Crowdsourced privacy-preserved feature tagging of short home videos for machine learning ASD detection.
Ontology highlight
ABSTRACT:
SUBMITTER: Washington P
PROVIDER: S-EPMC8027393 | biostudies-literature |
REPOSITORIES: biostudies-literature
Similar Datasets
Project description:Probabilistic record linkage is a process used to bring together person-based records from within the same dataset (de-duplication) or from disparate datasets using pairwise comparisons and matching probabilities. The linkage strategy and associated match probabilities are often estimated through investigations into data quality and manual inspection. However, as privacy-preserved datasets comprise encrypted data, such methods are not possible. In this paper, we present a method for estimating the probabilities and threshold values for probabilistic privacy-preserved record linkage using Bloom filters.Our method was tested through a simulation study using synthetic data, followed by an application using real-world administrative data. Synthetic datasets were generated with error rates from zero to 20% error. Our method was used to estimate parameters (probabilities and thresholds) for de-duplication linkages. Linkage quality was determined by F-measure. Each dataset was privacy-preserved using separate Bloom filters for each field. Match probabilities were estimated using the expectation-maximisation (EM) algorithm on the privacy-preserved data. Threshold cut-off values were determined by an extension to the EM algorithm allowing linkage quality to be estimated for each possible threshold. De-duplication linkages of each privacy-preserved dataset were performed using both estimated and calculated probabilities. Linkage quality using the F-measure at the estimated threshold values was also compared to the highest F-measure. Three large administrative datasets were used to demonstrate the applicability of the probability and threshold estimation technique on real-world data.Linkage of the synthetic datasets using the estimated probabilities produced an F-measure that was comparable to the F-measure using calculated probabilities, even with up to 20% error. Linkage of the administrative datasets using estimated probabilities produced an F-measure that was higher than the F-measure using calculated probabilities. Further, the threshold estimation yielded results for F-measure that were only slightly below the highest possible for those probabilities.The method appears highly accurate across a spectrum of datasets with varying degrees of error. As there are few alternatives for parameter estimation, the approach is a major step towards providing a complete operational approach for probabilistic linkage of privacy-preserved datasets.
| S-EPMC5504757 | biostudies-other
Project description:Autism Spectrum Disorder is a neuropsychiatric condition affecting 53 million children worldwide and for which early diagnosis is critical to the outcome of behavior therapies. Machine learning applied to features manually extracted from readily accessible videos (e.g., from smartphones) has the potential to scale this diagnostic process. However, nearly unavoidable variability in video quality can lead to missing features that degrade algorithm performance. To manage this uncertainty, we evaluated the impact of missing values and feature imputation methods on two previously published autism detection classifiers, trained on standard-of-care instrument scoresheets and tested on ratings of 140 children videos from YouTube. We compare the baseline method of listwise deletion to classic univariate and multivariate techniques. We also introduce a feature replacement method that, based on a score, selects a feature from an expanded dataset to fill-in the missing value. The replacement feature selected can be identical for all records (general) or automatically adjusted to the record considered (dynamic). Our results show that general and dynamic feature replacement methods achieve a higher performance than classic univariate and multivariate methods, supporting the hypothesis that algorithmic management can maintain the fidelity of video-based diagnostics in the face of missing values and variable video quality.
| S-EPMC7719177 | biostudies-literature
Project description:BackgroundAutism 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.ObjectiveUsing 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.MethodsAlthough 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.ResultsUsing 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.ConclusionsThese 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.
| S-EPMC6505375 | biostudies-literature
Project description:The Internet of Things (IoT) is a widely hyped concept, with its focus on the connection of smart devices to the Internet rather than on people. IoT for consumers is often called the smart home market, and a large part of that market consists of home security devices. Consumers are often motivated to purchase smart home security devices to prevent burglaries, which they fear may lead to damage to their property or threats to their families. However, they also understand that IoT home security devices may be a threat to the privacy of their personal information. To determine the relative roles of fear and privacy concerns in the decision to purchase IoT home security devices, we conducted a survey of American consumers. We used the Theory of Reasoned Action as the theoretical basis for the study. We found that fear positively affected consumer attitudes toward purchasing smart home security devices, while concerns about privacy negatively affected attitudes. We found that attitudes toward purchase, the opinions of important others, and experience with burglaries all affected intent to purchase. We also found that the relationship between privacy concerns and intent to purchase is completely mediated by attitudes, while fear has both direct and indirect effects on intent.
| S-EPMC8454981 | biostudies-literature
Project description:With the increase in applications of machine learning methods in drug design and related fields, the challenge of designing sound test sets becomes more and more prominent. The goal of this challenge is to have a realistic split of chemical structures (compounds) between training, validation and test set such that the performance on the test set is meaningful to infer the performance in a prospective application. This challenge is by its own very interesting and relevant, but is even more complex in a federated machine learning approach where multiple partners jointly train a model under privacy-preserving conditions where chemical structures must not be shared between the different participating parties. In this work we discuss three methods which provide a splitting of a data set and are applicable in a federated privacy-preserving setting, namely: a. locality-sensitive hashing (LSH), b. sphere exclusion clustering, c. scaffold-based binning (scaffold network). For evaluation of these splitting methods we consider the following quality criteria (compared to random splitting): bias in prediction performance, classification label and data imbalance, similarity distance between the test and training set compounds. The main findings of the paper are a. both sphere exclusion clustering and scaffold-based binning result in high quality splitting of the data sets, b. in terms of compute costs sphere exclusion clustering is very expensive in the case of federated privacy-preserving setting.
| S-EPMC8650276 | biostudies-literature
Project description:Liquid-phase transmission electron microscopy (TEM) has been recently applied to materials chemistry to gain fundamental understanding of various reaction and phase transition dynamics at nanometer resolution. However, quantitative extraction of physical and chemical parameters from the liquid-phase TEM videos remains bottlenecked by the lack of automated analysis methods compatible with the videos' high noisiness and spatial heterogeneity. Here, we integrate, for the first time, liquid-phase TEM imaging with our customized analysis framework based on a machine learning model called U-Net neural network. This combination is made possible by our workflow to generate simulated TEM images as the training data with well-defined ground truth. We apply this framework to three typical systems of colloidal nanoparticles, concerning their diffusion and interaction, reaction kinetics, and assembly dynamics, all resolved in real-time and real-space by liquid-phase TEM. A diversity of properties for differently shaped anisotropic nanoparticles are mapped, including the anisotropic interaction landscape of nanoprisms, curvature-dependent and staged etching profiles of nanorods, and an unexpected kinetic law of first-order chaining assembly of concave nanocubes. These systems representing properties at the nanoscale are otherwise experimentally inaccessible. Compared to the prevalent image segmentation methods, U-Net shows a superior capability to predict the position and shape boundary of nanoparticles from highly noisy and fluctuating background-a challenge common and sometimes inevitable in liquid-phase TEM videos. We expect our framework to push the potency of liquid-phase TEM to its full quantitative level and to shed insights, in high-throughput and statistically significant fashion, on the nanoscale dynamics of synthetic and biological nanomaterials.
| S-EPMC7453571 | biostudies-literature
Project description:BackgroundThe protection of private data is a key responsibility for research studies that collect identifiable information from study participants. Limiting the scope of data collection and preventing secondary use of the data are effective strategies for managing these risks. An ideal framework for data collection would incorporate feature engineering, a process where secondary features are derived from sensitive raw data in a secure environment without a trusted third party.ObjectiveThis study aimed to compare current approaches based on how they maintain data privacy and the practicality of their implementations. These approaches include traditional approaches that rely on trusted third parties, and cryptographic, secure hardware, and blockchain-based techniques.MethodsA set of properties were defined for evaluating each approach. A qualitative comparison was presented based on these properties. The evaluation of each approach was framed with a use case of sharing geolocation data for biomedical research.ResultsWe found that approaches that rely on a trusted third party for preserving participant privacy do not provide sufficiently strong guarantees that sensitive data will not be exposed in modern data ecosystems. Cryptographic techniques incorporate strong privacy-preserving paradigms but are appropriate only for select use cases or are currently limited because of computational complexity. Blockchain smart contracts alone are insufficient to provide data privacy because transactional data are public. Trusted execution environments (TEEs) may have hardware vulnerabilities and lack visibility into how data are processed. Hybrid approaches combining blockchain and cryptographic techniques or blockchain and TEEs provide promising frameworks for privacy preservation. For reference, we provide a software implementation where users can privately share features of their geolocation data using the hybrid approach combining blockchain with TEEs as a supplement.ConclusionsBlockchain technology and smart contracts enable the development of new privacy-preserving feature engineering methods by obviating dependence on trusted parties and providing immutable, auditable data processing workflows. The overlap between blockchain and cryptographic techniques or blockchain and secure hardware technologies are promising fields for addressing important data privacy needs. Hybrid blockchain and TEE frameworks currently provide practical tools for implementing experimental privacy-preserving applications.
| S-EPMC6712958 | biostudies-literature
Project description:Lentiviral (LV) vectors have emerged as powerful tools for transgene delivery ex vivo but in vivo gene therapy applications involving LV vectors have faced a number of challenges, including the low efficiency of transgene delivery, a lack of tissue specificity, immunogenicity to both the product encoded by the transgene and the vector, and the inactivation of the vector by the human complement cascade. To mitigate these issues, several engineering approaches, involving the covalent modification of vector particles or the incorporation of specific protein domains into the vector's envelope, have been tested. Short synthetic oligonucleotides, including aptamers bound to the surface of LV vectors, may provide a novel means with which to retarget LV vectors to specific cells and to shield these vectors from neutralization by sera. The purpose of this study was to develop strategies to tether nucleic acid sequences, including short RNA sequences, to LV vector particles in a specific and tight fashion. To bind short RNA sequences to LV vector particles, a bacteriophage lambda N protein-derived RNA binding domain (λN), fused to the measles virus hemagglutinin protein, was used. The λN protein bound RNA sequences bearing a boxB RNA hairpin. To test this approach, we used an RNA aptamer specific to the human epidermal growth factor receptor (EGFR), which was bound to LV vector particles via an RNA scaffold containing a boxB RNA motif. The results obtained confirmed that the EGFR-specific RNA aptamer bound to cells expressing EGFR and that the boxB containing the RNA scaffold was bound specifically to the λN RNA binding domain attached to the vector. These results show that LV vectors can be equipped with nucleic acid sequences to develop improved LV vectors for in vivo applications.
| S-EPMC8508951 | biostudies-literature
Project description:MotivationClassification of individuals into disease or clinical categories from high-dimensional biological data with low prediction error is an important challenge of statistical learning in bioinformatics. Feature selection can improve classification accuracy but must be incorporated carefully into cross-validation to avoid overfitting. Recently, feature selection methods based on differential privacy, such as differentially private random forests and reusable holdout sets, have been proposed. However, for domains such as bioinformatics, where the number of features is much larger than the number of observations p?n , these differential privacy methods are susceptible to overfitting.MethodsWe introduce private Evaporative Cooling, a stochastic privacy-preserving machine learning algorithm that uses Relief-F for feature selection and random forest for privacy preserving classification that also prevents overfitting. We relate the privacy-preserving threshold mechanism to a thermodynamic Maxwell-Boltzmann distribution, where the temperature represents the privacy threshold. We use the thermal statistical physics concept of Evaporative Cooling of atomic gases to perform backward stepwise privacy-preserving feature selection.ResultsOn simulated data with main effects and statistical interactions, we compare accuracies on holdout and validation sets for three privacy-preserving methods: the reusable holdout, reusable holdout with random forest, and private Evaporative Cooling, which uses Relief-F feature selection and random forest classification. In simulations where interactions exist between attributes, private Evaporative Cooling provides higher classification accuracy without overfitting based on an independent validation set. In simulations without interactions, thresholdout with random forest and private Evaporative Cooling give comparable accuracies. We also apply these privacy methods to human brain resting-state fMRI data from a study of major depressive disorder.Availability and implementationCode available at http://insilico.utulsa.edu/software/privateEC .Contactbrett-mckinney@utulsa.edu.Supplementary informationSupplementary data are available at Bioinformatics online.
| S-EPMC5870708 | biostudies-literature
Project description:The complex genetic architecture of Autism Spectrum Disorder (ASD) and its heterogeneous phenotype makes molecular diagnosis and patient prognosis challenging tasks. To establish more precise genotype-phenotype correlations in ASD, we developed a novel machine-learning integrative approach, which seeks to delineate associations between patients' clinical profiles and disrupted biological processes, inferred from their copy number variants (CNVs) that span brain genes. Clustering analysis of the relevant clinical measures from 2446 ASD cases in the Autism Genome Project identified two distinct phenotypic subgroups. Patients in these clusters differed significantly in ADOS-defined severity, adaptive behavior profiles, intellectual ability, and verbal status, the latter contributing the most for cluster stability and cohesion. Functional enrichment analysis of brain genes disrupted by CNVs in these ASD cases identified 15 statistically significant biological processes, including cell adhesion, neural development, cognition, and polyubiquitination, in line with previous ASD findings. A Naive Bayes classifier, generated to predict the ASD phenotypic clusters from disrupted biological processes, achieved predictions with a high precision (0.82) but low recall (0.39), for a subset of patients with higher biological Information Content scores. This study shows that milder and more severe clinical presentations can have distinct underlying biological mechanisms. It further highlights how machine-learning approaches can reduce clinical heterogeneity by using multidimensional clinical measures, and establishes genotype-phenotype correlations in ASD. However, predictions are strongly dependent on patient's information content. Findings are therefore a first step toward the translation of genetic information into clinically useful applications, and emphasize the need for larger datasets with very complete clinical and biological information.
| S-EPMC7026098 | biostudies-literature