Acentech (United States)

Prognostic models in oncology have a profound impact on personalized cancer care and patient profiling, but tend to be heterogeneously developed and implemented in narrow patient cohorts. Here, we develop and benchmark multiple machine learning models to predict survival in pan-cancer and 16 single-cancer settings using a de-identified clinico-genomic database of 28,079 US patients with cancer. We identify key predictors of cancer prognosis, including 15 shared across seven or more cancer types, revealing strong consistency in cancer prognostic factors. We demonstrate that pan-cancer models generally outperform or match single-cancer models in predicting survival and risk stratifying patients, especially in smaller cancer cohorts, suggesting a unique transfer learning advantage of pan-cancer models. This work demonstrates the potential of pan-cancer approaches in enhancing the accuracy and applicability of prognostic models in oncology, paving the way for more personalized and effective cancer care strategies.

<bold>BACKGROUND:</bold> Chronic Obstructive Pulmonary Disease (COPD) is a global health concern, and accurate diagnosis is crucial for early intervention and improving patient quality of life, while also reducing the risk of exacerbations and healthcare costs. <bold>OBJECTIVE:</bold> To quantify diagnostic performance of N-Tidal Diagnose (NTD) machine learning models in a cohort of known COPD patients and healthy controls. <bold>METHODS:</bold> Capnography data was collected using 75 seconds of tidal breathing through TidalSense’s N-Tidal capnometer from a cross-sectional observational data set, which included 51 COPD subjects and 51 healthy controls. The performance of NTD 1 v1.0 software’s machine learning models, which perform diagnostic classification using 25 geometric capnography features, was measured. <bold>RESULTS:</bold> <fig><object-id>erj;66/suppl_69/PA1460/F1</object-id><object-id>F1</object-id><object-id>F1</object-id><caption> <italic>Figure 1.</italic> Confusion matrix evaluating the diagnostic performance of the NTD models. </caption><graphic></graphic></fig> The models were unable to diagnose eight patients who fell within the ‘indeterminate’ category with prediction probability close to 0.5. For all other participants, the NTD software yielded the following performance metrics: sensitivity 0.960, specificity 0.932, positive predictive value 0.953, and negative predictive value 0.941. <bold>CONCLUSIONS:</bold> The N-Tidal Diagnose ML models were able to accurately distinguish COPD subjects from healthy controls. This suggests that NTD could be used as a rapid and accessible alternative to spirometry for rapid point-of-care testing of COPD.

Today Underwater Radiated Noise (or URN) is not just about keeping submarines and combatant surface vessels undetectable. Since the mid 1990’s, vessel URN moved into the public domain for many purposes. Approaching 30 years as an open topic, much has been learned, many new standards have been issued, and many more are being developed. As an example, the standard for general acoustic terms, ANSI S1.1 was first published in 1960 with roots back to 1942. Whereas the standard for general underwater acoustic terms, ISO-18405 was just published in 2017, as much as 75 years later. Methodologies for measurement of URN have been codified by ANSI, ISO, and ship classification societies. Europe has written directives for URN limits; Canada is in the process of similar limits and U.S. environmentalists would like to follow. The International Maritime Organization (IMO, a UN subsidiary) has significant guidelines on vessel URN. This paper will highlight where and when URN became commercially relevant. The current metrics, methods, and other guidelines for measurement of ship source sound levels will be addressed. Finally, what will the future bring for shipboard URN limitations.

High frequency vibration transmission through steel coil springs due to surge and wave resonance effects is a concern when isolating sources of high frequency vibration, such as screw chillers. Air springs are sometimes used in these situations as an alternative to steel springs because they do not exhibit this high frequency transmission behavior. However, they must be well maintained and kept at the proper inflation. Low natural frequency, high static deflection elastomer isolators have recently been introduced to the market and these isolators provide an alternative to air springs while avoiding these maintenance needs. A case study on the use of these isolators for a screw chiller will be presented.

Abstract Neurofibromatosis type 1 (NF1) is a tumor predisposition syndrome caused by loss of function of the neurofibromin protein. Malignant peripheral nerve sheath tumors (MPNSTs) are a rare and deadly sarcoma with few therapeutic options that are the leading cause of death for patients with NF1. To date, no targeted therapies have been approved for MPNST treatment, highlighting the need for an understanding of adaptive signaling mechanisms that drive resistance. We developed a preclinical model of drug resistance using a cross-over drug holiday design and evaluated patterns of response and resistance to MEK and AKT inhibitors, two pathways that are dysregulated by loss of neurofibromin. We show that the mTOR and YAP/TEAD pathways are activated by MEK inhibitor exposure, yet blockade of these pathways in resistant MPNST PDX models does not significantly reduce tumor growth, despite strong in vitro synergy between trametinib and the novel TEAD inhibitor, GNE-7883. Using spatial transcriptomics, we uncovered phenotypic inertia as a key mechanism of drug resistance in MPNST, in addition to signaling plasticity. Further, we found that resistance is mediated by sustained ERK, YAP, and MYC driven transcriptional programs. In the future, preclinical studies should focus on addressing intratumoral heterogeneity and how it evolves over time.

Above-grade fitness centers can be an annoying source of vibration disturbance in buildings. Their potential impact is best evaluated at the design stage before people begin to use its treadmills, free weights, and exercise classes. Often the building does not yet exist and structural models provide the best way to evaluate potential impacts. The authors recently had the opportunity to compare finite element frequency response function (FRF) predictions to test data obtained in the field. The driving point accelerance predictions showed very good agreement with the test data up to 100 Hz. Cross-accelerance predictions from one point to another similarly showed good agreement with the test data. The test data also confirmed reciprocity when the drive and response points were interchanged. Test data in the field were obtained using both an instrumented force hammer and a drop weight with an accelerometer. Both methods produced consistent experimental FRFs. This study shows that a well-conceived finite element model can produce reasonable predictions of low-frequencystructural response.