arXiv:2511.15485v1 Announce Type: cross 
Abstract: Parkinson's disease is a neurodegenerative disorder that can be very tricky to diagnose and treat. Such early symptoms can include tremors, wheezy breathing, and changes in voice quality as critical indicators of neural damage. Notably, there has been growing interest in utilizing changes in vocal attributes as markers for the detection of PD early on. Based on this understanding, the present paper was designed to focus on the acoustic feature analysis based on voice recordings of patients diagnosed with PD and healthy controls (HC). In this paper, we introduce a novel classification and visualization model known as CustNetGC, combining a Convolutional Neural Network (CNN) with Custom Network Grad-CAM and CatBoost to enhance the efficiency of PD diagnosis. We use a publicly available dataset from Figshare, including voice recordings of 81 participants: 40 patients with PD and 41 healthy controls. From these recordings, we extracted the key spectral features: L-mHP and Spectral Slopes. The L-mHP feature combines three spectrogram representations: Log-Mel spectrogram, harmonic spectrogram, and percussive spectrogram, which are derived using Harmonic-Percussive Source Separation (HPSS). Grad-CAM was used to highlight the important regions in the data, thus making the PD predictions interpretable and effective. Our proposed CustNetGC model achieved an accuracy of 99.06% and precision of 95.83%, with the area under the ROC curve (AUC) recorded at 0.90 for the PD class and 0.89 for the HC class. Additionally, the combination of CatBoost, a gradient boosting algorithm, enhanced the robustness and the prediction performance by properly classifying PD and non-PD samples. Therefore, the results provide the potential improvement in the CustNetGC system in enhancing diagnostic accuracy and the interpretability of the Parkinson's Disease prediction model.

تم تطوير نموذج جديد يسمى CustNetGC لتحسين التنبؤ بمرض باركنسون (PD) من خلال تحليل السمات الصوتية. يجمع هذا النموذج بين شبكة عصبية تلافيفية (CNN) وCustom Network Grad-CAM وCatBoost، باستخدام تسجيلات صوتية من 81 مشاركًا، بما في ذلك 40 مريضًا بمرض PD و41 شخصًا سليمًا. تهدف الدراسة إلى تحسين الكشف المبكر عن مرض PD من خلال تحليل الميزات الصوتية.

Se ha desarrollado un nuevo modelo llamado CustNetGC para mejorar la predicción de la enfermedad de Parkinson (EP) mediante el análisis de atributos vocales. Este modelo integra una red neuronal convolucional (CNN) con Custom Network Grad-CAM y CatBoost, utilizando grabaciones de voz de 81 participantes, incluidos 40 pacientes con EP y 41 controles sanos. El estudio tiene como objetivo mejorar la detección temprana de la EP a través del análisis de características acústicas.

Un nouveau modèle appelé CustNetGC a été développé pour améliorer la prédiction de la maladie de Parkinson (MP) en analysant les attributs vocaux. Ce modèle intègre un réseau de neurones convolutifs (CNN) avec Custom Network Grad-CAM et CatBoost, utilisant des enregistrements vocaux de 81 participants, dont 40 patients atteints de MP et 41 témoins sains. L'étude vise à améliorer la détection précoce de la MP grâce à l'analyse des caractéristiques acoustiques.

A novel model called CustNetGC has been developed to enhance the prediction of Parkinson's disease (PD) by analyzing vocal attributes. This model integrates a Convolutional Neural Network (CNN) with Custom Network Grad-CAM and CatBoost, utilizing voice recordings from 81 participants, including 40 PD patients and 41 healthy controls. The study aims to improve early detection of PD through acoustic feature analysis.

A Novel CustNetGC Boosted Model with Spectral Features for Parkinson's Disease Prediction

arXiv:2601.08240v1 Announce Type: cross 
Abstract: Diabetic retinopathy (DR), affecting millions globally with projections indicating a significant rise, poses a severe blindness risk and strains healthcare systems. Diagnostic complexity arises from visual symptom overlap with conditions like age-related macular degeneration and hypertensive retinopathy, exacerbated by high misdiagnosis rates in underserved regions. This study introduces TIMM-ProRS, a novel deep learning framework integrating Vision Transformer (ViT), Convolutional Neural Network (CNN), and Graph Neural Network (GNN) with multi-modal fusion. TIMM-ProRS uniquely leverages both retinal images and temporal biomarkers (HbA1c, retinal thickness) to capture multi-modal and temporal dynamics. Evaluated comprehensively across diverse datasets including APTOS 2019 (trained), Messidor-2, RFMiD, EyePACS, and Messidor-1 (validated), the model achieves 97.8\% accuracy and an F1-score of 0.96, demonstrating state-of-the-art performance and outperforming existing methods like RSG-Net and DeepDR. This approach enables early, precise, and interpretable diagnosis, supporting scalable telemedical management and enhancing global eye health sustainability.

تم تقديم إطار جديد للتعلم العميق يسمى TIMM-ProRS لتحسين التنبؤ وتصنيف المخاطر لاعتلال الشبكية السكري (DR)، وهي حالة تهدد رؤية الملايين حول العالم. يجمع هذا الإطار بين تقنيات المحولات البصرية، والشبكات العصبية التلافيفية، والشبكات العصبية البيانية، مستخدمًا كل من صور الشبكية والبيانات الحيوية الزمنية لتحقيق معدل دقة مرتفع يبلغ 97.8% عبر مجموعات بيانات متعددة.

Se ha presentado un nuevo marco de aprendizaje profundo llamado TIMM-ProRS para mejorar el pronóstico y la estratificación del riesgo de la retinopatía diabética (RD), una condición que amenaza la visión de millones de personas en todo el mundo. Este marco integra tecnologías de Vision Transformer, Red Neuronal Convolucional y Red Neuronal Gráfica, utilizando tanto imágenes retinianas como biomarcadores temporales para lograr una alta tasa de precisión del 97.8% en múltiples conjuntos de datos.

Un nouveau cadre d'apprentissage profond nommé TIMM-ProRS a été introduit pour améliorer le pronostic et la stratification des risques de la rétinopathie diabétique (RD), une affection qui menace la vision de millions de personnes dans le monde. Ce cadre intègre des technologies de Vision Transformer, de Réseau de Neurones Convolutifs et de Réseau de Neurones Graphiques, utilisant à la fois des images rétiniennes et des biomarqueurs temporels pour atteindre un taux de précision élevé de 97,8 % sur plusieurs ensembles de données.

A novel deep learning framework named TIMM-ProRS has been introduced to enhance the prognosis and risk stratification of diabetic retinopathy (DR), a condition that threatens the vision of millions worldwide. This framework integrates Vision Transformer, Convolutional Neural Network, and Graph Neural Network technologies, utilizing both retinal images and temporal biomarkers to achieve a high accuracy rate of 97.8% across multiple datasets.

Temporal-Enhanced Interpretable Multi-Modal Prognosis and Risk Stratification Framework for Diabetic Retinopathy (TIMM-ProRS)

arXiv:2601.08618v1 Announce Type: new 
Abstract: Multiple binary responses arise in many modern data-analytic problems. Although fitting separate logistic regressions for each response is computationally attractive, it ignores shared structure and can be statistically inefficient, especially in high-dimensional and class-imbalanced regimes. Low-rank models offer a natural way to encode latent dependence across tasks, but existing methods for binary data are largely likelihood-based and focus on pointwise classification rather than ranking performance. In this work, we propose a unified framework for learning with multiple binary responses that directly targets discrimination by minimizing a surrogate loss for the area under the ROC curve (AUC). The method aggregates pairwise AUC surrogate losses across responses while imposing a low-rank constraint on the coefficient matrix to exploit shared structure. We develop a scalable projected gradient descent algorithm based on truncated singular value decomposition. Exploiting the fact that the pairwise loss depends only on differences of linear predictors, we simplify computation and analysis. We establish non-asymptotic convergence guarantees, showing that under suitable regularity conditions, leading to linear convergence up to the minimax-optimal statistical precision. Extensive simulation studies demonstrate that the proposed method is robust in challenging settings such as label switching and data contamination and consistently outperforms likelihood-based approaches.

دراسة جديدة نشرت على arXiv تقدم إطارًا قويًا لتقدير الرتبة المنخفضة للتعلم مع استجابات ثنائية متعددة، باستخدام خسارة AUC الزوجية لتعزيز أداء التمييز. تتناول هذه الطريقة قيود نماذج الانحدار اللوجستي التقليدية من خلال التقاط الهياكل المشتركة عبر المهام، خاصة في السيناريوهات عالية الأبعاد وغير المتوازنة من حيث الفئات.

Un nuevo estudio publicado en arXiv introduce un marco de estimación robusta de bajo rango para el aprendizaje con múltiples respuestas binarias, utilizando la pérdida AUC por pares para mejorar el rendimiento de discriminación. Este método aborda las limitaciones de los modelos de regresión logística tradicionales al capturar estructuras compartidas entre tareas, especialmente en escenarios de alta dimensión y desbalance de clases.

Une nouvelle étude publiée sur arXiv présente un cadre d'estimation à faible rang robuste pour l'apprentissage avec plusieurs réponses binaires, utilisant la perte AUC pairwise pour améliorer la performance de discrimination. Cette méthode aborde les limitations des modèles de régression logistique traditionnels en capturant les structures partagées entre les tâches, en particulier dans des scénarios à haute dimension et déséquilibrés par classe.

A new study published on arXiv introduces a robust low-rank estimation framework for learning with multiple binary responses, utilizing pairwise AUC loss to enhance discrimination performance. This method addresses the limitations of traditional logistic regression models by capturing shared structures across tasks, particularly in high-dimensional and class-imbalanced scenarios.

A Novel CustNetGC Boosted Model with Spectral Features for Parkinson's Disease Prediction

Was this article worth reading? Share it

Airparser

LucidQuery AI

Thinkability

Hypertune

Supanote

DrawMy.Pet

Ready to build your own newsroom?