scholarly journals Crowdsourced privacy-preserved feature tagging of short home videos for machine learning ASD detection

2021 ◽  
Vol 11 (1) ◽  
Author(s):  
Peter Washington ◽  
Qandeel Tariq ◽  
Emilie Leblanc ◽  
Brianna Chrisman ◽  
Kaitlyn Dunlap ◽  
...  
Keyword(s):  
Machine Learning ◽  
At Risk ◽  
High Performance ◽  
High Sensitivity ◽  
Autism Spectrum ◽  
Privacy Preserving ◽  
Machine Learning Classification ◽  
Home Videos ◽  
Video Feature ◽  
Pitch Shift

AbstractStandard medical diagnosis of mental health conditions requires licensed experts who are increasingly outnumbered by those at risk, limiting reach. We test the hypothesis that a trustworthy crowd of non-experts can efficiently annotate behavioral features needed for accurate machine learning detection of the common childhood developmental disorder Autism Spectrum Disorder (ASD) for children under 8 years old. We implement a novel process for identifying and certifying a trustworthy distributed workforce for video feature extraction, selecting a workforce of 102 workers from a pool of 1,107. Two previously validated ASD logistic regression classifiers, evaluated against parent-reported diagnoses, were used to assess the accuracy of the trusted crowd’s ratings of unstructured home videos. A representative balanced sample (N = 50 videos) of videos were evaluated with and without face box and pitch shift privacy alterations, with AUROC and AUPRC scores > 0.98. With both privacy-preserving modifications, sensitivity is preserved (96.0%) while maintaining specificity (80.0%) and accuracy (88.0%) at levels comparable to prior classification methods without alterations. We find that machine learning classification from features extracted by a certified nonexpert crowd achieves high performance for ASD detection from natural home videos of the child at risk and maintains high sensitivity when privacy-preserving mechanisms are applied. These results suggest that privacy-safeguarded crowdsourced analysis of short home videos can help enable rapid and mobile machine-learning detection of developmental delays in children.

scholarly journals Crowdsourced feature tagging for scalable and privacy-preserved autism diagnosis

2020 ◽  
Author(s):  
Peter Washington ◽  
Qandeel Tariq ◽  
Emilie Leblanc ◽  
Brianna Chrisman ◽  
Kaitlyn Dunlap ◽  
...  
Keyword(s):  
Machine Learning ◽  
Clinical Performance ◽  
Privacy Preserving ◽  
Machine Learning Classification ◽  
Home Videos ◽  
Video Feature ◽  
The Common ◽  
Mobile Assessment ◽  
Pitch Shift

ABSTRACT Standard medical diagnosis of mental health conditions often requires licensed experts who are increasingly outnumbered by those at risk, limiting reach. We test the hypothesis that a trustworthy crowd of non-experts can efficiently label features needed for accurate machine learning detection of the common childhood developmental disorder autism. We implement a novel process for creating a trustworthy distributed workforce for video feature extraction, selecting a workforce of 102 workers from a pool of 1,107. Two previously validated binary autism logistic regression classifiers were used to evaluate the quality of the curated crowd’s ratings on unstructured home videos. A clinically representative balanced sample (N=50 videos) of videos were evaluated with and without face box and pitch shift privacy alterations, with AUROC and AUPRC scores >0.98. With both privacy-preserving modifications, sensitivity is preserved (96.0%) while maintaining specificity (80.0%) and accuracy (88.0%) at levels that exceed classification methods without alterations. We find that machine learning classification from features extracted by a curated nonexpert crowd achieves clinical performance for pediatric autism videos and maintains acceptable performance when privacy-preserving mechanisms are applied. These results suggest that privacy-based crowdsourcing of short videos can be leveraged for rapid and mobile assessment of behavioral health.

scholarly journals Feature Extraction from MRI ADC Images for Brain Tumor Classification Using Machine Learning Techniques

2022 ◽  
Author(s):  
Sahan M. Vijithananda ◽  
Mohan L. Jayatilake ◽  
Badra Hewavithana ◽  
Teresa Gonçalves ◽  
Luis M. Rato ◽  
...  
Keyword(s):  
Machine Learning ◽  
Feature Selection ◽  
Brain Tumor ◽  
Brain Tumors ◽  
High Performance ◽  
Selection Process ◽  
Parameter Tuning ◽  
Machine Learning Techniques ◽  
F Test ◽  
Machine Learning Classification

Abstract Background: Diffusion-weighted (DW) imaging is a well-recognized magnetic resonance imaging (MRI) technique that is being routinely used in brain examinations in modern clinical radiology practices. This study focuses on extracting demographic and texture features from MRI Apparent Diffusion Coefficient (ADC) images of human brain tumors, identifying the distribution patterns of each feature and applying Machine Learning (ML) techniques to differentiate malignant from benign brain tumors.Methods: This prospective study was carried out using 1599 labeled MRI brain ADC image slices, 995 malignant, 604 benign from 195 patients who were radiologically diagnosed and histopathologically confirmed as brain tumor patients.The demographics, mean pixel values, skewness, kurtosis, features of Grey Level Co-occurrence Matrix (GLCM), mean, variance, energy, entropy, contrast, homogeneity, correlation, prominence and shade, were extracted from MRI ADC images of each patient.At the feature selection phase, the validity of the extracted features were measured using ANOVA f-test. Then, these features were used as input to several Machine Learning classification algorithms and the respective models were assessed.Results: According to the results of ANOVA f-test feature selection process, two attributes: skewness (3.34) and GLCM homogeneity (3.45) scored the lowest ANOVA f-test scores. Therefore both features were excluded in continuation of the experiment. From the different tested ML algorithms, the Random Forest classifier was chosen to build the final ML model since it presented the highest accuracy. The final model was able to predict malignant and benign neoplasms with an 90.41% accuracy after the hyper parameter tuning process.Conclusion: This study concludes that the above mentioned features (except skewness and GLCM homogeneity) are informative to identify and differentiate malignant from benign brain tumors. Moreover, they enable the development of a high-performance ML model that has the ability to assist in the decision-making steps of brain tumor diagnosis process, prior to attempting invasive diagnostic procedures such as brain biopsies.

The CBCL 1.5–5 and the identification of preschoolers with autism in Italy

2011 ◽  
Vol 20 (4) ◽  
pp. 329-338 ◽  
Author(s):  
F. Muratori ◽  
A. Narzisi ◽  
R. Tancredi ◽  
A. Cosenza ◽  
S. Calugi ◽  
...  
Keyword(s):  
At Risk ◽  
Logistic Regression ◽  
High Sensitivity ◽  
Autism Spectrum ◽  
Attention Problems ◽  
Control Group ◽  
Developmental Problems ◽  
Primary Care Settings ◽  
Lower Specificity ◽  
Almost All

Aims.To study the potential use of child behaviour checklist (CBCL) 1.5–5 scales for the early identification of preschoolers at risk of autism.Methods.CBCL scores of three groups of preschoolers were compared: (1) an experimental group of 101 preschoolers with autism spectrum disorder (ASD); (2) a control group of 95 preschoolers with other psychiatric disorders (OPD); (3) a control group of 117 preschoolers with typical development (TD). One-way analysis of variance (ANOVA), logistic regression with odds ratio (OR) and receiver operating characteristic (ROC) analyses were performed.Results.ANOVA revealed that ASD and OPD had significantly higher scores in almost all CBCL scales than TD. ASD presented significantly higher scores than OPD on Withdrawn, Attention Problems and Pervasive Developmental Problems (PDP) scales. Logistic regression analysis demonstrated that these same CBCL scales have validity in predicting the presence of an ASD towards both TD and OPD. ROC analysis indicated high sensitivity and specificity for PDP (0.85 and 0.90) and Withdrawn (0.89 and 0.92) scales when ASD is compared to TD. Specificity (0.60 for PDP and 0.65 for Withdrawn) decreases when comparing ASD and OPDConclusions.The PDP and Withdrawn scales have a good predictive validity so that they could be proposed as a first-level tool to identify preschoolers at risk of autism in primary care settings. Problems regarding the lower specificity when comparing ASDv. OPD are discussed.

Predictors of Turnover Intention in U.S. Federal Government Workforce: Machine Learning Evidence That Perceived Comprehensive HR Practices Predict Turnover Intention

2020 ◽  
pp. 009102602097756
Author(s):  
In Gu Kang ◽  
Ben Croft ◽  
Barbara A. Bichelmeyer
Keyword(s):  
Machine Learning ◽  
Job Satisfaction ◽  
At Risk ◽  
Turnover Intention ◽  
Regression Tree ◽  
Classification And Regression Tree ◽  
Federal Employees ◽  
Organizational Policies ◽  
Machine Learning Classification ◽  
Hr Practices

This study aims to identify important predictors of turnover intention and to characterize subgroups of U.S. federal employees at high risk for turnover intention. Data were drawn from the 2018 Federal Employee Viewpoint Survey (FEVS, unweighted N = 598,003), a nationally representative sample of U.S. federal employees. Machine learning Classification and Regression Tree (CART) analyses were conducted to predict turnover intention and accounted for sample weights. CART analyses identified six at-risk subgroups. Predictor importance scores showed job satisfaction was the strongest predictor of turnover intention, followed by satisfaction with organization, loyalty, accomplishment, involvement in decisions, likeness to job, satisfaction with promotion opportunities, skill development opportunities, organizational tenure, and pay satisfaction. Consequently, Human Resource (HR) departments should seek to implement comprehensive HR practices to enhance employees’ perceptions on job satisfaction, workplace environments and systems, and favorable organizational policies and supports and make tailored interventions for the at-risk subgroups.

scholarly journals Diabetes Prediction Using Machine Learning Techniques

2021 ◽  
pp. 150-152
Author(s):  
Seyma Kiziltas Koc ◽  
Mustafa Yeniad
Keyword(s):  
Machine Learning ◽  
Support Vector Machine ◽  
High Performance ◽  
Nearest Neighbor ◽  
Classification Performance ◽  
Machine Learning Techniques ◽  
Support Vector ◽  
Classification Algorithms ◽  
K Nearest Neighbor ◽  
Machine Learning Classification

Technologies which are used in the healthcare industry are changing rapidly because the technology is evolving to improve people's lifestyles constantly. For instance, different technological devices are used for the diagnosis and treatment of diseases. It has been revealed that diagnosis of disease can be made by computer systems with developing technology.Machine learning algorithms are frequently used tools because of their high performance in the field of health as well as many field. The aim of this study is to investigate different machine learning classification algorithms that can be used in the diagnosis of diabetes and to make comparative analyzes according to the metrics in the literature. In the study, seven classification algorithms were used in the literature. These algorithms are Logistic Regression, K-Nearest Neighbor, Multilayer Perceptron, Random Forest, Decision Trees, Support Vector Machine and Naive Bayes. Firstly, classification performance of algorithms are compared. These comparisons are based on accuracy, sensitivity, precision, and F1-score. The results obtained showed that support vector machine algorithm had the highest accuracy with 78.65%.

scholarly journals Predicting Breast Cancer Classification Using Various Machine Learning Classification Algorithm

2020 ◽  
Vol 6 (12) ◽  
pp. 282-285
Author(s):  
Nishant Bansal Nidhi Sengar and Amita Goe
Keyword(s):  
Breast Cancer ◽  
Machine Learning ◽  
Cancer Diagnosis ◽  
High Performance ◽  
Breast Cancer Diagnosis ◽  
Second Primary ◽  
Machine Learning Classification ◽  
University Of Wisconsin ◽  
Sensitivity Specificity ◽  
The University

Cancer diagnosis is one among the foremost studied problems within the medical domain. Several researchers have focused so as to enhance performance and achieve to get satisfactory results. Breast cancer[1] represents the second primary explanation for cancer deaths in women today and has become the foremost common cancer among women both within the developed and therefore the developing world in the last years. Breast cancer diagnosis is used to categorize the patients among benign (lacks ability to invade neighbouring tissue) from malignant (ability to invade neighbouring tissue) categories. In this study, the diagnosis of breast cancer from mammograms is complemented by using various classification techniques. In artificial intelligence, machine learning is a discipline which allows to the machine to evolve through a process. Machine learning[2] is widely utilized in bio-informatics and particularly in carcinoma diagnosis. This paper explores the various data processing approaches using Classification which may be applied on carcinoma data to create deep predictions. Besides this, this study predicts the simplest Model yielding high performance by evaluating dataset on various classifiers.[4-8] The results that are obtained through the research are assessed on various parameters like Accuracy, RMSE Error, Sensitivity, Specificity etc. Our work is going to be performed on the WBCD database (Wisconsin carcinoma Database) [12]obtained by the university of Wisconsin Hospital.

Abstract 15783: A Simplified Machine Learning Algorithm for Identification of Patients At-risk for Wild-type Transthyretin Amyloid Cardiomyopathy

Circulation ◽  
2020 ◽  
Vol 142 (Suppl_3) ◽  
Author(s):  
Stephen B Heitner ◽  
Ahmad Masri ◽  
Miriam R. Elman ◽  
Birol Emir ◽  
Kim D. Nolen ◽  
...  
Keyword(s):  
Machine Learning ◽  
At Risk ◽  
Learning Algorithm ◽  
High Sensitivity ◽  
Recursive Feature Elimination ◽  
Wild Type ◽  
Snomed Ct ◽  
Retrospective Case ◽  
Referral Center ◽  
Amyloid Cardiomyopathy

Introduction: Wild-type transthyretin amyloid cardiomyopathy (wtATTR-CM) is a progressive, life-threatening, increasingly recognized but underdiagnosed cause of heart failure (HF). A previously validated machine learning (ML) model trained on medical claims data from 300 million US patients predicted wtATTR-CM among individuals diagnosed with HF with high sensitivity and specificity. Here, we simplified the ML model by reducing the number of predictive variables and tested it in a cohort of patients with confirmed wtATTR-CM at a large academic amyloid referral center. Methods: A retrospective, case-control study was conducted using electronic health records (EHR) from a random 1:1 sample of patients diagnosed with wtATTR-CM (cases) and non-amyloid HF (controls) at OHSU (Jul 2005-Nov 2019). Inclusion criteria were age ≥50 years; HF diagnosis (based on ICD-10 codes/SNOMED CT); and ≥1 of the following: ≥12 months of medical history, ≥5 clinical visits, or ≥10 documented diagnosis codes. The original 1,871 variables were systematically reduced to 15 based on recursive feature elimination and clinical relevance to ATTR-CM. After confirmation of the full ML model algorithm performance, the simplified model, validated against Optum EHR, was applied to the OHSU cohort of patients with wtATTR-CM. Results: Of 25,233 patients who met study criteria, 38 (0.2%) had wtATTR-CM and were evaluated along with 38 patients with non-amyloid HF. Performance of the simplified ML model was consistent with the previously validated model, with an ROC AUC of 0.812 and 0.804, respectively, and improved at lower thresholds (Table). Conclusions: A simplified ML algorithm to estimate the empirical probability of wtATTR-CM in patients with HF performed well at an academic amyloid referral center. This may serve as a practical approach to aid physicians in identifying HF patients who may be at-risk for wtATTR-CM. Additional studies are needed to confirm these findings in larger cohorts.

scholarly journals Detecting Developmental Delay and Autism Through Machine Learning Models Using Home Videos of Bangladeshi Children: Development and Validation Study (Preprint)

2019 ◽  
Author(s):  
Qandeel Tariq ◽  
Scott Lanyon Fleming ◽  
Jessey Nicole Schwartz ◽  
Kaitlyn Dunlap ◽  
Conor Corbin ◽  
...  
Keyword(s):  
Machine Learning ◽  
Developmental Delays ◽  
Area Under The Curve ◽  
Population Level ◽  
Autism Spectrum ◽  
Risk Scores ◽  
Future Research ◽  
Home Videos ◽  
Shapley Values ◽  
Children Development

BACKGROUND Autism spectrum disorder (ASD) is currently diagnosed using qualitative methods that measure between 20-100 behaviors, can span multiple appointments with trained clinicians, and take several hours to complete. In our previous work, we demonstrated the efficacy of machine learning classifiers to accelerate the process by collecting home videos of US-based children, identifying a reduced subset of behavioral features that are scored by untrained raters using a machine learning classifier to determine children’s “risk scores” for autism. We achieved an accuracy of 92% (95% CI 88%-97%) on US videos using a classifier built on five features. OBJECTIVE Using videos of Bangladeshi children collected from Dhaka Shishu Children’s Hospital, we aim to scale our pipeline to another culture and other developmental delays, including speech and language conditions. METHODS Although our previously published and validated pipeline and set of classifiers perform reasonably well on Bangladeshi videos (75% accuracy, 95% CI 71%-78%), this work improves on that accuracy through the development and application of a powerful new technique for adaptive aggregation of crowdsourced labels. We enhance both the utility and performance of our model by building two classification layers: The first layer distinguishes between typical and atypical behavior, and the second layer distinguishes between ASD and non-ASD. In each of the layers, we use a unique rater weighting scheme to aggregate classification scores from different raters based on their expertise. We also determine Shapley values for the most important features in the classifier to understand how the classifiers’ process aligns with clinical intuition. RESULTS Using these techniques, we achieved an accuracy (area under the curve [AUC]) of 76% (SD 3%) and sensitivity of 76% (SD 4%) for identifying atypical children from among developmentally delayed children, and an accuracy (AUC) of 85% (SD 5%) and sensitivity of 76% (SD 6%) for identifying children with ASD from those predicted to have other developmental delays. CONCLUSIONS These results show promise for using a mobile video-based and machine learning–directed approach for early and remote detection of autism in Bangladeshi children. This strategy could provide important resources for developmental health in developing countries with few clinical resources for diagnosis, helping children get access to care at an early age. Future research aimed at extending the application of this approach to identify a range of other conditions and determine the population-level burden of developmental disabilities and impairments will be of high value.

scholarly journals Machine learning approach for early detection of autism by combining questionnaire and home video screening

2018 ◽  
Vol 25 (8) ◽  
pp. 1000-1007 ◽  
Author(s):  
Halim Abbas ◽  
Ford Garberson ◽  
Eric Glover ◽  
Dennis P Wall
Keyword(s):  
Machine Learning ◽  
Clinical Study ◽  
Early Detection ◽  
Low Cost ◽  
Clinical Analysis ◽  
Autism Spectrum ◽  
Training Data ◽  
Screening Tools ◽  
Home Video ◽  
Home Videos

Abstract Background Existing screening tools for early detection of autism are expensive, cumbersome, time- intensive, and sometimes fall short in predictive value. In this work, we sought to apply Machine Learning (ML) to gold standard clinical data obtained across thousands of children at-risk for autism spectrum disorder to create a low-cost, quick, and easy to apply autism screening tool. Methods Two algorithms are trained to identify autism, one based on short, structured parent-reported questionnaires and the other on tagging key behaviors from short, semi-structured home videos of children. A combination algorithm is then used to combine the results into a single assessment of higher accuracy. To overcome the scarcity, sparsity, and imbalance of training data, we apply novel feature selection, feature engineering, and feature encoding techniques. We allow for inconclusive determination where appropriate in order to boost screening accuracy when conclusive. The performance is then validated in a controlled clinical study. Results A multi-center clinical study of n = 162 children is performed to ascertain the performance of these algorithms and their combination. We demonstrate a significant accuracy improvement over standard screening tools in measurements of AUC, sensitivity, and specificity. Conclusion These findings suggest that a mobile, machine learning process is a reliable method for detection of autism outside of clinical settings. A variety of confounding factors in the clinical analysis are discussed along with the solutions engineered into the algorithms. Final results are statistically limited and will benefit from future clinical studies to extend the sample size.

Sign in / Sign up

Export Citation Format

Share Document