More Like this

1. A Deep Learning-Based Real-time Seizure Detection System

   https://doi.org/10.1109/SPMB50085.2020.9353623  

   Shawki, N. ; Elseify, T. ; Cap, T. ; Shah, V. ; Obeid, I. ; Picone, J. ( December 2020 , Proceedings of the IEEE Signal Processing in Medicine and Biology Symposium (SPMB))

   Obeid, Iyad Selesnick (Ed.)

   Electroencephalography (EEG) is a popular clinical monitoring tool used for diagnosing brain-related disorders such as epilepsy [1]. As monitoring EEGs in a critical-care setting is an expensive and tedious task, there is a great interest in developing real-time EEG monitoring tools to improve patient care quality and efficiency [2]. However, clinicians require automatic seizure detection tools that provide decisions with at least 75% sensitivity and less than 1 false alarm (FA) per 24 hours [3]. Some commercial tools recently claim to reach such performance levels, including the Olympic Brainz Monitor [4] and Persyst 14 [5]. In this abstract, we describe our efforts to transform a high-performance offline seizure detection system [3] into a low latency real-time or online seizure detection system. An overview of the system is shown in Figure 1. The main difference between an online versus offline system is that an online system should always be causal and has minimum latency which is often defined by domain experts. The offline system, shown in Figure 2, uses two phases of deep learning models with postprocessing [3]. The channel-based long short term memory (LSTM) model (Phase 1 or P1) processes linear frequency cepstral coefficients (LFCC) [6] features from each EEG channel separately. We use the hypotheses generated by the P1 model and create additional features that carry information about the detected events and their confidence. The P2 model uses these additional features and the LFCC features to learn the temporal and spatial aspects of the EEG signals using a hybrid convolutional neural network (CNN) and LSTM model. Finally, Phase 3 aggregates the results from both P1 and P2 before applying a final postprocessing step. The online system implements Phase 1 by taking advantage of the Linux piping mechanism, multithreading techniques, and multi-core processors. To convert Phase 1 into an online system, we divide the system into five major modules: signal preprocessor, feature extractor, event decoder, postprocessor, and visualizer. The system reads 0.1-second frames from each EEG channel and sends them to the feature extractor and the visualizer. The feature extractor generates LFCC features in real time from the streaming EEG signal. Next, the system computes seizure and background probabilities using a channel-based LSTM model and applies a postprocessor to aggregate the detected events across channels. The system then displays the EEG signal and the decisions simultaneously using a visualization module. The online system uses C++, Python, TensorFlow, and PyQtGraph in its implementation. The online system accepts streamed EEG data sampled at 250 Hz as input. The system begins processing the EEG signal by applying a TCP montage [8]. Depending on the type of the montage, the EEG signal can have either 22 or 20 channels. To enable the online operation, we send 0.1-second (25 samples) length frames from each channel of the streamed EEG signal to the feature extractor and the visualizer. Feature extraction is performed sequentially on each channel. The signal preprocessor writes the sample frames into two streams to facilitate these modules. In the first stream, the feature extractor receives the signals using stdin. In parallel, as a second stream, the visualizer shares a user-defined file with the signal preprocessor. This user-defined file holds raw signal information as a buffer for the visualizer. The signal preprocessor writes into the file while the visualizer reads from it. Reading and writing into the same file poses a challenge. The visualizer can start reading while the signal preprocessor is writing into it. To resolve this issue, we utilize a file locking mechanism in the signal preprocessor and visualizer. Each of the processes temporarily locks the file, performs its operation, releases the lock, and tries to obtain the lock after a waiting period. The file locking mechanism ensures that only one process can access the file by prohibiting other processes from reading or writing while one process is modifying the file [9]. The feature extractor uses circular buffers to save 0.3 seconds or 75 samples from each channel for extracting 0.2-second or 50-sample long center-aligned windows. The module generates 8 absolute LFCC features where the zeroth cepstral coefficient is replaced by a temporal domain energy term. For extracting the rest of the features, three pipelines are used. The differential energy feature is calculated in a 0.9-second absolute feature window with a frame size of 0.1 seconds. The difference between the maximum and minimum temporal energy terms is calculated in this range. Then, the first derivative or the delta features are calculated using another 0.9-second window. Finally, the second derivative or delta-delta features are calculated using a 0.3-second window [6]. The differential energy for the delta-delta features is not included. In total, we extract 26 features from the raw sample windows which add 1.1 seconds of delay to the system. We used the Temple University Hospital Seizure Database (TUSZ) v1.2.1 for developing the online system [10]. The statistics for this dataset are shown in Table 1. A channel-based LSTM model was trained using the features derived from the train set using the online feature extractor module. A window-based normalization technique was applied to those features. In the offline model, we scale features by normalizing using the maximum absolute value of a channel [11] before applying a sliding window approach. Since the online system has access to a limited amount of data, we normalize based on the observed window. The model uses the feature vectors with a frame size of 1 second and a window size of 7 seconds. We evaluated the model using the offline P1 postprocessor to determine the efficacy of the delayed features and the window-based normalization technique. As shown by the results of experiments 1 and 4 in Table 2, these changes give us a comparable performance to the offline model. The online event decoder module utilizes this trained model for computing probabilities for the seizure and background classes. These posteriors are then postprocessed to remove spurious detections. The online postprocessor receives and saves 8 seconds of class posteriors in a buffer for further processing. It applies multiple heuristic filters (e.g., probability threshold) to make an overall decision by combining events across the channels. These filters evaluate the average confidence, the duration of a seizure, and the channels where the seizures were observed. The postprocessor delivers the label and confidence to the visualizer. The visualizer starts to display the signal as soon as it gets access to the signal file, as shown in Figure 1 using the “Signal File” and “Visualizer” blocks. Once the visualizer receives the label and confidence for the latest epoch from the postprocessor, it overlays the decision and color codes that epoch. The visualizer uses red for seizure with the label SEIZ and green for the background class with the label BCKG. Once the streaming finishes, the system saves three files: a signal file in which the sample frames are saved in the order they were streamed, a time segmented event (TSE) file with the overall decisions and confidences, and a hypotheses (HYP) file that saves the label and confidence for each epoch. The user can plot the signal and decisions using the signal and HYP files with only the visualizer by enabling appropriate options. For comparing the performance of different stages of development, we used the test set of TUSZ v1.2.1 database. It contains 1015 EEG records of varying duration. The any-overlap performance [12] of the overall system shown in Figure 2 is 40.29% sensitivity with 5.77 FAs per 24 hours. For comparison, the previous state-of-the-art model developed on this database performed at 30.71% sensitivity with 6.77 FAs per 24 hours [3]. The individual performances of the deep learning phases are as follows: Phase 1’s (P1) performance is 39.46% sensitivity and 11.62 FAs per 24 hours, and Phase 2 detects seizures with 41.16% sensitivity and 11.69 FAs per 24 hours. We trained an LSTM model with the delayed features and the window-based normalization technique for developing the online system. Using the offline decoder and postprocessor, the model performed at 36.23% sensitivity with 9.52 FAs per 24 hours. The trained model was then evaluated with the online modules. The current performance of the overall online system is 45.80% sensitivity with 28.14 FAs per 24 hours. Table 2 summarizes the performances of these systems. The performance of the online system deviates from the offline P1 model because the online postprocessor fails to combine the events as the seizure probability fluctuates during an event. The modules in the online system add a total of 11.1 seconds of delay for processing each second of the data, as shown in Figure 3. In practice, we also count the time for loading the model and starting the visualizer block. When we consider these facts, the system consumes 15 seconds to display the first hypothesis. The system detects seizure onsets with an average latency of 15 seconds. Implementing an automatic seizure detection model in real time is not trivial. We used a variety of techniques such as the file locking mechanism, multithreading, circular buffers, real-time event decoding, and signal-decision plotting to realize the system. A video demonstrating the system is available at: https://www.isip.piconepress.com/projects/nsf_pfi_tt/resources/videos/realtime_eeg_analysis/v2.5.1/video_2.5.1.mp4. The final conference submission will include a more detailed analysis of the online performance of each module. ACKNOWLEDGMENTS Research reported in this publication was most recently supported by the National Science Foundation Partnership for Innovation award number IIP-1827565 and the Pennsylvania Commonwealth Universal Research Enhancement Program (PA CURE). Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the official views of any of these organizations. REFERENCES [1] A. Craik, Y. He, and J. L. Contreras-Vidal, “Deep learning for electroencephalogram (EEG) classification tasks: a review,” J. Neural Eng., vol. 16, no. 3, p. 031001, 2019. https://doi.org/10.1088/1741-2552/ab0ab5. [2] A. C. Bridi, T. Q. Louro, and R. C. L. Da Silva, “Clinical Alarms in intensive care: implications of alarm fatigue for the safety of patients,” Rev. Lat. Am. Enfermagem, vol. 22, no. 6, p. 1034, 2014. https://doi.org/10.1590/0104-1169.3488.2513. [3] M. Golmohammadi, V. Shah, I. Obeid, and J. Picone, “Deep Learning Approaches for Automatic Seizure Detection from Scalp Electroencephalograms,” in Signal Processing in Medicine and Biology: Emerging Trends in Research and Applications, 1st ed., I. Obeid, I. Selesnick, and J. Picone, Eds. New York, New York, USA: Springer, 2020, pp. 233–274. https://doi.org/10.1007/978-3-030-36844-9_8. [4] “CFM Olympic Brainz Monitor.” [Online]. Available: https://newborncare.natus.com/products-services/newborn-care-products/newborn-brain-injury/cfm-olympic-brainz-monitor. [Accessed: 17-Jul-2020]. [5] M. L. Scheuer, S. B. Wilson, A. Antony, G. Ghearing, A. Urban, and A. I. Bagic, “Seizure Detection: Interreader Agreement and Detection Algorithm Assessments Using a Large Dataset,” J. Clin. Neurophysiol., 2020. https://doi.org/10.1097/WNP.0000000000000709. [6] A. Harati, M. Golmohammadi, S. Lopez, I. Obeid, and J. Picone, “Improved EEG Event Classification Using Differential Energy,” in Proceedings of the IEEE Signal Processing in Medicine and Biology Symposium, 2015, pp. 1–4. https://doi.org/10.1109/SPMB.2015.7405421. [7] V. Shah, C. Campbell, I. Obeid, and J. Picone, “Improved Spatio-Temporal Modeling in Automated Seizure Detection using Channel-Dependent Posteriors,” Neurocomputing, 2021. [8] W. Tatum, A. Husain, S. Benbadis, and P. Kaplan, Handbook of EEG Interpretation. New York City, New York, USA: Demos Medical Publishing, 2007. [9] D. P. Bovet and C. Marco, Understanding the Linux Kernel, 3rd ed. O’Reilly Media, Inc., 2005. https://www.oreilly.com/library/view/understanding-the-linux/0596005652/. [10] V. Shah et al., “The Temple University Hospital Seizure Detection Corpus,” Front. Neuroinform., vol. 12, pp. 1–6, 2018. https://doi.org/10.3389/fninf.2018.00083. [11] F. Pedregosa et al., “Scikit-learn: Machine Learning in Python,” J. Mach. Learn. Res., vol. 12, pp. 2825–2830, 2011. https://dl.acm.org/doi/10.5555/1953048.2078195. [12] J. Gotman, D. Flanagan, J. Zhang, and B. Rosenblatt, “Automatic seizure detection in the newborn: Methods and initial evaluation,” Electroencephalogr. Clin. Neurophysiol., vol. 103, no. 3, pp. 356–362, 1997. https://doi.org/10.1016/S0013-4694(97)00003-9. 

   more » « less
2. Soil health through farmers’ eyes: Toward a better understanding of how farmers view, value, and manage for healthier soils

   https://doi.org/10.2489/jswc.2023.00058  

   Irvine, R. ; Houser, M. ; Marquart-Pyatt, S.T. ; Bogar, G. ; Bolin, L.G. ; Browning, E. Grennan ; Evans, S.E. ; Howard, M.M. ; Lau, J.A. ; Lennon, J.T. ( January 2023 , Journal of Soil and Water Conservation)

   Abstract: Improved soil health (SH) is critical in achieving agricultural resilience and miti- gating climate risks. Whether SH management practices are widely used depends greatly on US farmers’ voluntary decision-making. Toward understanding this point, much research has addressed factors that contribute to the adoption (or lack thereof) of SH-promoting practices, but less is known in terms of farmers’ perceptions of SH itself and the corresponding man- agement practices they see as related to achieving SH. To offer introductory insight on this knowledge gap and support better buy-in from farmers toward positive SH outcomes, our research draws upon qualitative interviews with 91 farmers across three key agricultural states in the Midwest (Illinois, Indiana, and Michigan). We develop a more detailed understanding of farmers’ views on SH, and why and how they manage for it. Nearly all interviewed farmers were familiar with the concept of SH and most viewed it favorably. A minority of farmers lacked familiarity with the term “SH” yet still managed for it. Skeptics of SH largely cited uncertainties related to over-zealous messaging by proponents of SH or lack of evidence for the return on investment of SH practices. Overall, farmers’ perceptions of SH largely aligned with the scientific community’s understanding of soils being a dynamic system, though farmers most dominantly defined SH by its biological component. Farmers perceived a host of benefits of SH, most often noting benefits to production, followed by improvements in physical aspects of the soil such as erosion control and increased organic matter. Notably, pro- duction and sustainability benefits were often cited together, suggesting that SH management is increasingly seen as a “win-win” by farmers. Additionally, we found that many farmers view themselves as active participants in SH outcomes and believe their management choices are indicators of positive SH outcomes, regardless of the practices they employ, including some strategies (such as tillage or tile drainage) that do not align with scientifically documented approaches to improving SH. Our findings show that farmers report engaging in an array of SH management practices that target both biotic and abiotic components of soils, and often use multiple practices in tandem to promote SH on their farms. Achieving better SH in agricultural production in the future will require engaging farmers in SH management by tailoring outreach and communication strategies to align with the perspectives and language farmers themselves use to conceptualize SH. 

   more » « less
3. Maximizing the consistency between regional and global reference frames utilizing inheritance of seasonal displacement parameters

   https://doi.org/10.1007/s00190-022-01594-0  

   Gómez, Demián D. ; Bevis, Michael G. ; Caccamise, Dana J. ( February 2022 , Journal of Geodesy)

   Abstract Although modern global geometric reference frames (GRFs) such as the International Terrestrial Reference Frame (ITRF) can be used anywhere on Earth, regional reference frames (RRFs) are still used to densify geodetic control and optimize solutions for continental-scale areas and national purposes. Such RRFs can be formed by densifying the ITRF, utilizing GPS / GNSS stations common to both the ITRF and the RRF. It is possible to attach a RRF to a GRF by ensuring that some or all of the coefficients of the trajectory models in the RRF are ‘inherited’ from the trajectory models that define the GRF. This can be done on an epoch-by-epoch basis, or (our preference) via transformations that operate simultaneously in space and time. This paper documents inconsistencies in the densification of ITRF that arise when the common stations’ trajectory models ignore periodic displacements. This results in periodic coordinate biases in the RRF. We describe a generalized procedure to minimize this inconsistency when realizing any RRF aligned to the ITRF or any other ‘primary’ frame. We show the method used to realize the Argentine national frame Posiciones Geodésicas Argentinas (POSGAR) and discuss our results. Discrepancies in the periodic motion amplitudes in the vertical were reduced from 4 mm to less than 1 mm for multiple stations after applying our technique. We also propose adopting object-oriented programming terminology to describe the relationship between different reference frames, such as a regional and a global frame. This terminology assists in describing and understanding the hierarchy in geodetic reference frames. 

   more » « less
4. Responding to incorrect ideas: science graduate teaching assistants’ operationalization of error framing and undergraduate students’ perception

   https://doi.org/10.1186/s40594-023-00398-8  

   Wan, Tong ; Doty, Constance M. ; Geraets, Ashley A. ; Saitta, Erin K. H. ; Chini, Jacquelyn J. ( January 2023 , International Journal of STEM Education)

   In college science laboratory and discussion sections, student-centered active learning strategies have been implemented to improve student learning outcomes and experiences. Research has shown that active learning activities can increase student anxiety if students fear that they could be negatively evaluated by their peers. Error framing (i.e., to frame errors as natural and beneficial to learning) is proposed in the literature as a pedagogical tool to reduce student anxiety. However, little research empirically explores how an instructor can operationalize error framing and how error framing is perceived by undergraduate students. To bridge the gap in the literature, we conducted a two-stage study that involved science graduate teaching assistants (GTAs) and undergraduate students. In stage one, we introduced cold calling (i.e., calling on non-volunteering students) and error framing to 12 chemistry and 11 physics GTAs. Cold calling can increase student participation but may increase student anxiety. Error framing has the potential to mitigate student anxiety when paired with cold calling. GTAs were then tasked to rehearse cold calling paired with error framing in a mixed-reality classroom simulator. We identified GTA statements that aligned with the definition of error framing. In stage two, we selected a few example GTA error framing statements and interviewed 13 undergraduate students about their perception of those statements.

   In the simulator, all the GTAs rehearsed cold calling multiple times while only a few GTAs made error framing statements. A thematic analysis of GTAs’ error framing statements identified ways of error indication (i.e., explicit and implicit) and framing (i.e., natural, beneficial, and positive acknowledgement). Undergraduate student interviews revealed specific framing and tone that are perceived as increasing or decreasing student comfort in participating in classroom discourse. Both undergraduate students and some GTAs expressed negative opinions toward responses that explicitly indicate student mistakes. Undergraduate students’ perspectives also suggest that error framing should be implemented differently depending on whether errors have already occurred.

   Error framing is challenging for science GTAs to implement. GTAs’ operationalizations of error framing in the simulator and undergraduate students’ perceptions contribute to defining and operationalizing error framing for instructional practice. To increase undergraduate student comfort in science classroom discourse, GTAs can use implicit error indication. In response to students’ incorrect answers, GTAs can positively frame students’ specific ideas rather than discussing broadly how errors are natural or beneficial.

    

   more » « less
5. Assortment Optimization and Pricing Under the Multinomial Logit Model with Impatient Customers: Sequential Recommendation and Selection

   https://doi.org/10.1287/opre.2021.2127  

   Gao, Pin ; Ma, Yuhang ; Chen, Ningyuan ; Gallego, Guillermo ; Li, Anran ; Rusmevichientong, Paat ; Topaloglu, Huseyin ( September 2021 , Operations Research)

   We develop a variant of the multinomial logit model with impatient customers and study assortment optimization and pricing problems under this choice model. In our choice model, a customer incrementally views the assortment of available products in multiple stages. The patience level of a customer determines the maximum number of stages in which the customer is willing to view the assortments of products. In each stage, if the product with the largest utility provides larger utility than a minimum acceptable utility, which we refer to as the utility of the outside option, then the customer purchases that product right away. Otherwise, the customer views the assortment of products in the next stage as long as the customer’s patience level allows the customer to do so. Under the assumption that the utilities have the Gumbel distribution and are independent, we give a closed-form expression for the choice probabilities. For the assortment-optimization problem, we develop a polynomial-time algorithm to find the revenue-maximizing sequence of assortments to offer. For the pricing problem, we show that, if the sequence of offered assortments is fixed, then we can solve a convex program to find the revenue-maximizing prices, with which the decision variables are the probabilities that a customer reaches different stages. We build on this result to give a 0.878-approximation algorithm when both the sequence of assortments and the prices are decision variables. We consider the assortment-optimization problem when each product occupies some space and there is a constraint on the total space consumption of the offered products. We give a fully polynomial-time approximation scheme for this constrained problem. We use a data set from Expedia to demonstrate that incorporating patience levels, as in our model, can improve purchase predictions. We also check the practical performance of our approximation schemes in terms of both the quality of solutions and the computation times. 

   more » « less