In Journal of neural engineering ; h5-index 52.0

OBJECTIVE : Understanding and differentiating brain states is an important task in the field of cognitive neuroscience with applications in health diagnostics (such as detecting neurotypical development vs. Autism Spectrum or coma/vegetative state vs. locked-in state). Electroencephalography (EEG) analysis is a particularly useful tool for this task as EEG data can detect millisecond-level changes in brain activity across a range of frequencies in a non-invasive and relatively inexpensive fashion. The goal of this study is to apply machine learning methods to EEG data in order to classify visual language comprehension across multiple participants.

APPROACH : 26-channel EEG was recorded for 24 Deaf participants while they watched videos of sign language sentences played in time-direct and time-reverse formats to simulate interpretable vs. uninterpretable sign language, respectively. Sparse Optimal Scoring (SOS) was applied to EEG data in order to classify which type of video a participant was watching, time-direct or time-reversed. The use of SOS also served to reduce the dimensionality of the features to improve model interpretability.

MAIN RESULTS : The analysis of frequency-domain EEG data resulted in an average out-of-sample classification accuracy of 98.89%, which was far superior to the time-domain analysis. This high classification accuracy suggests this model can accurately identify common neural responses to visual linguistic stimuli.

SIGNIFICANCE : The significance of this work is in determining necessary and sufficient neural features for classifying the high-level neural process of visual language comprehension across multiple participants.

Ford Linda Katherine Wood, Borneman Joshua, Krebs Julia, Malaia Evguenia, Ames Brendan

2021-Jan-13

Discriminant Analysis, EEG, Optimal Scoring, classification, sign language

In International journal of sports physiology and performance ; h5-index 49.0

PURPOSE : To evaluate the relative importance and predictive ability of salivary immunoglobulin A (s-IgA) measures with regards to upper respiratory illness (URI) in youth athletes.

METHODS : Over a 38-week period, 22 youth athletes (age = 16.8 [0.5] y) provided daily symptoms of URI and 15 fortnightly passive drool saliva samples, from which s-IgA concentration and secretion rate were measured. Kernel-smoothed bootstrapping generated a balanced data set with simulated data points. The random forest algorithm was used to evaluate the relative importance (RI) and predictive ability of s-IgA concentration and secretion rate with regards to URI symptoms present on the day of saliva sampling (URIday), within 2 weeks of sampling (URI2wk), and within 4 weeks of sampling (URI4wk).

RESULTS : The percentage deviation from average healthy s-IgA concentration was the most important feature for URIday (median RI 1.74, interquartile range 1.41-2.07). The average healthy s-IgA secretion rate was the most important feature for URI4wk (median RI 0.94, interquartile range 0.79-1.13). No feature was clearly more important than any other when URI symptoms were identified within 2 weeks of sampling. The values for median area under the curve were 0.68, 0.63, and 0.65 for URIday, URI2wk, and URI4wk, respectively.

CONCLUSIONS : The RI values suggest that the percentage deviation from average healthy s-IgA concentration may be used to evaluate the short-term risk of URI, while the average healthy s-IgA secretion rate may be used to evaluate the long-term risk. However, the results show that neither s-IgA concentration nor secretion rate can be used to accurately predict URI onset within a 4-week window in youth athletes.

Sawczuk Thomas, Jones Ben, Welch Mitchell, Beggs Clive, Scantlebury Sean, Till Kevin

2021-Jan-13

adolescent, immune function, machine learning, monitoring

In Computers in biology and medicine

Peng Gengyou, Dong Hua, Liang Tong, Li Ling, Liu Jun

2021-Jan-05

Acetic acid test, Automatic diagnosis, Cervical screening, Colposcopy image, Deep learning, Multimodal feature change

In Tuberculosis (Edinburgh, Scotland)

Qin Zhi Zhen, Naheyan Tasneem, Ruhwald Morten, Denkinger Claudia M, Gelaw Sifrash, Nash Madlen, Creswell Jacob, Kik Sandra Vivian

2021-Jan-04

Artificial intelligence, Chest X-ray, Computer automated detection, Deep learning, Diagnostic, Tuberculosis

In Journal of neuroscience methods

BACKGROUND : Mild traumatic brain injury (mTBI) strongly associates with chronic neurodegenerative impairments such as post-traumatic stress disorder (PTSD) and mild cognitive impairment. Early detection of concussive events would significantly enhance the understanding of head injuries and provide better guidance for urgent diagnoses and the best clinical practices for achieving full recovery.

NEW METHOD : A smart helmet was developed with a single embedded fiber Bragg grating (FBG) sensor for real-time sensing of blunt-force impact events to helmets. The transient signals provide both magnitude and directional information about the impact event, and the data can be used for training machine learning (ML) models.

RESULTS : The FBG-embedded smart helmet prototype successfully achieved real-time sensing of concussive events. Transient data "fingerprints" consisting of both magnitude and direction of impact, were found to correlate with types of blunt-force impactors. Trained ML models were able to accurately predict (R² ∼ 0.90) the magnitudes and directions of blunt-force impact events from data not used for model training.

COMPARISON WITH EXISTING METHODS : The combination of the smart helmet data with analyses using ML models provides accurate predictions of the types of impactors that caused the events, as well as the magnitudes and the directions of the impact forces, which are unavailable using existing devices.

CONCLUSION : This work resulted in an ML-assisted, FBG-embedded smart helmet for real-time identification of concussive events using a highly accurate multi-metric strategy. The use of ML-FBG smart helmet systems can serve as an early-stage intervention strategy during and immediately following a concussive event.

Zhuang Yiyang, Yang Qingbo, Han Taihao, O’Malley Ryan, Kumar Aditya, Gerald Rex E, Huang Jie

2021-Jan-10

Blunt-force impact-induced brain injury, Concussive events, Fiber Bragg grating, Fiber-optic sensor, Machine learning, Mild traumatic brain injury

In Chest ; h5-index 81.0

RATIONALE : Asthma exacerbations result in significant health and economic burden but is difficult to predict.

OBJECTIVES : We aim to develop machine learning (ML) models with large-scale outpatient data to predict asthma exacerbations.

METHODS : We analyzed data extracted from electronic health records (EHR) on asthmatics followed at the Cleveland Clinic from 2010 to 2018. Demographic information, comorbidities, lab values and asthma medications were included as covariates. Three different models were built with Logistic regression, random forest, and gradient boosting decision tree to predict: (1) non-severe asthma exacerbation requiring oral glucocorticoid burst, (2) emergency department (ED) visits and (3) hospitalizations.

MEASUREMENTS AND MAIN RESULTS : Out of 60,302 patients, 19,772 (32.8%) had at least one non-severe exacerbation requiring oral glucocorticoid burst, 1,748 (2.9%) requiring ED visit, and 902 (1.5%) requiring hospitalization. Non-severe exacerbation, ED visit and hospitalization were best predicted by Light Gradient Boosting Machine (LightGBM), an algorithm used in machine learning to fit predictive analytic models, and had an area under the receiver operator curve of 0.71 (95%CI: 0.70-0.72), 0.88 (95%CI 0.86-0.89) and 0.85 (95%CI: 0.82-0.88) respectively. Risk factors for all three outcomes include age, long acting beta agonist, high dose inhaled glucocorticoid or chronic oral glucocorticoid therapy. In subgroup analysis of 9,448 patients with spirometry data, low FEV1 and FEV/FVC ratio were identified as top risk factors for asthma exacerbation, ED visits and hospitalization. However, adding pulmonary function tests did not improve models' prediction performance.

CONCLUSIONS : Models built with ML algorithm from real-world outpatient EHR data accurately predict asthma exacerbation and can be incorporated into clinical decision tools to enhance outpatient care and prevent adverse outcomes.

Zein Joe G, Wu Chao-Ping, Attaway Amy H, Zhang Peng, Nazha Aziz

2021-Jan-10