Receive a weekly summary and discussion of the top papers of the week by leading researchers in the field.

Radiology Radiology

Applications of artificial intelligence in clinical management, research and health administration: imaging perspectives with a focus on hemophilia.

In Expert review of hematology

INTRODUCTION : Joints of persons with hemophilia are frequently affected by repetitive hemarthrosis. In this paper concepts, perks and quirks of the use of artificial intelligence (AI), machine learning (ML) and deep learning are reviewed within clinical and research contexts of hemophilia and other blood-induced disorders' patient care, targeted to the imaging diagnosis of hemophilic joints, under the perspective of different stakeholders (radiologists, hematologists, nurses, physiotherapists, technologists, researchers, managers and patients/caregivers).

AREAS COVERED : Rubrics that determine the suitability of the utilization of AI in blood-induced disorders' patient care, including diagnosis and follow-up of patients are discussed, focusing on features in which AI can replace or augment the role of radiology in the clinical management and in research of patients. Insights on features in the design and conduct of AI projects in which the human intervention remains critical are provided.

EXPERT OPINION : The author discusses research concepts in radiogenomics, and challenges of the utilization of AI in different healthcare fields such as patient safety, data sharing and privacy regulations, workforce education and future jobs' shortage. Finally, the author proposes alternatives and potential solutions to mitigate challenges in successfully deploying ML algorithms into clinical practice.

Doria Andrea S

2023-Mar-20

Artificial Intelligence, arthropathy, augmentation, hemophilia, imaging, joints, machine learning, radiogenomics, regulations

Surgery Surgery

Quantifying and Improving the Performance of Speech Recognition Systems on Dysphonic Speech.

In Otolaryngology--head and neck surgery : official journal of American Academy of Otolaryngology-Head and Neck Surgery

OBJECTIVE : This study seeks to quantify how current speech recognition systems perform on dysphonic input and if they can be improved.

STUDY DESIGN : Experimental machine learning methods based on a retrospective database.

SETTING : Single academic voice center.

METHODS : A database of dysphonic speech recordings was created and tested against 3 speech recognition platforms. Platform performance on dysphonic voice input was compared to platform performance on normal voice input. A custom speech recognition model was trained on voice from patients with spasmodic dysphonia or vocal cord paralysis. Custom model performance was compared to base model performance.

RESULTS : All platforms performed well on normal voice, and 2 platforms performed significantly worse on dysphonic speech. Accuracy metrics on dysphonic speech returned values of 84.55%, 88.57%, and 93.56% for International Business Machines (IBM) Watson, Amazon Transcribe, and Microsoft Azure, respectively. The secondary analysis demonstrated that the lower performance of IBM Watson and Amazon Transcribe was driven by performance on spasmodic dysphonia and vocal fold paralysis. Thus, a custom model was built to increase the accuracy of these pathologies on the Microsoft platform. Overall, the performance of the custom model on dysphonic voices was 96.43% and on normal voices was 97.62%.

CONCLUSION : Current speech recognition systems generally perform worse on dysphonic speech than on normal speech. We theorize that poor performance is a consequence of a lack of dysphonic voices in each platform's original training dataset. We address this limitation with transfer learning used to increase the performance of these systems on all dysphonic speech.

Hidalgo Lopez Julio C, Sandeep Shelly, Wright MaKayla, Wandell Grace M, Law Anthony B

2023-Jan-24

artificial intelligence, dysphonia, laryngology, transfer learning, voice

General General

Intelligent Cubic-Designed Piezoelectric Node (iCUPE) with Simultaneous Sensing and Energy Harvesting Ability toward Self-Sustained Artificial Intelligence of Things (AIoT).

In ACS nano ; h5-index 203.0

The evolution of artificial intelligence of things (AIoT) drastically facilitates the development of a smart city via comprehensive perception and seamless communication. As a foundation, various AIoT nodes are experiencing low integration and poor sustainability issues. Herein, a cubic-designed intelligent piezoelectric AIoT node iCUPE is presented, which integrates a high-performance energy harvesting and self-powered sensing module via a micromachined lead zirconate titanate (PZT) thick-film-based high-frequency (HF)-piezoelectric generator (PEG) and poly(vinylidene fluoride-co-trifluoroethylene) (P(VDF-TrFE)) nanofiber thin-film-based low-frequency (LF)-PEGs, respectively. The LF-PEG and HF-PEG with specific frequency up-conversion (FUC) mechanism ensures continuous power supply over a wide range of 10-46 Hz, with a record high power density of 17 mW/cm3 at 1 g acceleration. The cubic design allows for orthogonal placement of the three FUC-PEGs to ensure a wide range of response to vibrational energy sources from different directions. The self-powered triaxial piezoelectric sensor (TPS) combined with machine learning (ML) assisted three orthogonal piezoelectric sensing units by using three LF-PEGs to achieve high-precision multifunctional vibration recognition with resolutions of 0.01 g, 0.01 Hz, and 2° for acceleration, frequency, and tilting angle, respectively, providing a high recognition accuracy of 98%-100%. This work proves the feasibility of developing a ML-based intelligent sensor for accelerometer and gyroscope functions at resonant frequencies. The proposed sustainable iCUPE is highly scalable to explore multifunctional sensing and energy harvesting capabilities under diverse environments, which is essential for AIoT implementation.

Huang Manjuan, Zhu Minglu, Feng Xiaowei, Zhang Zixuan, Tang Tianyi, Guo Xinge, Chen Tao, Liu Huicong, Sun Lining, Lee Chengkuo

2023-Mar-20

artificial intelligence of things (AIoT), machine learning, piezoelectric generator, self-powered sensor, status monitoring

Ophthalmology Ophthalmology

Analysis of risk and protective factors associated with retinal nerve fiber layer defect in a Chinese adult population.

In International journal of ophthalmology

AIM : To investigate the risk and protective factors associated with the retinal nerve fiber layer defect (RNFLD) in a Chinese adult population.

METHODS : This study was a cross-sectional population-based investigation including employees and retirees of a coal mining company in Kailuan City, Hebei Province. All the study participants underwent a comprehensive systemic and ophthalmic examination. RNFLD was diagnosed on fundus photographs. Binary logistic regression was used to investigate the risk and protective factors associated with the RNFLD.

RESULTS : The community-based study included 14 440 participants. There were 10 473 participants in our study, including 7120 males (68.0%) and 3353 females (32.0%). The age range was 45-108y, averaging 59.56±8.66y. Totally 568 participants had RNFLD and the prevalence rate was 5.42%. A higher prevalence of RNFLD was associated with older age [P<0.001, odds ratio (OR): 1.032; 95% confidence interval (CI): 1.018-1.046], longer axial length (P=0.010, OR: 1.190; 95%CI: 1.042-1.359), hypertension (P=0.007, OR: 0.639; 95%CI: 0.460-0.887), and diabetes mellitus (P=0.019, OR: 0.684; 95%CI: 0.499-0.939). The protective factors of RNFLD were visual acuity (P=0.038, OR: 0.617; 95%CI: 0.391-0.975), and central anterior chamber depth (P=0.046, OR: 0.595; 95%CI: 0.358-0.990).

CONCLUSION : In our cross-sectional community-based study, with an age range of 45-108y, RNFLD is associated with older age, longer axial length, hypertension, and diabetes mellitus. The protective factors of RNFLD are visual acuity and central anterior chamber depth. These can help to predict and evaluate RNFLD related diseases and identify high-risk populations early.

Wang Ye-Nan, Wang Ya-Xing, Zhou Jin-Qiong, Wan Qian-Qian, Fang Li-Jian, Wang Hai-Wei, Yang Jing-Yan, Dong Li, Wang Jin-Yuan, Yang Xuan, Yan Yan-Ni, Wang Qian, Wu Shou-Ling, Chen Shuo-Hua, Zhu Jing-Yuan, Wei Wen-Bin, Jonas Jost B

2023

age, axial length, central anterior chamber depth, diabetes mellitus, hypertension, retinal nerve fiber layer, retinal nerve fiber layer defect, visual acuity

Surgery Surgery

Objective Pharyngeal Phenotyping in Obstructive Sleep Apnea With High-Resolution Manometry.

In Otolaryngology--head and neck surgery : official journal of American Academy of Otolaryngology-Head and Neck Surgery

OBJECTIVE : Drug-induced sleep endoscopy (DISE) is a commonly used diagnostic tool for surgical procedural selection in obstructive sleep apnea (OSA), but it is expensive, subjective, and requires sedation. Here we present an initial investigation of high-resolution pharyngeal manometry (HRM) for upper airway phenotyping in OSA, developing a software system that reliably predicts pharyngeal sites of collapse based solely on manometric recordings.

STUDY DESIGN : Prospective cross-sectional study.

SETTING : An academic sleep medicine and surgery practice.

METHODS : Forty participants underwent simultaneous HRM and DISE. A machine learning algorithm was constructed to estimate pharyngeal level-specific severity of collapse, as determined by an expert DISE reviewer. The primary outcome metrics for each level were model accuracy and F1-score, which balances model precision against recall.

RESULTS : During model training, the average F1-score across all categories was 0.86, with an average weighted accuracy of 0.91. Using a holdout test set of 9 participants, a K-nearest neighbor model trained on 31 participants attained an average F1-score of 0.96 and an average accuracy of 0.97. The F1-score for prediction of complete concentric palatal collapse was 0.86.

CONCLUSION : Our findings suggest that HRM may enable objective and dynamic mapping of the pharynx, opening new pathways toward reliable and reproducible assessment of this complex anatomy in sleep.

Kent David T, Scott William C, Ye Cheng, Fabbri Daniel

2023-Jan-29

drug-induced sleep endoscopy, high-resolution manometry, obstructive sleep apnea, sleep surgery

General General

Movement Optimization for a Cyborg Cockroach in a Bounded Space Incorporating Machine Learning.

In Cyborg and bionic systems (Washington, D.C.)

Cockroaches can traverse unknown obstacle-terrain, self-right on the ground and climb above the obstacle. However, they have limited motion, such as less activity in light/bright areas and lower temperatures. Therefore, the movement of the cyborg cockroaches needs to be optimized for the utilization of the cockroach as a cyborg insect. This study aims to increase the search rate and distance traveled by cockroaches and reduce the stop time by utilizing automatic stimulation from machine learning. Multiple machine learning classifiers were applied to classify the offline binary classification of the cockroach movement based on the inertial measuring unit input signals. Ten time-domain features were chosen and applied as the classifier inputs. The highest performance of the classifiers was implemented for the online motion recognition and automatic stimulation provided to the cerci to trigger the free walking motion of the cockroach. A user interface was developed to run multiple computational processes simultaneously in real time such as computer vision, data acquisition, feature extraction, automatic stimulation, and machine learning using a multithreading algorithm. On the basis of the experiment results, we successfully demonstrated that the movement performance of cockroaches was importantly improved by applying machine learning classification and automatic stimulation. This system increased the search rate and traveled distance by 68% and 70%, respectively, while the stop time was reduced by 78%.

Ariyanto Mochammad, Refat Chowdhury Mohammad Masum, Hirao Kazuyoshi, Morishima Keisuke

2023