARHGEF19 and/or RAC2 gene activities contributed to impaired MAPK signaling and disrupted actin and cytoskeleton organization and were arguably key contributors to the onset of tumorigenesis in NSCLC. We hypothesize that targeted measures to restore aberrant ARHGEF19 and/or RAC2 functions could conceivably rescue the cancerous phenotype in NSCLC. Our findings offer promising avenues for early predictive biomarker discovery, targeted therapeutic intervention and improved clinical outcomes in NSCLC.

Stochastic-HD: Leveraging Stochastic Computing on the Hyper-Dimensional Computing Pipeline.

Brain-inspired Hyper-dimensional(HD) computing is a novel and efficient computing paradigm. However, highly parallel architectures such as Processing-in-Memory(PIM) are bottle-necked by reduction operations required such as accumulation. To reduce this bottle-neck of HD computing in PIM, we present Stochastic-HD that combines the simplicity of operations in Stochastic Computing (SC) with the complex task solving capabilities of the latest HD computing algorithms. Stochastic-HD leverages deterministic SC, which enables all of HD operations to be done as highly parallel bitwise operations and removes all reduction operations, thus improving the throughput of PIM. To this end, we propose an in-memory hardware design for Stochastic-HD that exploits its high level of parallelism and robustness to approximation. Our hardware uses in-memory bitwise operations along with associative memory-like operations to enable a fast and energy-efficient implementation. With Stochastic-HD, we were able to reach a comparable accuracy with the Baseline-HD. Furthermore, by proposing an integrated Stochastic-HD retraining approach Stochastic-HD is able to reduce the accuracy loss to just 0.3%. We additionally accelerate the retraining process in our hardware design to create an end-to-end accelerator for Stochastic-HD. Finally, we also add support for HD Clustering to Stochastic-HD, which is the first to map the HD Clustering operations to the stochastic domain. As compared to the best PIM design for HD, Stochastic-HD is also 4.4% more accurate and 43.1× more energy-efficient.

Defect dynamics in two-dimensional black phosphorus under argon ion irradiation.

Fundamental understanding of the atomic-scale mechanisms underlying production, accumulation, and temporal evolution of defects in phosphorene during noble-gas ion irradiation is crucial to design efficient defect engineering routes to fabricate next-generation materials for energy technologies. Here, we employed classical molecular dynamics (CMD) simulations using a reactive force field to unravel the effect of defect dynamics on the structural changes in a monolayer of phosphorene induced by argon-ion irradiation, and its subsequent relaxation during post-radiation annealing treatment. Analysis of our CMD trajectories using unsupervised machine learning methods showed that radiation fluence strongly influences the types of defect that form, their dynamics, and their relaxation mechanisms during subsequent annealing. Low ion fluences yielded a largely crystalline sheet featuring isolated small voids (up to 2 nm), Stone-Wales defects, and mono-/di-vacancies; while large nanopores (∼10 nm) can form beyond a critical fluence of ∼1014 ions per cm2. During post-radiation annealing, we found two distinct relaxation mechanisms, depending on the fluence level. The isolated small voids (1-2 nm) formed at low ion-fluences heal via local re-arrangement of rings, which is facilitated by a cooperative mechanism involving a series of atomic motions that include thermal rippling, bond formation, bond rotation, angle bending and dihedral twisting. On the other hand, damaged structures obtained at high fluences exhibit pronounced coalescence of nanopores mediated by 3D networks of P-centered tetrahedra. These findings provide new perspectives to use ion beams to precisely control the concentration and distribution of specific defect types in phosphorene for emerging applications in electronics, batteries, sensing, and neuromorphic computing.

Long-Term Functional Outcomes of Endoscopic Decompression with Destandau Technique for Lumbar Canal Stenosis.

STUDY DESIGN: Retrospective study of patients with lumbar canal stenosis (LCS) operated using endoscopic unilateral laminotomy with bilateral decompression (ULBD). PURPOSE: This study aimed to provide a detailed description of the technique of endoscopic decompression in LCS along with a description of the surgical anatomy and its advantages. We also discuss the clinical outcomes in patients operated using this technique. OVERVIEW OF LITERATURE: In 1999, the results with the use of microscopic ULBD were published. Microscopic/microendoscopic decompression using tubular retractor system showed good to excellent results in studies that compared such techniques with midline decompression. The first description of the use of endoscope in spine surgery was in 1988 when it was used for discectomy. With advancements and familiarity with the techniques, full endoscopic surgery has found application in LCS treatment. METHODS: The clinical records of 953 patients who were operated between 1998 and 2008 were analyzed in 2018. Along with patient characteristics, information about return to daily activities, complication rates, and functional outcomes using Prolo score was assessed. RESULTS: L4-L5 was the most common level for which surgery was performed. Two-level decompression was performed in 116 patients; 89.5% patients were able to return to their daily activities after 2 weeks. Functional outcomes as per the Prolo score were reported by patients as excellent, good, and poor in 89.85%, 1.59%, and 8.55%, respectively. Repeat surgery was required at same level in 16 patients and at a different level in 21 patients. Total 605 patients (63.49%) were symptom-free during the 70-month followup, while 344 complained of residual back pain, and four complained of persistent leg pain. CONCLUSIONS: ULBD using the Endospine system achieves adequate decompression in most cases and is a good alternative to open laminectomy, with the advantage of avoiding damage to the structural integrity of the spine and preserving soft tissue attachments.

Sclerosing Epithelioid Fibrosarcoma of the Coccyx: A Case Report and Review of Literature.

Coccydynia in adult patients is not uncommon and is frequently neglected. Coccydynia is mostly associated with fall on buttocks. In long-standing cases, coccydynia can be debilitating. Rarely coccydynia can be due to more sinister causes and surgeons should be aware of all differential diagnosis. We present a case of an elderly female who presented with a complaint of pain over coccyx which was not subsiding with conventional treatment methods. Biopsy was done and a diagnosis of sclerosing epitheloid fibrosarcoma was made. We describe an unusual case of coccydynia secondary to this tumour with the histopathology finding and surgical management.