| Volume 29, Issue 1 | |
| Classification of COVID-19 Using Temporal and Spectral Features of Cough Sounds | 5-18 |
| Biruk Abera Tessema | |
| doi: 10.7546/ijba.2025.29.1.000931 | |
| [ +/- how to cite ][ +/- abstract ][ full text ] | |
| Biruk Abera Tessema (2025) Classification of COVID-19 Using Temporal and Spectral Features of Cough Sounds, Int J Bioautomation, 29 (1), 5-18, doi: 10.7546/ijba.2025.29.1.000931 | |
| Abstract: Chest X-ray and computed tomography scan play a major role in the diagnosis of lung diseases, including coronavirus disease (COVID-19). However, their cost, the obstacles to their implementation in health facilities in small settlements of developing countries, and the limitations of their use for daily assessment due to the risk of repeated radiation dose, greatly limit their application. In response to the search for safe, simple, rapid, non-invasive, and cost-effective promising alternatives for the diagnosis of COVID-19, researchers in the field are increasingly turning to the analysis of human respiratory sound signals, including cough, breathing, and voice sounds. This is due to the direct connection of the respiratory sound signals with the lungs. Despite the detection efficiency obtained in earlier related works, further studies are still needed on the ability of breath sounds to provide meaningful information about COVID-19. This study used 2660 samples of cough sounds (1 330 recordings from healthy subjects and 1 330 recordings from subjects infected with COVID-19) from the CoughVid dataset, to train models for the classification of the COVID-19 disease. An attempt has been made to classify COVID-19 using different machine-learning models. Temporal and spectral features were extracted from the amplitude spectrum of cough sound signals, and evaluated using a periodogram, and those with higher discriminative power were selected. 1862 cough sound recordings were used for training and 798 cough sound recordings were used to test the model. On the test set, the final optimized model achieved classification accuracy, sensitivity, and specificity of 97.87%, 97.90%, and 97.85%, respectively. The experimental results of the study showed that the proposed method provides significant accuracy for classifying the COVID-19 disease, making it a reliable decision-support tool in healthcare settings where reverse transcription polymerase chain reaction is not available and test kits are scarce.
Keywords: Coronavirus, COVID-19, Feature extraction, Hyperparameter optimization | |
| Empirical Study on Myopia Identification Using CNN Hereditary Model for Resource Constrained Ophthalmology | 19-32 |
| Aqila Nazifa, Manisha Shivaram Joshi, Soumya Ramani | |
| doi: 10.7546/ijba.2025.29.1.000961 | |
| [ +/- how to cite ][ +/- abstract ][ full text ] | |
| Aqila Nazifa, Manisha Shivaram Joshi, Soumya Ramani (2025) Empirical Study on Myopia Identification Using CNN Hereditary Model for Resource Constrained Ophthalmology, Int J Bioautomation, 29 (1), 19-32, doi: 10.7546/ijba.2025.29.1.000961 | |
| Abstract: Refractive errors, which include myopia, hyperopia, presbyopia, and astigmatism, are common vision problems that result in blurred vision when light rays are not focused correctly on the retinal plane. Diagnosis and classification of refractive errors are essential for providing appropriate corrective measures such as glasses or contact lenses. The key objective of this research is to establish an efficient and fast approach to identifying a refractive defect and categorizing them. Leveraging the capabilities of modern technology, we utilize a smartphone’s camera to capture pictures of the red reflex in the eye. During capturing, the photos are processed using recent image processing techniques to identify any irregularities or asymmetries that may indicate refractive errors. By comparing our method to other current models, we hope to illustrate the advantage of our Hereditary model, which combines a random forest and a convolutional neural network, in accurately diagnosing and classifying refractive errors. Additionally, the proposed approach can serve as a foundation in order to do additional research and development in machine learning and image processing methods improvements for the classification of ocular disorders.
Keywords: Refractive error, Myopia, Red reflex, Image processing, Machine learning, Hereditary model | |
| Deep Learning Classification of Simulated Surface EMG Signals across Maximum Voluntary Contraction Levels | 33-50 |
| Radhouane Hammachi, Samia Belkacem, Noureddine Messaoudi, Raïs El’hadi Bekka | |
| doi: 10.7546/ijba.2025.29.1.000988 | |
| [ +/- how to cite ][ +/- abstract ][ full text ] | |
| Radhouane Hammachi, Samia Belkacem, Noureddine Messaoudi, Raïs El’hadi Bekka (2025) Deep Learning Classification of Simulated Surface EMG Signals across Maximum Voluntary Contraction Levels, Int J Bioautomation, 29 (1), 33-50, doi: 10.7546/ijba.2025.29.1.000988 | |
| Abstract: Electromyography (EMG) is a fundamental tool in diagnosing neuromuscular disorders (NMD). Due to the complex nature of EMG signals, different approaches, based on artificial intelligence and machine learning, were developed for EMG signal analysis and NMD diagnosis. Considering the critical role of maximum voluntary contraction (MVC) as a fundamental metric in assessing muscle fatigue, in this work, classification of simulated surface EMG (sEMG) into MVC levels is performed. Unlike previous studies, which focus primarily on binary classification of fatigue and non-fatigue states, our approach employs a deep convolutional neural network for the classification of sEMG signals into ten MVC levels, where the model outputs categorical predictions, with each class representing a specific MVC level. sEMG signals were generated using a computer muscle model that we developed using MATLAB, which allows for greater control over variability, ensuring robustness and generalizability of the model. The obtained results demonstrate that the model achieved high performance in differentiating between the ten classes (MVC levels), with an accuracy, F1-score, recall, and precision of 88.88%, 88.75%, 88.80% and 88.86%, respectively. These findings reveal that the model can accurately differentiate across MVC levels, indicating a potential method for accurate assessment of muscle fatigue intensity.
Keywords: Artificial intelligence (AI), Diagnosis, Electromyography (EMG), Muscle fatigue | |
| Exploring Equilibrium Points in a Long-term Glucose-insulin Model for Type I Diabetes: MPC Application in Automated Insulin Delivery Systems Using Functional Insulin Therapy Tools | 51-76 |
| Amor Hamata, Salim Aissi | |
| doi: 10.7546/ijba.2025.29.1.000990 | |
| [ +/- how to cite ][ +/- abstract ][ full text ] | |
| Amor Hamata, Salim Aissi (2025) Exploring Equilibrium Points in a Long-term Glucose-insulin Model for Type I Diabetes: MPC Application in Automated Insulin Delivery Systems Using Functional Insulin Therapy Tools, Int J Bioautomation, 29 (1), 51-76, doi: 10.7546/ijba.2025.29.1.000990 | |
| Abstract: This study explores a novel approach to regulate blood glucose levels in individuals with type I diabetes, employing the widely used model predictive control (MPC) strategy in type 1 diabetes mellitus therapy and clinical trials. The MPC algorithm is implemented based on Magdelaine’s long-term glucose-insulin model, which encompasses real-life characteristics often absent in other prevalent models. The control strategy is evaluated through simulations involving 10 virtual patients from existing literature. The simulations encompass fasting scenarios and a closed-loop control scenario involving three meals. MPC results are compared to those of the “optimal” conventional insulin daily injections therapy (open-loop treatment), especially under “aggressive conditions” including elevated initial blood glucose levels, substantial carbohydrate intake, closely spaced meal times, and incorporating a time delay between plasma glucose concentration and its subcutaneous measurement. The MPC algorithm demonstrated remarkable efficacy in glycemic control for 80% of patients, achieving an average time-in-range percentage exceeding 80% with no hypoglycemic episodes. This aligns with the American Diabetes Association’s recommendation of spending at least 70% of the time in the target range for effective glycemic control and maintaining an average time spent in hypoglycemia of less than 4%. However, the same MPC controller exhibited suboptimal performance for two patients, with an average time spent in hypoglycemia exceeding 8%. These findings underscore the need for individualized adjustments of MPC parameters or alternative control strategies to optimize glycemic management in all patients.
Keywords: Type I diabetes, Equilibrium points, Open-loop therapy, Closed-loop control therapy, Model predictive control, Functional insulin therapy tools, Time in range | |
| Detection and Classification of Diabetic Retinopathy Using Modified Inception V3 | 77-92 |
| Abini M. A., S. Sridevi Sathya Priya | |
| doi: 10.7546/ijba.2025.29.1.001004 | |
| [ +/- how to cite ][ +/- abstract ][ full text ] | |
| Abini M. A., S. Sridevi Sathya Priya (2025) Detection and Classification of Diabetic Retinopathy Using Modified Inception V3, Int J Bioautomation, 29 (1), 77-92, doi: 10.7546/ijba.2025.29.1.001004 | |
| Abstract: In the last decade, the prevalence of diabetic retinopathy (DR), a sight-threatening medical condition of diabetes mellitus, has markedly increased, impacting millions globally. The conventional method of diagnosing and classifying this condition was done through physical and detailed examination of fundus images by ophthalmologists, a process prone to human error and time-consuming. To overcome this challenge, artificial intelligence, particularly deep learning algorithms, has taken a position in automating the diagnosis of diabetic eye disease and categorization from fundus images. Various studies have confirmed the effectiveness of convolutional neural networks in this task, with Inception V3 emerging as a particularly successful architecture. In our current work, we adduce a novel approach for DR detection and categorizing it utilizing the Inception V3 architecture on fundus images. The learning rate is modified between le-3, le-4, le-5, and le-6 to investigate various optimization approaches. Our model, trained on the Asia Pacific Tele Ophthalmology Society datasets, accomplished an accuracy of 91.64% for a learning rate of le-5, which outperforms current approaches in diagnosing the five phases of DR.
Keywords: Deep learning, Convolutional neural networks, Retina images, Inception V3, Diabetic retinopathy | |
| Why Wind Energy? | 93-99 |
| Ivan Popchev | |
| doi: 10.7546/ijba.2025.29.1.001056 | |
| [ +/- how to cite ][ +/- abstract ][ full text ] | |
| Ivan Popchev (2025) Why Wind Energy?, Int J Bioautomation, 29 (1), 93-99, doi: 10.7546/ijba.2025.29.1.001056 | |
| Abstract:
Keywords: Book Review |
Sponsored by National Science Fund of Bulgaria, Grant No КП-06-НП6-14, 2025
© 2025, BAS, Institute of Biophysics and Biomedical Engineering