Unwanted Information Research Articles

Frequency and reasons that parents decline genetic testing for critically ill neonates

PurposeCurrent literature reports strong support among parents for genetic testing for ill neonates; yet, some parents decline this testing for unknown reasons. We aimed to document the proportion of parents who decline, describe their clinical and demographic characteristics, and categorize their rationales. MethodsWe reviewed medical records to collect and compare clinical and demographic information for patients whose parents consented to and declined recommended genetic testing. We also conducted brief interviews with parents who declined testing to discover their rationales. ResultsFifty-one of 247 parents (21%) declined recommended genetic testing. The most common reason for declining, cited by 83% of parents interviewed, was that the testing felt irrelevant to the problems they saw as most important. The second most common reason, cited by 63%, was worrying that the testing might yield unwanted information. Compared with parents who consented, those who declined were more likely to be making the decision for a child with a prenatally diagnosed condition (P = .022) or congenital anomaly (P = .029) and to have private health insurance (P = .031). ConclusionParents who decline genetic testing for ill neonates provide an alternate appraisal of benefits and harms which should be incorporated into informing future parents considering these tests.

Suppressing the Maintenance of Information in Working Memory Alters Long-term Memory Traces.

The sensory recruitment hypothesis conceptualizes information in working memory as being activated representations of information in long-term memory. Accordingly, changes made to an item in working memory would be expected to influence its subsequent retention. Here, we tested the hypothesis that suppressing information from working memory, which can reduce short-term access to that information, may also alter its long-term neural representation. We obtained fMRI data (n = 25; 13 female / 12 male participants) while participants completed a working memory removal task with scene images as stimuli, followed by a final surprise recognition test of the examined items. We applied a multivariate pattern analysis to the data to quantify the engagement of suppression on each trial, to track the contents of working memory during suppression, and to assess representational changes afterward. Our analysis confirms previous reports that suppression of information in working memory involves focused attention to target and remove unwanted information. Furthermore, our findings provide new evidence that even a single dose of suppression of an item in working memory can (if engaged with sufficient strength) produce lasting changes in its neural representation, particularly weakening the unique, item-specific features, which leads to forgetting. Our study sheds light on the underlying mechanisms that contribute to the suppression of unwanted thoughts and highlights the dynamic interplay between working memory and long-term memory.

"Holy Cow, Where do I Sign up?" Attitudes of Military Veterans Toward Epigenomic Biomarker Toxic Exposure Testing.

After the Sergeant First Class Heath Robinson Honoring Our Promise to Address Comprehensive Toxics Promise to Address Comprehensive Toxics (PACT) Act in 2022, there has been a great interest in studying toxic exposures encountered during military service. Development of epigenomic biomarkers for exposures could facilitate understanding of exposure-related health effects, but such testing could also provide unwanted information. We explored attitudes toward epigenomic biomarker research and the potential to test for past exposures using semistructured interviews with Veterans (n = 22) who experienced potentially harmful exposures. Twenty Veterans said they would hypothetically want to receive epigenomic information related to their toxic exposures and potential health impacts as part of a research study. Veterans identified 9 potential benefits, including promoting insights concerning intergenerational health, identification of early health interventions, and additional knowledge or explanation for their experiences. Sixteen participants noted potential risks, including psychological distress, receiving nonactionable, uncertain, or inaccurate results, and privacy and discrimination risks. Ten participants identified at least 1 condition in their children that they thought could be related to their exposure and most said they would be interested in receiving research results related to their children's and grandchildren's risk. Results suggest that Veterans might welcome benefits of epigenomic research related to military exposures, yet have some concerns about potential negative impacts.

PULMO: Precise utterance-level modeling for speech anti-spoofing

In recent years, most state-of-the-art approaches for spoofed speech detection have been based on convolutional neural networks (CNNs). Most neural networks, including CNNs, are trained in minibatch units, where all input data in each minibatch must have the same shape. Therefore, for minibatch training, each utterance is first either padded or truncated because utterances are variable-length sequences and thus cannot be directly fed into networks in minibatch units. However, modeling either a padded or truncated utterance, rather than the original one, makes it unfeasible to capture the entire context as is: padding could propagate even unwanted information, like artifacts, in the original utterance, and truncation inevitably loses some information. With these information distortions, model could get stuck in a suboptimal solution. To fill this gap, we proposeÚ a method for precise utterance-level modeling that enables minibatch-wise utterance-level modeling of variable-length utterances while minimizing the information distortions. The proposed method comprises sequence segmentation followed by segment aggregation. Sequence segmentation feeds variable-length utterances in the minibatch unit by decomposing each of them into fixed-length segments, which enables parallel processing of variable-length utterances without the uncertainty in input length. Segment aggregation plays a role in aggregating the segment embeddings by utterance to encode the entire information of each utterance. The experimental results of the evaluation trials of ASVspoof 2019 and 2021 indicate that the proposed method shows up to 84.9 % and 97.6 % relative equal error rate reductions on logical and physical access scenarios, respectively. Furthermore, the proposed method reduced the FLOPs for an epoch by 6 %.

Information avoidance: An interchangeable strategy of self-protection

Across two studies, using five samples (N = 1,850), we examined whether health information avoidance—the deliberate decision to remain ignorant of available but unwanted personal health information—serves a defensive purpose and is interchangeable with other defensive strategies. We tested this idea by examining the relationship between health information avoidance—both as a dispositional tendency and deliberate decision—and feedback derogation. In Study 1, we experimentally demonstrated that a situation known to reduce defensiveness—self-uncertainty—decreased both proactive avoidance and reactive defensiveness relative to a control group. Study 2 demonstrated, across four samples, that people with a greater defensive orientation toward personal health information were more likely to derogate health information. These results are consistent with the idea that feedback derogation replaced the decision to avoid feedback. Together, results suggest that health information avoidance is likely part of a broader self-protective system and is replaceable with other motivated self-protection strategies.

Adversarial analysis of similarity-based sign prediction

Adversarial social network analysis explores how social links can be altered or otherwise manipulated to hinder unwanted information collection. To date, however, problems of this kind have not been studied in the context of signed networks in which links have positive and negative labels. Such formalism is often used to model social networks with positive links indicating friendship or support and negative links indicating antagonism or opposition.In this work, we present a computational analysis of the problem of attacking sign prediction in signed networks, whereby the aim of the attacker (a network member) is to hide from the defender (an analyst) the signs of a target set of links by removing the signs of some other, non-target, links. While the problem turns out to be NP-hard if either local or global similarity measures are used for sign prediction, we provide a number of positive computational results, including an FPT-algorithm for eliminating common signed neighborhood and heuristic algorithms for evading local similarity-based link prediction in signed networks.

The right to be forgotten in the “global village” paradigm: on the issue of implementation

The advent of the information era in the history of mankind also caused a change in the paradigm of social relations, transforming their format, according to Marshall McLuhan's apt phrase, into a "global village". In this paradigm, the right to be forgotten becomes an important means of protecting a person from uncontrolled dissemination of unwanted information about him. The article reveals the nature of the right to be forgotten and its relationship with the phenomenon of privacy, analyzes the provisions of European legislation and practice. Attention is drawn to the fact that in Ukraine, despite the existing need to have this tool for the protection of personally sensitive information, no legislative algorithm for the effective implementation of the right in question has been developed. It is noted that, from a legal and technical point of view, the implementation of the right to be forgotten is quite difficult, since the recognition of the genetic connection of the state of being forgotten with the right to the inviolability of private life in the “global village” paradigm forces the law enforcer to always seek a reasonable balance between a person's desire to limit access to a certain information and other basic human rights and freedoms (freedom of expression, the right of users to access any form of information, etc.). Given the situation of legal uncertainty, the article concludes that there are two ways to close this gap. The first of them is to speed up work on the adoption of the new Law of Ukraine “On the Protection of Personal Data”, the provisions of which would be devoted, among other things, to practical issues of implementation and protection of the right to be forgotten; the second is to develop a separate regulatory legal act, fully devoted to the analyzed law. At the same time, when working out an appropriate mechanism for the effective implementation of the right to be forgotten, one should take into account those aspects that in practice lead to paradoxical situations in their content, such as: the mismatch of traditional physical borders with the borders of the “global village”, the Streisand effect, the disproportionality of the efforts and the achieved result, the conflict of the person's past with the possibility of their further socialization in society, etc.).

The role of prior familiarisation and meaningfulness of verbal and visual stimuli on directed forgetting

ABSTRACT Intentional forgetting of unwanted information is a crucial cognitive function that is often studied with directed forgetting (DF) procedure, whereby cuing some study materials with Forget (F) instruction impairs their memory compared to cuing with Remember (R) instruction. This study investigates how the nature of information (verbal or pictorial), its semantic significance (meaningful or meaningless), and the degree of prior episodic familiarity influence DF. Before the DF phase, stimuli were familiarised by pre-exposing them 0, 2, or 6 times in a prior preview phase. Finally, memory for all items was assessed with old/new recognition test. Experiment 1 employed words, Experiment 2 utilised fractal images, Experiment 3 featured both meaningful and meaningless object images, and Experiment 4 used words and nonwords. Our results indicate that materials that produced better memory performance are not always harder to intentionally forget. Previewed items showed reduced DF compared to non-previewed items regardless of the nature of information, and meaningless stimuli are challenging to intentionally forget regardless of their degrees of familiarisation unless they are meaningless verbal materials. Collectively, the results highlight the importance of joint consideration of the stimulus format, its meaningfulness, and its episodic familiarity in understanding conditions that interact with intentional forgetting.

Computation of Fuzzy Linear Regression Model using Simulation Data

Regression analysis is a powerful technique for determining the causal influence on population outcome, although it is susceptible to outliers. It also oversimplifies the real-world data and problem, as data is rarely linearly separable. This paper analyses the simulation data of a fuzzy linear regression model (FLRM) model, which can aid in minimizing the interference of unwanted information, hence improving the precision of results. The aim of the fuzzy linear regression model (FLRM) is used to determine the best prediction model with the lowest measurement error. The model provides a fundamental mathematical and statistical framework for acknowledging the data imprecision. This study has applied the fuzzy linear regression model (FLRM) to evaluate the data in 15 rows of respondent models. The study implemented measurement error of cross-validation technique which is mean square error (MSE) and root mean square error (RMSE), to enhance data accuracy. Microsoft Excel and Matlab were applied to obtain the result. The simulation result indicates that comparing models with two measurement errors should be used to determine the optimal results. MSE and RMSE value with six parameters of a degree of fitting H-value have been calculated in the study. It concludes that a degree of fitting H-value of 0.0 is proven as a good model with the lowest MSE and RMSE measurement errors.

Proactive selective attention across competition contexts

Selective attention is a cognitive function that helps filter out unwanted information. Theories such as the biased competition model (Desimone & Duncan, 1995) explain how attentional templates bias processing towards targets in contexts where multiple stimuli compete for resources. However, it is unclear how the anticipation of different levels of competition influences the nature of attentional templates, in a proactive fashion. In this study, we used electroencephalography (EEG) to investigate how the anticipated demands of attentional selection (either high or low stimuli competition contexts) modulate target-specific preparatory brain activity and its relationship with task performance. To do so, participants performed a sex/gender judgment task in a cue-target paradigm where, depending on the block, target and distractor stimuli appeared simultaneously (high competition) or sequentially (low competition). Multivariate Pattern Analysis (MVPA) showed that, in both competition contexts, there was a preactivation of the target category to select, with a ramping-up profile at the end of the preparatory interval. However, cross-classification showed no generalization across competition conditions, suggesting different preparatory formats. Notably, time-frequency analyses showed differences between anticipated competition demands, with higher theta band power for high than low competition, which mediated the impact of subsequent stimuli competition on behavioral performance. Overall, our results show that, whereas preactivation of the internal templates associated with the category to select are engaged in advance in high and low competition contexts, their underlying neural patterns differ. In addition, these codes could not be associated with theta power, suggesting that they reflect different preparatory processes. The implications of these findings are crucial to increase our understanding of the nature of top-down processes across different contexts.

Learning to Unlearn: Instance-Wise Unlearning for Pre-trained Classifiers

Since the recent advent of regulations for data protection (e.g., the General Data Protection Regulation), there has been increasing demand in deleting information learned from sensitive data in pre-trained models without retraining from scratch. The inherent vulnerability of neural networks towards adversarial attacks and unfairness also calls for a robust method to remove or correct information in an instance-wise fashion, while retaining the predictive performance across remaining data. To this end, we consider instance-wise unlearning, of which the goal is to delete information on a set of instances from a pre-trained model, by either misclassifying each instance away from its original prediction or relabeling the instance to a different label. We also propose two methods that reduce forgetting on the remaining data: 1) utilizing adversarial examples to overcome forgetting at the representation-level and 2) leveraging weight importance metrics to pinpoint network parameters guilty of propagating unwanted information. Both methods only require the pre-trained model and data instances to forget, allowing painless application to real-life settings where the entire training set is unavailable. Through extensive experimentation on various image classification benchmarks, we show that our approach effectively preserves knowledge of remaining data while unlearning given instances in both single-task and continual unlearning scenarios.

A Robust and Simple Method for Filling in Masked Data in Astronomical Images

Astronomical images often have regions with missing or unwanted information, such as bad pixels, bad columns, cosmic rays, masked objects, or residuals from imperfect model subtractions. In certain situations it can be essential, or preferable, to fill in these regions. Most existing methods use low order interpolations for this task. In this paper a method is described that uses the full information that is contained in the pixels just outside masked regions. These edge pixels are extrapolated inwards, using iterative median filtering. This leads to a smoothly varying spatial resolution within the filled-in regions, and ensures seamless transitions between masked pixels and good pixels. Gaps in continuous, narrow features can be reconstructed with high fidelity, even if they are large. The method is implemented in maskfill, an open-source MIT licensed Python package (https://github.com/dokkum/maskfill). Its performance is illustrated with several examples, and compared to several alternative interpolation schemes.

Real-Time Cucumber Target Recognition in Greenhouse Environments Using Color Segmentation and Shape Matching

Accurate identification of fruits in greenhouse environments is an essential need for the precise functioning of agricultural robots. This study presents a solution to the problem of distinguishing cucumber fruits from their stems and leaves, which often have similar colors in their natural environment. The proposed algorithm for cucumber fruit identification relies on color segmentation and form matching. First, we get the boundary details from the acquired image of the cucumber sample. The edge information is described and reconstructed by utilizing a shape descriptor known as the Fourier descriptor in order to acquire a matching template image. Subsequently, we generate a multi-scale template by amalgamating computational and real-world data. The target image is subjected to color conditioning in order to enhance the segmenacktation of the target region inside the HSV color space. Then, the segmented target region is compared to the multi-scale template based on its shape. The method of color segmentation decreases the presence of unwanted information in the target image, hence improving the effectiveness of shape matching. An analysis was performed on a set of 200 cucumber photos that were obtained from the field. The findings indicate that the method presented in this study surpasses conventional recognition algorithms in terms of accuracy and efficiency, with a recognition rate of up to 86%. Moreover, the system has exceptional proficiency in identifying cucumber targets within greenhouses. This attribute renders it a great resource for offering technical assistance to agricultural robots that operate with accuracy.

GoogleNet transfer learning with improved gorilla optimized kernel extreme learning machine for accurate detection of asphalt pavement cracks

Asphalt cracks are the initial indication and major pavement structural distress in the field of civil engineering because it might potentially intimidate road traffic and highway safety. Managing and maintaining pavement structures are the two crucial issues faced eventually by road administrators. It is essential to take immediate action regarding pavement structural damages to prevent its severity. In recent times, there established numerous deep learning-based image processing methods to detect pavement cracks, but due to the presence of interference factors such as noises and road markings in the images, those methods failed to provide high accuracy and efficiency. Therefore, with the aim to yield high accuracy of crack detection, this article designs a novel asphalt pavement crack identification system “GoogleNet transfer learning with improved gorilla optimized kernel extreme learning machine” (GNet TL with IGT-KELM). The road crack images used for perusal are acquired from NHA12D dataset, in which it consists of 40 asphalt pavement images with two different viewpoints. For the purpose of data balancing, some non-crack images are gathered by field survey. The images with large noise distortions and unwanted background information influence crack detection performance so that the preprocessing steps are executed before applying it to the prediction system. The most significant crack and non-crack features from the images are extracted using GNet TL model. With the extracted features, the KELM is well-trained to detect cracks and non-cracks separately. To increase crack detection performance of the classifier, an improved gorilla troop optimizer is introduced to optimize the KELM parameters. The experimental finding reveals that the proposed detection mechanism achieves crack recognition accuracy of about 98.93%, which is considerably greater than compared baseline approaches.

Hybridized Clustering Algorithm and Ensemble Learning for Monitoring Paddy Crop Growth Analysis

In food security control and agricultural management, crop monitoring is important to improve the crop yield rate and ensure the paddy growth. The paddy crop is monitored via remote sensing satellite images that cover the entire cultivating region. The satellite images are affected due to satellite movement and human errors, which causes over fitting issues. In addition, the remote images are facing difficulties while managing system robustness. The research issue is overcome by applying the Hybridized Clustering Algorithm-based Ensemble Learning Hybridized Clustering Algorithm with Ensemble Learning (HCA-EL) approach. Initially, the satellite images are handled to remove geometric, radiometric, and atmospheric corrections using pre-processing techniques. This process uses the Ground Control Points (GCP) and histogram process to remove unwanted information. Then vegetation index value data points are obtained to progress the overall crop monitoring. The data points are analyzed using spectral clustering based on the eigenvector and eigenvalue, minimizing the dimensionality issues. Finally, the crop monitoring criteria, such as soil quality and water levels are computed to improve paddy growth. The monitoring process is done with the help of the Ensemble Learning Extreme Networks (ELEN), which uses the voting criteria to identify the quality parameters. The HCA-EL approach-based crop monitoring process was implemented using the Python tool with respective performance metrics like Accuracy, Error Rate, MCC and F1-Score.

DIGITAL TRANSFORMATION OF THE GLOBAL EDUCATION MARKET

The global education market is undergoing significant changes due to modern digitalisation. The latest technologies are significantly changing learning technologies, the structure of education, international flows and their volume in general. The subject of the study is to assess the impact of digital transformation on the global world market. The aim of this article is to analyse the development processes of the global market for educational services in the context of modern challenges, in particular digital transformation, and its consequences for the higher education market. The methodology of the study included research, a review of scientific literature, graphic visualisation of digitalisation. The methods of system analysis and generalisation have made it possible to identify the main directions of transformation of the global and regional educational markets, to characterise the dynamics of the spread of the latest forms of education and the growth of their profitability, to confirm the turbulence and instability of internationalisation trends and the intensification of competition between traditional and new providers of educational services. Research has shown that the online education sector is developing and expanding rapidly. The amount of capital raised and its profitability are increasing, confirming the prospects of these investments. There is a growing variety of online education providers, which generally creates significant competition for traditional institutions such as universities. Educational products are becoming more diversified and there is a growing trend towards personalised and adaptive learning on online educational platforms using artificial intelligence and machine learning technologies. In recent years, there has been a significant increase in the number of students enrolled in various online courses, and there is a growing demand for distance learning and flexible, personalised learning options. The global education market is characterised by unstable dynamics of internationalisation due to various external factors such as the COVID-19 pandemic, military aggression, demographic factors, political and economic instability. Competition between traditional and new providers for students, market share and financial resources is intensifying. In general, despite the negative factors, the number of foreign students tends to increase, albeit rather unevenly across countries. The widespread use of ICT in education and other activities can pose potential threats and challenges: increased competition and lower quality of education, spread of unwanted and anti-social information.

An Improved and Optimized Gated Recurrent Unit and Long Short-Term Memory Model for Fake News Detection

This study presents a novel approach for detecting counterfeit news, employing an advanced hybrid model that integrates Enhanced Gated Recurrent Unit and Long Short-Term Memory networks, termed as IGRU-LSTM. Initially, the database is assembled from the Information Security and Object Technology (ISOT) database and Wikipedia databases. From the database, the real and fake news is detected by considering news reviews. The dataset may contain unwanted information line URLs and symbols, which should be corrected to achieve efficient fake news detection. So, the pre-processing technique should be considered such as special symbol removal, URLS removal, upper to lower case conversion and replace contractions. After that, the pre-processed data is sent to the embedding procedure for word embedding. Finally, the IGRU-LSTM classifier is utilized for classifying real and fake news detection. In the combined GRU-LSTM framework, we incorporate the Enhanced Wild Horse Optimisation (EWHO) algorithm to optimize the selection of optimal weighting parameters. We utilize MATLAB for implementing this method. To evaluate the effectiveness of our approach, we analyze key performance metrics like precision, recall, and accuracy, and compare them with established methods including CNN-PSO, CNN-FO, and standard CNN.

How Resources Combining Expertise and Social Support Help Breastfeeding Women Address Self-Doubt and Increase Breastfeeding Self-Efficacy: A Mixed-Methods Study

ABSTRACT Existing research on parental information literacy focuses on parents’ ability to evaluate information. This approach does not account for conflicting or unwanted information and obscures the emotional impact of unsought information. We aimed to (1) document the sources women use most frequently for accessing breastfeeding information, (2) explore the reasons women choose some sources over others, (3) test the relationship between source characteristics and breastfeeding self-efficacy, and (4) determine the extent to which four source characteristics – competence, trustworthiness, goodwill, and social support – predict breastfeeding self-efficacy. This study was conducted in two phases. The first phase consisted of interviews and focus groups, which we analyzed inductively using a grounded theory approach. In phase two, we conducted a cross-sectional survey of women who were currently breastfeeding, collecting data on the source credibility and social support associated with the source they use most frequently and the participants’ breastfeeding self-efficacy. We used multiple regression to analyze our survey data. The frequency with which our participants encountered conflicting and unsought prompted them to view sources that combine expertise and social support as ideal. However, these sources are often difficult to access, leading to a reliance on online sources. This compromise has implications for breastfeeding tenure, as both competence and social support predicted breastfeeding self-efficacy. This study demonstrates how emotional aspects of information seeking shape women’s preferred sources, how accessibility limits the use of preferred sources, and how source competence and social support influence breastfeeding self-efficacy.

Redactable consortium blockchain with access control: Leveraging chameleon hash and multi-authority attribute-based encryption

A redactable blockchain allows authorized individuals to remove or replace undesirable content, offering the ability to remove illegal or unwanted information. Access control is a mechanism that limits data visibility and ensures that only authorized users can decrypt and access encrypted information, playing a crucial role in addressing privacy concerns and securing the data stored on a blockchain. Redactability and access control are both essential components when implementing a regulated consortium blockchain in real-world situations to ensure the secure sharing of data while removing undesirable content. We propose a decentralized consortium blockchain system prototype that supports redactability and access control. Through the development of a prototype blockchain system, we investigate the feasibility of combining these approaches and demonstrate that it is possible to implement a redactable blockchain with access control in a consortium blockchain setting.

Robust hyperparameter tuned deep Elman neural network for the diagnosis of osteosarcoma on histology images

Osteosarcomas are a type of bone tumour that can develop anywhere in the bone but most typically do so around the metaphyseal growth plates at the ends of long bones. Death rates can be lowered by early detection. Manual osteosarcoma identification can be difficult and requires specialised knowledge. With the aid of contemporary technology, medical photographs may now be automatically analysed and categorised, enabling quicker and more effective data processing. This paper proposes a novel hyperparameter-tuned deep learning (DL) approach for predicting osteosarcoma on histology images with effective feature selection mechanism which aims to improve the prediction accuracy of the classification system for bone tumor detection. The proposed system mainly consists of ‘6’ phases: data collection, preprocessing, segmentation, feature extraction, feature selection, and classification. Firstly, the dataset of histology images is gathered from openly available sources. Then Median Filtering (MEF) is utilized as the preprocessing step that enhances the quality of the input images for accurate prediction by eliminating unwanted information from them. Afterwards, the pre-processed image was segmented using Harmonic Mean-based Otsu Thresholding (HMOTH) approach to obtain the tumor-affected regions from the pre-processed data. Then the features from the segmented tumor portions are extracted using the Self-Attention Mechanism-based MobileNet (SAMMNet) model. A Van der Corput sequence and Adaptive Inertia Weight included Reptile Search Optimization Algorithm (VARSOA) is used to select the more relevant features from the extracted features. Finally, a Hyperparameter-Tuned Deep Elman Neural Network (HTDENN) is utilized to diagnose and classify osteosarcoma, in which the hyperparameters of the neural network are obtained optimally using the VARSOA. The proposed HTDENN attains the higher accuracy of 0.9531 for the maximum of 200 epochs, whereas the existing DENN, MLP, RF, and SVM attains the accuracies of 0.9492, 0.9427, 0.9413, and 0.9387. Likewise, the proposed model attains the better results for precision (0.9511), f-measure (0.9423), sensitivity (0.9345) and specificity (0.9711) than the existing approaches for the maximum of 200 epochs. Simulation outcomes proved that the proposed model outperforms existing research frameworks for osteosarcoma prediction and classification.