Journal Articles
Permanent URI for this collection
Browse
Recent Submissions
1 - 5 of 185
-
PublicationAI-optimized electrochemical aptasensors for stable, reproducible detection of neurodegenerative diseases, cancer, and coronavirus(Elsevier, 2025)AI-optimized electrochemical aptasensors are transforming diagnostic testing by offering high sensitivity, selectivity, and rapid response times. Leveraging data-driven AI techniques, these sensors provide a non-invasive, cost-effective alternative to traditional methods, with applications in detecting molecular biomarkers for neurodegenerative diseases, cancer, and coronavirus. The performance metrics outlined in the comparative table illustrate the significant advancements enabled by AI integration. Sensitivity increases from 60 to 75 % in ordinary aptasensors to 85–95 %, while specificity improves from 70-80 % to 90–98 %. This enhanced performance allows for ultra-low detection limits, such as 10 fM for carcinoembryonic antigen (CEA) and 20 fM for mucin-1 (MUC1) using Electrochemical Impedance Spectroscopy (EIS), and 1 pM for prostate-specific antigen (PSA) with Differential Pulse Voltammetry (DPV). Similarly, Square Wave Voltammetry (SWV) and potentiometric sensors have detected alpha-fetoprotein (AFP) at 5 fM and epithelial cell adhesion molecule (EpCAM) at 100 fM, respectively. AI integration also enhances reproducibility, reduces false positives and negatives (from 15-20 % to 5–10 %), and significantly decreases response times (from 10-15 s to 2–3 s). These advancements improve data processing speeds (from 10 to 20 min per sample to 2–5 min) and calibration accuracy (<2 % margin of error compared to 5–10 %), while expanding application scope to multi-target biomarker detection. This review highlights how these advancements position AI-optimized electrochemical aptasensors as powerful tools for personalized treatment, point-of-care testing, and continuous health monitoring. Despite a higher cost ($500-$1,500/unit), their enhanced portability and diagnostic performance promise to revolutionize healthcare, environmental monitoring, and food safety, ultimately improving public health outcomes.
-
PublicationEEG signal processing using deep learning for motor imagery tasks: Leveraging signal images(Springer, 2025)A novel approach to processing electroencephalography (EEG) signals has emerged, leveraging the utilization of signal images. The application of deep learning techniques in bypassing complex signal and image processing tasks has generated significant interest in this field. However, challenges remain in signal image processing, particularly in handling significant features and image sizes. This study presents a comprehensive investigation of EEG motor imagery signal processing, focusing on the classification of three tasks: eating, drinking, and seeking assistance. Fast Fourier Transform (FFT) is employed to extract signal image features, which are subsequently utilized in a deep learning framework. EEG data were collected from five subjects, and four transfer functions of deep learning models, namely VGG16, VGG19, ResNet50, and ResNet101, were employed for training and classification purposes. The performance of the four models was meticulously evaluated and compared. Notably, VGG16 exhibited superior performance in accurately classifying the EEG motor imagery tasks, achieving an impressive accuracy of 90%, sensitivity of 84%, and specificity of 92%. In conclusion, this study underscores the efficacy of EEG signal image processing through deep learning-based classification techniques. The findings highlight the potential of utilizing signal images in EEG analysis for motor imagery tasks, thereby contributing to the advancement of brain-computer interface technology and enhancing our understanding of neural dynamics.
-
PublicationA wearable 3D printed microfluidic device for sweat-sensing application(Springer Science and Business Media, 2024-12)This study focuses on developing a wearable microfluidic device (WMD) using stereolithography (SLA) 3D printing for sweat collection. The use of the SLA technique, particularly in achieving rapid fabrication, printing smooth surfaces, and creation of channels with dimensions below 1 mm. However, it is quite challenging to integrate the SLA 3D printed WMD with a sensor for real-time sweat analysis using a traditional bonding method. In addition, an SLA conventional resin is non-water-washable and is made from a polymer material that tends to cause a hydrophobic effect on the microchannel surface. In this work, a reversible bonding method through mechanical clamping was applied to enable easy assembly and disassembly of the WMD integrated with a sensor. A water-washable clear resin was used to provide a hydrophilic surface, allowing for effective fluid handling. The fluid delivery into the sensor's channel was efficient, taking only 0.06 s after the fluid flowed out at the outlet channel, and it sufficiently covered the entire surface of the sensor. This work also found that closed channels can be created up to 0.6 mm after fine-tuning the minimum achievable using the SLA printer. The dimensions of the printed WMD resulted in a size tolerance difference of 0.05–0.35 mm compared to the 3D model design, indicating a discrepancy of less than 1%. These capabilities promise to advance WMD and enable cutting-edge research in sweat analysis and related fields.
-
PublicationDetermination of probability density, position and momentum uncertainties, and information theoretic measures using a class of inversely quadratic Yukawa potential(Nature Research, 2025)This study utilizes the Nikiforov-Uvarov method to solve the Schrödinger equation for the class of inversely quadratic Yukawa potential (CIQYP), deriving both the energy equation and the normalized wave function. Shannon entropy and Fisher information in both position and momentum spaces are analyzed for low-energy states using the wave function. The Bialynicki-Birula-Mycielski and Stam-Cramer-Rao inequalities are satisfied for the Shannon and Fisher information entropies, illustrating the complementary uncertainties inherent in position and momentum in quantum mechanics. The study underscores the interplay between position and momentum Fisher entropies, reinforcing the Heisenberg uncertainty principle, which imposes limits on the precise simultaneous measurement of conjugate variables. Eigenvalues of the CIQYP for three diatomic molecules (N₂, O₂, and NO) are obtained using their respective data, revealing that the bound state energy spectra of these diatomic molecules increase as both the principal quantum number and angular momentum quantum number rise. Expectation values were numerically determined, and the potential model simplifies to the Kratzer potential under specific boundary conditions, thereby ensuring analytical accuracy. The energy spectra of diatomic molecules such as I₂ and CO are examined, showing that for a fixed principal quantum number, the energy spectrum increases with increasing angular momentum quantum number, in very good agreement with previously obtained results using different analytical methods.
-
PublicationAI-powered MMI fiber sensors for wide-range refractive index detection using neural networks algorithm(Elsevier, 2025-03)This research presents an artificial intelligence (AI)-driven machine learning (ML) approach for accurately measuring refractive index (RI) values across both lower and higher regimes than the fiber material's RI, using a simple single multimode interference (MMI) fiber sensor. The sensor configuration consists of a no-core fiber (NCF) segment between two single-mode fiber (SMF) sections. A Bilayer Neural Network (BNN) regression model is employed to predict both low refractive index (LRI) and high refractive index (HRI) regimes, achieving a broad dynamic measurement range from 1.3000 RIU to 1.3900 RIU for LRI regime and from 1.4600 RIU to 1.5500 RIU for HRI regime. The model demonstrates 99.7% accuracy and a low root mean square error (RMSE) of 0.0044, ensuring that predicted RI values closely match actual measurements without any RI ambiguity. Furthermore, the all-silica NCF structure is inherently resistant to temperature fluctuations, enabling its deployment in environments with varying temperatures without requiring additional temperature compensation mechanisms.