Machine Learning Methods for Dysgraphia Screening with Online Handwriting Features

Kunhoth, Jayakanth; Saleh, Moutaz; Akbari, Younus

doi:10.1109/icca56443.2022.10039584

Cited by 4 publications

(5 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…RENT acquires information on selection stability for each feature and utilizes this information for the selection of the final set of features. RENT, or slight modifications thereof, has been shown to be useful specifically for high-dimensional datasets consisting of more features than samples and has been applied in life-science research (44)(45)(46).…”

Section: Feature Selection Using Rentrepeated Elastic Net Techniquementioning

confidence: 99%

Head and neck cancer treatment outcome prediction: a comparison between machine learning with conventional radiomics features and deep learning radiomics

Huynh,

Groendahl,

Tomic

et al. 2023

Front. Med.

View full text Add to dashboard Cite

BackgroundRadiomics can provide in-depth characterization of cancers for treatment outcome prediction. Conventional radiomics rely on extraction of image features within a pre-defined image region of interest (ROI) which are typically fed to a classification algorithm for prediction of a clinical endpoint. Deep learning radiomics allows for a simpler workflow where images can be used directly as input to a convolutional neural network (CNN) with or without a pre-defined ROI.PurposeThe purpose of this study was to evaluate (i) conventional radiomics and (ii) deep learning radiomics for predicting overall survival (OS) and disease-free survival (DFS) for patients with head and neck squamous cell carcinoma (HNSCC) using pre-treatment 18F-fluorodeoxuglucose positron emission tomography (FDG PET) and computed tomography (CT) images.Materials and methodsFDG PET/CT images and clinical data of patients with HNSCC treated with radio(chemo)therapy at Oslo University Hospital (OUS; n = 139) and Maastricht University Medical Center (MAASTRO; n = 99) were collected retrospectively. OUS data was used for model training and initial evaluation. MAASTRO data was used for external testing to assess cross-institutional generalizability. Models trained on clinical and/or conventional radiomics features, with or without feature selection, were compared to CNNs trained on PET/CT images without or with the gross tumor volume (GTV) included. Model performance was measured using accuracy, area under the receiver operating characteristic curve (AUC), Matthew’s correlation coefficient (MCC), and the F1 score calculated for both classes separately.ResultsCNNs trained directly on images achieved the highest performance on external data for both endpoints. Adding both clinical and radiomics features to these image-based models increased performance further. Conventional radiomics including clinical data could achieve competitive performance. However, feature selection on clinical and radiomics data lead to overfitting and poor cross-institutional generalizability. CNNs without tumor and node contours achieved close to on-par performance with CNNs including contours.ConclusionHigh performance and cross-institutional generalizability can be achieved by combining clinical data, radiomics features and medical images together with deep learning models. However, deep learning models trained on images without contours can achieve competitive performance and could see potential use as an initial screening tool for high-risk patients.

show abstract

Section: Feature Selection Using Rentrepeated Elastic Net Techniquementioning

confidence: 99%

Head and neck cancer treatment outcome prediction: a comparison between machine learning with conventional radiomics features and deep learning radiomics

Huynh,

Groendahl,

Tomic

et al. 2023

Front. Med.

View full text Add to dashboard Cite

show abstract

“…In the evaluation, it is essential to select the appropriate metrics based on the specific problem and the available data and consider the results together to obtain a complete review of the performance of the algorithm; in Figure 3, the flow chart used for the process is presented for an algorithm evaluation [29]. In addition, it is essential to consider that the review of algorithms for detecting abnormal patterns in children must be carried out in collaboration with professionals who are specialized in writing and child development.…”

Section: Discussionmentioning

confidence: 99%

Detection of Abnormal Patterns in Children’s Handwriting by Using an Artificial-Intelligence-Based Method

2023

View full text Add to dashboard Cite

Using camera-based algorithms to detect abnormal patterns in children’s handwriting has become a promising tool in education and occupational therapy. This study analyzes the performance of a camera- and tablet-based handwriting verification algorithm to detect abnormal patterns in handwriting samples processed from 71 students of different grades. The study results revealed that the algorithm saw abnormal patterns in 20% of the handwriting samples processed, which included practices such as delayed typing speed, excessive pen pressure, irregular slant, and lack of word spacing. In addition, it was observed that the detection accuracy of the algorithm was 95% when comparing the camera data with the abnormal patterns detected, which indicates a high reliability in the results obtained. The highlight of the study was the feedback provided to children and teachers on the camera data and any abnormal patterns detected. This can significantly impact students’ awareness and improvement of writing skills by providing real-time feedback on their writing and allowing them to adjust to correct detected abnormal patterns.

show abstract

“…Likewise, an additional research effort focuses on dysgraphia as a crucial learning disability impacting writing abilities and potentially exerting adverse effects on the academic performance of children if not identified in the early stages [17]. The main focus is to propose automated methods for the diagnosis of dysgraphia through the analysis of handwriting, specifically using the kinematics and dynamics of writing.…”

Section: Literature Reviewmentioning

confidence: 99%

“…The accuracy metric was used to evaluate the overall proportion of correct predictions made by the system. This metric is calculated using (1), this involves the total of true positives and true negatives divided by the overall number of cases. Accuracy provides an overall measure of system performance in terms of correct predictions over the entire dataset [22].…”

Section: Accuracymentioning

confidence: 99%

“…Learning disorders such as dyslexia, dysgraphia, and dyscalculia pose significant challenges to the educational process of children. These learning disabilities affect approximately 10% of children worldwide [1]. Early diagnosis of these disorders is crucial to minimize disruptions in learning development and to provide appropriate treatment.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Expert system for diagnosing learning disorders in children

Andrade-Arenas,

Yactayo-Arias

2024

IJECE

View full text Add to dashboard Cite

Given the urgent need for early detection of learning disorders such as dysgraphia, dyslexia, and dyscalculia in children, this study aimed to evaluate an expert system developed in Python to facilitate early diagnosis of these disorders. The background highlights the importance of providing parents, educators, and health professionals with an effective tool for early detection of these disorders. In 21 simulated cases, the system showed impressive performance with an accuracy rate of 95%, a precision of 100%, a sensitivity of 93%, and a specificity of 100%. Furthermore, the acceptability evaluation, conducted with 15 parents selected by convenience sampling, showed a high level of satisfaction, with an overall mean of 4.78 and a standard deviation of 0.45, indicating consistency in responses. In conclusion, this study confirms the effectiveness of the expert system in the early diagnosis of learning disabilities, providing parents, educators, and health professionals with a valuable tool. Despite these encouraging results, the need for additional research is recognized to address limitations and improve the external validity of the system to ensure its widespread utility and adaptability in real clinical settings.

show abstract

Machine Learning Methods for Dysgraphia Screening with Online Handwriting Features

Cited by 4 publications

References 18 publications

Head and neck cancer treatment outcome prediction: a comparison between machine learning with conventional radiomics features and deep learning radiomics

Head and neck cancer treatment outcome prediction: a comparison between machine learning with conventional radiomics features and deep learning radiomics

Detection of Abnormal Patterns in Children’s Handwriting by Using an Artificial-Intelligence-Based Method

Expert system for diagnosing learning disorders in children

Contact Info

Product

Resources

About