Acoustic surveillance for respiratory diseases: a prospective analysis of cough trends using artificial intelligence.

Syndromic surveillance for respiratory disease is limited by an inability to monitor its protean manifestation, cough. Advances in artificial intelligence provide the ability to passively monitor cough at individual and community levels. We hypothesized that changes in the aggregate number of coughs recorded among a sample could serve as a lead indicator for population incidence of respiratory diseases, particularly that of COVID-19. We enrolled over 900 people from the city of Pamplona (Spain) between 2020 and 2021 and used artificial intelligence cough detection software to monitor their cough. We collected nine person-years of cough aggregated data. Coughs per hour surged around the time cohort subjects sought medical care. There was a weak temporal correlation between aggregated coughs and the incidence of COVID-19 in the local population. We propose that a clearer correlation with COVID-19 incidence could be achieved with better penetration and compliance with cough monitoring.

Pengembangan Software Aplikasi Pendeteksian Objek Berbasis LiDAR (Light Detection and Ranging) (Development of Lidar-Based Object Detection Software)

Mobil adalah sarana transportasi yang kebutuhannya semakin tinggi. Hal ini tidak saja terjadi di luar negeri tapi juga di Indonesia. Namun demikian, keberadaan mobil saat ini dikeluhkan karena polusi yang dihasilkan dan juga tingkat kenyamannya. Harapan di masa mendatang sepertinya lebih mengarah pada hadirnya mobil listrik dengan tingkat polusi sangat rendah, serta kenyamanan dalam penggunaannya, seperti halnya mobil listrik otonom. Di negara maju gagasan ini sudah mulai akan direalisasikan, dan Indonesia sepertinya juga akan menghadapi situasi dimana mobil tersebut menjadi masif digunakan. Oleh sebab itu, kita harus menguasai teknologi kendaraan listrik otonom agar kita dapat memasuki era Mobility in Society 5.0. Salah satu bentuk teknologi terkait adalah sistem software pendeteksian objek berbasis LiDAR. Adakalanya software yang menyertai suatu alat tidak dapat menyediakan fasilitas yang beragam sesuai dengan kebutuhan aplikasi di lapangan. Hal ini dikarenakan keterbatasan yang diberikan oleh produsen alat tersebut, begitu pula dengan produk LiDAR 2D yang banyak dipasaran, contohnya YDLiDAR. Untuk keperluan aplikasi pendeteksian objek, software yang disediakan memiliki keterbatasan dalam hal penyimpanan data, fleksibilitas penyajian data dan kemampuan mereduksi derau yang muncul saat LiDAR tersebut dioperasikan pada kondisi tertentu. Untuk mengatasi kekurangan tersebut diatas, maka pada penelitian ini dikembangkan software pendeteksian objek berbasi LiDAR yang menambahkan fungsi-fungsi tersebut di atas, serta dapat diaplikasikan untuk pendeteksian objek dan pengenalan jarak. Secara umum sistem ini memadukan sistem software yang dikembangkan pada laptop dengan sistem hardware yang terdiri dari YDLiDAR G4 dan interface data serial. Sistem software ini juga dikembangkan dengan menggunakan bahasa pemrograman python. Hasil pengukuran menunjukkan bahwa kinerja software yang dikembangkan memiliki performansi visual yang baik, dapat menyimpan data hasil deteksi dengan durasi yang dapat ditentukan, serta kemampuan dalam menekan derau yang cukup baik. Kemampuan mereduksi noise dari sistem software ini dapat mereduksi error hingga 19,2%.

Deep learning algorithm in detecting intracranial hemorrhages on emergency computed tomographies

Background Highly accurate detection of intracranial hemorrhages (ICH) on head computed tomography (HCT) scans can prove challenging at high-volume centers. This study aimed to determine the number of additional ICHs detected by an artificial intelligence (AI) algorithm and to evaluate reasons for erroneous results at a level I trauma center with teleradiology services. Methods In a retrospective multi-center cohort study, consecutive emergency non-contrast HCT scans were analyzed by a commercially available ICH detection software (AIDOC, Tel Aviv, Israel). Discrepancies between AI analysis and initial radiology report (RR) were reviewed by a blinded neuroradiologist to determine the number of additional ICHs detected and evaluate reasons leading to errors. Results 4946 HCT (05/2020-09/2020) from 18 hospitals were included in the analysis. 205 reports (4.1%) were classified as hemorrhages by both radiology report and AI. Out of a total of 162 (3.3%) discrepant reports, 62 were confirmed as hemorrhages by the reference neuroradiologist. 33 ICHs were identified exclusively via RRs. The AI algorithm detected an additional 29 instances of ICH, missed 12.4% of ICH and overcalled 1.9%; RRs missed 10.9% of ICHs and overcalled 0.2%. Many of the ICHs missed by the AI algorithm were located in the subarachnoid space (42.4%) and under the calvaria (48.5%). 85% of ICHs missed by RRs occurred outside of regular working-hours. Calcifications (39.3%), beam-hardening artifacts (18%), tumors (15.7%), and blood vessels (7.9%) were the most common reasons for AI overcalls. ICH size, image quality, and primary examiner experience were not found to be significantly associated with likelihood of incorrect AI results. Conclusion Complementing human expertise with AI resulted in a 12.2% increase in ICH detection. The AI algorithm overcalled 1.9% HCT. Trial registration German Clinical Trials Register (DRKS-ID: DRKS00023593).

Analysis of Pungency Sensation Effects from an Oral Processing, Sensorial and Emotions Detection Perspective—Case Study with Grilled Pork Meat

Pungency is an interesting sensory stimulus analyzed from different perspectives, in particular the underpinning mechanisms of its sensation and perception. In this study, grilled pork meat coated with three types of hot sauces were investigated regarding its main food oral processing characteristics and evaluated using time-intensity and temporal dominance of pungency sensations methods analyzing the pungency descriptors and intensities. Besides these methods, facial expressions obtained from video capturing were subject to emotion detection. Mastication parameters showed a slight, but not statistically significant, trend of an increased number of chews and consumption time associated with pungency intensity, while saliva incorporation indicated an increasing trend depending on the pungency intensity, especially after 25 strokes and before swallowing. Both time intensity and temporal dominance of pungency sensations showed that the complexity of understanding these sensations is in relation to intensity and type. Finally, the use of emotion detection software in analyzing the faces of panelists during mastication confirmed the increase in non-neutral emotions associated with the increase in pungency intensity.

Abstract 1122‐000066: Number‐Needed‐to‐Review: A Novel Metric to Assess Triage Efficiency of Large Vessel Occlusion Detection Systems

Introduction : Endovascular thrombectomy is the gold standard treatment for acute ischemic strokes with large vessel occlusions (LVO). Manual image analysis is often time consuming and requires clinicians to be skilled in reading perfusion scans, as well as vessel images. RapidAI software has an automated processor to detect LVO of the middle cerebral artery and is analyzed in this study. A novel metric, number‐needed‐to‐review (NNR), is introduced to assess the clinical efficiency of this software. Methods : This is a retrospective review of patients with a suspected ischemic stroke and an image processed by RapidAI software from 11/1/2020 to 4/30/2021 at a regional hospital system. Only M1 LVOs were included. Sensitivities, specificities, positive predictive value (PPV), and negative predictive value (NPV) were calculated for the following: Rapid LVO detection, gaze deviation (GD), hyperdense sign (HDS), Tmax >6 seconds, and NIHSS at presentation. The NNR was calculated for an M1 occlusion. Results : 559 patients were included in this study. M1 occlusion was detected in 42 (7.5%) cases. Rapid LVO detection software was found to have a sensitivity of 71%, specificity of 94%, PPV of 49%, and NPV of 92% for a M1 occlusion. When both GD and HDS were combined with Rapid LVO, the specificity and PPV increased to 100% for a M1 occlusion. A negative LVO software combined with either a low (<15 mL on Tmax >6s) or high (<50 mL on Tmax >6s) Tmax threshold were found to have a specificity and PPV of 100% for no M1 occlusion. The combination of GD, HDS, Rapid LVO+ (for M1 occlusion) and Rapid LVO‐ with a low Tmax threshold (for no M1 occlusion) yielded 24 images NNR per 100 cases. When the combination of GD, HDS, Rapid LVO+ was combined with Rapid LVO‐ and a high Tmax threshold, the NNR per 100 cases was 16. With the addition of NIHSS<7 for the remaining cases in the high Tmax group, the NNR per 100 cases decreased to 9. Conclusions : The addition of GD and HDS to the Rapid LVO increases the specificity and PPV for a M1 occlusion. When combined with a negative Rapid LVO detection and either a low or high Tmax >6s threshold, the NNR is significantly decreased. As few as 9 images per 100 would be needed to be manually reviewed by a clinician during stroke triage.

Evaluating the Efficacy of Acoustic Metrics for Understanding Baleen Whale Presence in the Western North Atlantic Ocean

Soundscape analyses provide an integrative approach to studying the presence and complexity of sounds within long-term acoustic data sets. Acoustic metrics (AMs) have been used extensively to describe terrestrial habitats but have had mixed success in the marine environment. Novel approaches are needed to be able to deal with the added noise and complexity of these underwater systems. Here we further develop a promising approach that applies AM with supervised machine learning to understanding the presence and species richness (SR) of baleen whales at two sites, on the shelf and the slope edge, in the western North Atlantic Ocean. SR at both sites was low with only rare instances of more than two species (out of six species acoustically detected at the shelf and five at the slope) vocally detected at any given time. Random forest classification models were trained on 1-min clips across both data sets. Model outputs had high accuracy (&gt;0.85) for detecting all species’ absence in both sites and determining species presence for fin and humpback whales on the shelf site (&gt;0.80) and fin and right whales on the slope site (&gt;0.85). The metrics that contributed the most to species classification were those that summarized acoustic activity (intensity) and complexity in different frequency bands. Lastly, the trained model was run on a full 12 months of acoustic data from on the shelf site and compared with our standard acoustic detection software and manual verification outputs. Although the model performed poorly at the 1-min clip resolution for some species, it performed well compared to our standard detection software approaches when presence was evaluated at the daily level, suggesting that it does well at a coarser level (daily and monthly). The model provided a promising complement to current methodologies by demonstrating a good prediction of species absence in multiple habitats, species presence for certain species/habitat combinations, and provides higher resolution presence information for most species/habitat combinations compared to that of our standard detection software.

YOLO with adaptive frame control for real-time object detection applications

You only look once (YOLO) is being used as the most popular object detection software in many intelligent video applications due to its ease of use and high object detection precision. In addition, in recent years, various intelligent vision systems based on high-performance embedded systems are being developed. Nevertheless, the YOLO still requires high-end hardware for successful real-time object detection. In this paper, we first discuss real-time object detection service of the YOLO on AI embedded systems with resource constraints. In particular, we point out the problems related to real-time processing in YOLO object detection associated with network cameras, and then propose a novel YOLO architecture with adaptive frame control (AFC) that can efficiently cope with these problems. Through various experiments, we show that the proposed AFC can maintain the high precision and convenience of YOLO, and provide real-time object detection service by minimizing total service delay, which remains a limitation of the pure YOLO.