Limited Accuracy Research Articles

Efficient indoor localization by integrating RFID with adaptive emperor penguin colony-based deep Q learning

ABSTRACT Radio Frequency Identification (RFID) is exploited for localizing objects in an indoor environment. Indoor localization is a critical component of various applications, ranging from smart homes to industrial automation. Traditional methods often suffer from limitations in accuracy and scalability. The existing models found it complex to develop a map approach of the tag received signal strength indicator (RSSI) and the distance of the tag and reader in an indoor nature. This work presents an enhanced model that integrates RFID technology using deep Q learning (DQL) and adaptive emperor penguin colony optimizer (AEPCO) for improving the relationship between tag position and RFID signals. By utilizing RFID tags and readers, spatial data are collected and then processed through a proposed DQL-AEPCO designed to accurately estimate positions within indoor environments. Moreover, for enhancing the training quality, the dataset pre-processed using Gaussian filtering (GF) is presented for eliminating RSSI faults. The demonstration proved that the suggested DQL-AEPCO can able to locate the multi-tags with high stability and robustness and outperformed the conventional models.

Glomerular Filtration Rate Measurement Utilizing Transdermal Detection Methodology.

The well-known accuracy limitations of estimated GFR (eGFR) currently employed in clinical practice present barriers to optimal care for patients with, or at-risk for, decreased kidney function. A point-of-care glomerular filtration rate (GFR) measurement methodology has the potential to address these limitations. We prospectively assessed transdermal detection of the novel fluorescent GFR tracer agent relmapirazin in participants having normal or impaired kidney function across all human skin colors on the Fitzpatrick Skin Scale (FSS). A multi-center study comprising 74 participants with eGFR from normal to Stage 4 CKD was performed. Forty-six participants were FSS types I-III, and twenty-eight were FSS type IV-VI. A module containing an LED and photodetector to activate and collect transdermal relmapirazin fluorescence was attached adhesively to the upper chest of each participant. Relmapirazin (1.5mg/kg) was administered by intravenous push, and fluorescence emission was acquired for 12 hours. A two-compartment pharmacokinetic model fit the fluorescent intensity vs time data, and the fluorescence clearance rate (FCR) was extracted from the second (terminal) compartment. Plasma relmapirazin concentrations were measured contemporaneously and the corresponding plasma GFR for each participant was determined. Linear regression analysis was used to compare the FCR to the indexed plasma GFR. Participant age range was 23 to 80 years old, with 59% females. The two-compartment pharmacokinetic behavior was observed in the fluorescence intensity vs time data and a FCR was successfully deduced for every participant completing the 12-hour study. The FCR vs. the indexed plasma GFR yielded an excellent correlation over the range of GFR measured and for all skin colors with a r2 = 0.90 (95% confidence interval 0.85 to 0.94). No severe adverse events were reported. Point-of-care transdermal detection of the fluorescent GFR tracer agent relmapirazin was feasible in patients with normal to impaired kidney function and for a range of skin color types.

Current Status of Prostate Biopsy in Africa: Are We Ready for a Targeted Biopsy?

Prostate cancer remains a significant public health issue globally, with considerable disparities in diagnostic and management practices, especially in Africa. Traditional diagnostic methods such as transrectal ultrasound-guided biopsies have limitations in accuracy and are associated with potential complications. Emerging targeted biopsy techniques promise improved cancer detection rates and reduced morbidity but face adoption challenges across the African continent due to variable access to advanced imaging technologies and professional expertise. This study aims to evaluate the current practices of prostate biopsy techniques in Africa, focusing particularly on the readiness for and integration of targeted biopsy methods. It assesses the accessibility, quality, volume of procedures, and availability of advanced diagnostic tools across different regions. The survey included 58 centers across five African regions, representing a 70% response rate from an initial 83 invitations sent. Finger-guided biopsy was the most common method, used by 41 centers, while only one center reported using MRI fusion biopsy. Thirty-five centers routinely administered a rectal enema, and nearly all (n = 50) centers employed antibiotic prophylaxis, predominantly fluoroquinolones. For anesthesia, 12 centers used injectable lidocaine, and 38 centers used intrarectal lidocaine gel. The number of biopsy cores taken varied, with most centers taking 12, while others used fewer. The findings indicate a critical need for concerted efforts to bridge the gap in prostate cancer diagnostics and treatment in Africa. Enhancing the quality of prostate cancer care on the continent requires investments in training, infrastructure, and standardization of practices. Collaborative efforts towards adopting advanced diagnostic tools and methods are essential for aligning African practices with global standards, ultimately improving outcomes for prostate cancer patients.

Quantification of mycophenolic acid in plasma by isotope dilution liquid chromatography-tandem mass spectrometry candidate reference method.

Accurate measurements of plasma mycophenolic acid (MPA) are essential for therapeutic drug monitoring in transplant recipients and autoimmune diseases. The performance of plasma mycophenolic acid routine methods remains highly variable that calls for a candidate reference measurement procedure (cRMP) to improve the standardization of plasma mycophenolic acid measurements. In this study, sample preparation was based on protein precipitation with methanol followed by further dilution. The mycophenolic acid was quantified by the isotope dilution liquid chromatography-tandem mass spectrometry (ID-LC-MS/MS) with electrospray ionization in positive ion mode. According to the Clinical and Laboratory Standards Institute (CLSI) documents C62-A and C50-A, the basic analytical performance of the candidate reference method was verified, such as linearity, limit of quantification, matrix effect, precision, accuracy, and uncertainty. Moreover, the candidate reference measurement procedure was compared with the routine liquid chromatography-tandem mass spectrometry (LC-MS/MS) method in a clinical laboratory. Based on the data, the mycophenolic acid in human plasma was well detected by ID-LC-MS/MS. No apparent interferences were found with the mycophenolic acid measurement. The calibration curve for the mycophenolic acid was linear in the concentration range of 0.1-50μg/mL with a correlation coefficient of 0.9999 under the optimum experimental conditions. This method was sensitive because the low limit of quantitation (LOQ) was 0.05μg/mL. The recoveries of MPA were 98.11-98.95%. The intra-day and inter-day coefficients of variations (CV) of our method were ≤ 1.53% and ≤ 0.51%, respectively. No obvious matrix effect was observed. There was a good correlation between this method and the clinical routine LC-MS/MS method. To sum up, we established and validated a reliable plasma MPA method using ID-LC/MS/MS. The desirable accuracy and precision of this method enable it to serve as a promising cRMP to improve the standardization for plasma MPA routine measurements.

Machine learning of the dynamics of strain hardening based on contact transformations

Abstract Dislocation density-based models offer a physically grounded approach to modeling strain hardening in metal forming. Since these models are typically defined by Ordinary Differential Equations (ODEs), their accuracy is constrained by both, the model formulation and the parameter identification process. Machine Learning (ML) provides an alternative by allowing models to be constructed directly from experimental data, bypassing the accuracy limitations of explicitly defined models. However, applying ML to ODEs introduces the need for novel training techniques. This work presents a new approach for developing neural ODE models for flow curve description, utilizing a contact transformation to simplify the problem of learning an ODE into a learning a multivariate function. Graphical abstract

A Functionalized 3D-Printed Ti6Al4V "Cell Climbing Frame" Inspired by Marine Sponges to Recruit and Rejuvenate Autologous BMSCs in Osteoporotic Bone Repair.

Osteoporosis, characterized by low bone mass and high fracture risk, challenges orthopedic implant design. Conventional 3D-printed Ti6Al4V scaffolds are mechanically robust but suffer from poor bone regeneration in osteoporotic patients due to stress shielding and cellular senescence. In this study, a functionalized 3D-printed Ti6Al4V "Cell Climbing Frame" is developed, aiming to adapt to the mechanical microenvironment of osteoporosis, effectively recruit and support the adhesion and growth of autologous bone marrow mesenchymal stem cells (BMSCs), while rejuvenating senescent cells for improved bone regeneration. Inspired by marine sponges, the processing accuracy limitations of selective laser melting (SLM) technology is broke through innovatively constructing a hierarchical porous structure with macropores and micropores nested within each other. Results demonstrate that the unique hierarchical porous scaffold reduces the elastic modulus, facilitates blood penetration, and enhances cell adhesion and growth. Further surface functionalization with E7 peptides and exosomes promotes the attraction and rejuvenation of BMSCs and boosts migration, proliferation, and osteogenic differentiation in vitro. In vivo, the functionalized "Cell Climbing Frame" accelerates bone repair in osteoporotic rats, while delaying surrounding bone loss, enabling robust multi-stage osseointegration. This innovation advances 3D-printed regenerative implants for osteoporotic bone repair.

Auditory steady state response can predict declining EF in healthy elderly individuals

BackgroundThe aging population imposes significant economic and societal challenges, underscoring the need for early detection of individuals at risk of cognitive decline prior to the onset of clinical symptoms. This study explores the association between gamma-band Auditory Steady-State Responses (ASSRs) and subclinical cognitive decline using longitudinal data from healthy volunteers in the Metropolit Birth Cohort (MBC).MethodsLongitudinal recordings of cognitive test results and ASSRs at 40 Hz stimulation were analyzed. Generalized Linear Models (GLMs) were employed to determine the association between ASSR characteristics and cognitive performance with an emphasis on Executive Function (EF) at ages 61–68. Additionally, Vision Transformers (ViTs) were trained to distinguish between individuals with declining and stable cognitive performance.ResultsSubjects with declining cognitive performance through midlife showed a larger area of entrainment and delayed neural assembly of ASSRs compared to those with stable cognitive performance. These neurophysiological changes were correlated with poorer EF, as measured by the Stockings of Cambridge (SOC) task. The ViTs trained and cross-validated on time-frequency-transformed Electroencephalograms (EEGs) achieved an average cross-subject accuracy of 51.8% in identifying cognitive decline.ConclusionGamma-band ASSR characteristics are linked to early cognitive decline in middle-aged individuals, offering potential as biomarkers. However, the limited predictive accuracy of ML models emphasizes the need for further refinement to enhance their clinical applicability.

Performance of Hybrid Strategies Combining MDockPP and AlphaFold2 in CAPRI Rounds 47-55.

CAPRI challenges offer a range of blind tests for biomolecule interaction prediction. This study evaluates the performance of our prediction protocols for the human group Zou and the server group MDockPP in CAPRI rounds 47-55, highlighting the impact of AlphaFold2 (AF2) and the effectiveness of massive sampling approaches. Prior to AlphaFold2's release, our methods relied on homology modeling and docking-based protocols, achieving limited accuracy due to constraints in structural templates and inherent docking limitations. After AlphaFold2's public release, which demonstrated breakthrough accuracy in protein structure prediction, we integrated its multimer models and massive sampling techniques into our protocols. This integration significantly improved prediction accuracy, with human predictions increasing from 1 correct interface of 19 pre-AlphaFold2 to 4 of 8 post-AlphaFold2. The massive sampling approach further enhanced performance, particularly for targets T231 and T233, yielding medium-quality models that default parameters could not achieve.

Shedding light on development: Leveraging the new nightlights data to measure economic progress.

Nightlights (NTL) have been widely used as a proxy for economic activity, despite known limitations in accuracy and comparability, particularly with outdated Defense Meteorological Satellite Program (DMSP) data. The emergence of newer and more precise Visible Infrared Imaging Radiometer Suite (VIIRS) data offers potential, yet challenges persist due to temporal and spatial disparities between the two datasets. Addressing this, we employ a novel harmonized NTL dataset (VIIRS + DMSP), which provides the longest and most consistent database available to date. We evaluate the association between newly available harmonized NTL data and various indicators of economic activity at the subnational level across 34 countries in sub-Saharan Africa from 2004 to 2019. Specifically, we analyze the accuracy of the new NTL data in predicting socio-economic outcomes obtained from two sources: 1) nationally representative surveys, i.e., the household Wealth Index published by Demographic and Health Surveys, and 2) indicators derived from administrative records such as the gridded Human Development Index and Gross Domestic Product per capita. Our findings suggest that even after controlling for population density, the harmonized NTL remain a strong predictor of the wealth index. However, while urban areas show a notable association between harmonized NTL and the wealth index, this relationship is less pronounced in rural areas. Furthermore, we observe that NTL can also significantly explain variations in both GDP per capita and HDI at subnational levels.

Ultrasonography in the diagnosis of acute pancreatitis keeping computed tomography as a gold standard: Multi-center study.

Acute pancreatitis is considered as a major risk towards the health of the patient as the sudden incidence of the inflammation in the pancreas develops sufficient impacts on the physiological functionality of the adjacent tissues or the organs. Study Design: Cross-sectional study. Setting: PAF Hospital Islamabad. Period: November 2021 to February 2022. Methods: This study, conducted at PAF Hospital Islamabad, evaluated the diagnostic efficacy of ultrasound and CT imaging in 150 patients with acute pancreatitis. Following ethical approval and informed consent, patients aged 15-70 years with serum amylase levels above 400 U/L were included. Ultrasounds and CT scans were performed by experienced radiologists, and data were analyzed using MS Excel and SPSS. The study assessed sensitivity, specificity, predictive values, and diagnostic accuracy of both imaging modalities. Results were stratified by age, gender, BMI, and disease duration, with true positives and negatives identified based on consistent imaging results, ensuring a thorough evaluation of imaging effectiveness for acute pancreatitis diagnosis. Results: Patients were predominantly aged 36-55 (127 patients) and male (115 patients). USG showed limited diagnostic accuracy (37.6%) with a sensitivity of 41% and specificity of 35.1%. The study evaluated the diagnostic accuracy of ultrasound (USG) compared to CT scans in identifying acute pancreatitis. The positive predictive value (PPV) was 35%, and the negative predictive value (NPV) was 41%. With an overall diagnostic accuracy of 37.6%, the findings highlight USG’s limited reliability for diagnosing acute pancreatitis when compared to CT. The significance value was 0.072, influenced by a high rate of false positives. The results underscore the varying reliability of USG across different demographics. Conclusion: Ultrasonography (USG) shows limited reliability in diagnosing acute pancreatitis compared to CT, with a sensitivity of 41%, specificity of 35.1%, and overall accuracy of 37.6%. The high rate of false positives highlights that while USG can be a preliminary tool, CT remains essential for accurate diagnosis and effective clinical management.

Predictive diagnosis of El Niño based on empirical sea level anomaly forecasts

Abstract The aim of this paper is to present the skills of two statistical models in anticipating the development of El Niño based on sea level anomaly (SLA) forecasts with lead time up to 12 weeks. The models are: (1) the polynomial-harmonic model (PH) combined with the threshold autoregressive model (TAR), known as the PH+TAR, and (2) PH integrated with the multivariate autoregressive model (MAR), referred to as PH+MAR. Five powerful El Niño events are considered: 1997/1998, 2002/2003, 2006/2007, 2009/2010, 2015/2016. The performance of the prediction models is calculated in specific locations in the equatorial Pacific, i.e. centres of Niño 1+2, Niño 3, Niño 3.4 and Niño 4 regions. It is found that the SLA predictions hitting the El Niño peaks reveal different accuracy for dissimilar El Niño events, with the most skillful prognoses for El Niño 1997/1998. Two specific regions are identified in which the model performance fulfils the assumed accuracy limit of 5 cm, namely the Niño 1+2 and Niño 4 regions. In addition, the PH+MAR model performed better than the PH+TAR solution.

Force-Field Benchmark for Polydimethylsiloxane: Density, Heat Capacity, Isothermal Compressibility, Viscosity and Thermal Conductivity.

Polysiloxanes are versatile polymeric materials with widespread applications in industries ranging from electronics to biomedical devices because of their unique thermal and viscoelastic properties. Accurate molecular simulations of polysiloxanes are essential for understanding their broad applications from a microscopic perspective. However, the accuracy of these simulations is highly dependent on the quality of the force fields used. In this work, we present a comprehensive benchmark and development of force fields tailored for polydimethylsiloxane, which is one of the most widely used polysiloxane materials. Our focus is on their performance in predicting key thermophysical properties including density, heat capacities, isothermal compressibility, and transport properties such as viscosity and thermal conductivity. Experimental measurements are performed in parallel to rigorously validate simulation outcomes. Existing force fields for polydimethylsiloxane, including those derived for organic and inorganic systems, are systematically evaluated against experimental data to identify limitations in accuracy and transferability. Simulation results are compared extensively with experimental observations across a range of temperatures and pressures, revealing the strengths and shortcomings of these commonly utilized force fields for polydimethylsiloxane. Discrepancies between force field predictions and experimental measurements are analyzed in detail for thermodynamic and transport properties of polydimethylsiloxane. This benchmark study serves as a critical assessment of current force fields for polydimethylsiloxane and offers guidelines for their further development, enabling more reliable simulations of polysiloxane-based materials for various industrial applications.

SBCP-YOLO-R3D: Student Behavior Recognition and Visualization Framework Using Improved YOLO and R3D for Class Video

Real-time recognition and visualization of students' behaviors in face-to-face classrooms serve as pivotal indicators of learning engagement. However, current methods exhibit limitations in both real-time performance and accuracy. Additionally, in-depth studies have not been extensively conducted to evaluate learning status more conveniently by utilizing computer vision techniques. To address these issues, a novel Student Behavior Recognition and Dynamic Class Portraits Construction framework, named SBCP-YOLO-R3D, incorporating the StB-YOLO and R3D methods, has been proposed to detect student behaviors and construct class portraits. The developed framework comprises two layers: the StB-YOLO detection layer and the R3D classification layer. In the StB-YOLO detection layer, the Lightweight-SEAM (LW-SEAM) is incorporated into YOLOv5 to enhance the recognition of occluded students, by capturing contextual information and enhancing occlusion-related features. Moreover, a Double-SlideLoss function is devised, employing adaptive weighting mechanisms to strike an optimal balance between simple and challenging samples. In the R3D classification layer, the results generated by StB-YOLO are then processed using R3D to produce class portraits. Experiments conducted on the StuAct and SCB-DATASET3-S datasets demonstrate the effectiveness of the StB-YOLO. Compared with the baseline model, StB-YOLO increases the mAP by 3.1%.

Limited Accuracy and Utility of Surgeon’s Estimate of Gross Total Resection following Endoscopic Endonasal Transsphenoidal Surgery for Non-Hormone Producing Pituitary Macroadenomas

Limited Accuracy and Utility of Surgeon’s Estimate of Gross Total Resection following Endoscopic Endonasal Transsphenoidal Surgery for Non-Hormone Producing Pituitary Macroadenomas

Optimal design dimension configuration of high voltage electrodes for 400 KV bus post insulators

ABSTRACT The performance, reliability, and safety of 400 kV bus post insulators remain dependent on the design of high-potential electrodes, such as bus conductors, grading rings, and corona rings. Reduced localised stress and insulation failure are two benefits of proper dimensioning, which also helps regulate electric field stress along the insulator string. To mitigate these effects, corona ring dimensions must be assessed, but practical testing is expensive and challenging. Comprehensive knowledge of dielectric field stress analysis for these insulators is complex and costly to obtain. As a result, design engineers often rely on computer models that have been validated through experimental testing. While they are used to study field stress, traditional 2D approaches such as the Finite Element aprroache (FEM) and Boundary Element approach (BEM) have accuracy limits. In this study, a 400 kV porcelain bus post insulator was constructed using 3D simulation software using industry-standard design dimensions. The standard dimension configuration was compared with a various dimension configurations. The electrical simulations were carriedout using the COULOMB 3D tool, which employs BEM for accuracy and efficiency. Various case study dimensions are carried out and presented for deep field stress analysis of proposed insulator performance, which is major lacuna in experimental testing facility.

Predicting photovoltaic greenhouse irradiance at low-latitudes of plateau based on ultra-short-term time series

Accurate and reliable ultra-short-term prediction of solar irradiance in photovoltaic (PV) greenhouses at low-latitude plateau is essential to precisely control electricity consumption of greenhouse equipment and ensure high quality crop yields. However, the irradiance in the low-latitude plateau has problems such as poor data quality, limited short-term prediction accuracy, and insufficient ability to capture nonlinear characteristics. Therefore, in order to achieve efficient utilization of photovoltaic resources, this study proposed a new hybrid integrated model TTAO-CNN-BiGRU-Attention framework to predict ultra-short-term photovoltaic greenhouse irradiance in the region. Monthly and seasonal characteristics of irradiance in low-latitude plateau areas were analyzed by statistical methods. The performance of the proposed model was verified using 9 different models for 5 different data volumes and 4 different seasons. Comprehensive analysis results show that Total radiant instantaneous(TRI) demonstrates a seasonal trend, generally low in spring, high in summer and autumn, relatively stable in autumn and winter. The monthly trend initially increases and then decreases, reaching the highest value of the year in September. The scheme proposed in this paper makes full use of the advantages of CNN, BiGRU, Attention and TTAO, greatly improving the comprehensive prediction ability of the model. In predicting different data amounts, 1 year prediction performance was the best, with RMSE, MAE, MAPE and R2 reaching 70.61 W/m2, 31 W/m2, 9.3% and 95.84%, respectively. With regard to different seasons, autumn prediction performance was the best, with RMSE, MAE, MAPE and R2 reaching 66.27 W/m2, 31.02 W/m2, 8.37% and 95.87%, respectively. The TRI prediction curve of the proposed model was closer to the actual value than other comparison models. The study found that the TTAO-CNN-BiGRU-Attention model is more accurate and stable than many traditional models in predicting ultra-short-term TRI in low-latitude plateau photovoltaic greenhouses, which can provide a reference for the comprehensive performance of PV greenhouse irradiance prediction models and precise regulation of energy supply in the future.

Evaluating machine learning models for predictive analytics of liver disease detection using healthcare big data

Liver diseases rank among the most prevalent health issues globally, causing significant morbidity and mortality. Early detection of liver diseases allows for timely intervention, which can prevent the progression of such diseases to more severe stages such as cirrhosis or liver cancer. To this end, many machine learning models have been previously developed to early predict liver diseases among potential patients. However, each model has its accuracy and performance limitations. In this paper, we present a comprehensive comparison of three different machine learning models that can be employed to enhance the prediction and management of liver diseases. We utilize a big data set of 32,000 records to evaluate the performance of each model. First, we implement a preprocessing technique to rectify missing or corrupt data in liver disease datasets, ensuring data integrity. Afterwards, we compare the performance of three machine models: k-nearest neighbors (KNN), gaussian naive Bayes (Gaussian NB) and random forest (RF). We concluded that the RF algorithm demonstrates superior performance in our evaluation, excelling in both predictive accuracy and the ability to classify patients accurately regarding the presence of liver disease. Our results show that RF outperforms other models based on several performance metrics including accuracy: 97.3%, precision: 97%, recall: 96%, and f1-score: 95%.

Research on failure diagnosis analysis of plunger gas lift system using convolutional neural network with multi-scale channel attention mechanism based on wavelet transform

Plunger gas lift technology has been extensively utilized in unconventional gas fields, owing to its distinct engineering benefits. However, as development challenges intensify, fault diagnosis has grown more intricate, and conventional diagnostic techniques exhibit delays and inaccuracies. With advancements in computer science, machine learning methods have demonstrated their prowess in establishing robust correlations between data features and prediction outcomes. Consequently, this paper introduces a convolutional neural network fault diagnosis model that incorporates a multi-scale channel attention mechanism based on wavelet transform. This model dissects features across various scales using wavelet transform and leverages channel attention to adaptively select channels encompassing fault features, thereby enhancing diagnostic and recognition accuracy. Furthermore, the model integrates a hyperparameter search optimization algorithm to refine the model’s architecture and comprehensively bolster its generalization capability. A comparison with actual field data reveals that the fault diagnosis accuracy of the WT-MACNN model stands at 83.33%. Digital ablation experiments underscore the limited accuracy of the basic CNN model in fault diagnosis, but the sequential introduction of the channel attention mechanism and wavelet transform layers significantly elevates model performance. When juxtaposed with SVM, KNN, and decision tree models, the WT-MACNN model exhibits a diagnostic accuracy improvement of 73.81%, 57.13%, and 54.76%, respectively. Additionally, to assess the model’s adaptability under diverse well conditions, this study reacquired 128 sets of field data. The verification results indicate that the model’s prediction accuracy across different well conditions is 83.59%. Despite occasional misjudgments among certain operating conditions, the overall performance remains outstanding, offering dependable support for diagnosing real-world scenarios. The outcomes of numerical experiments highlight the profound advantages of the proposed deep learning model in terms of generalization ability and diagnostic accuracy, validating its superiority in plunger gas lift fault identification and carrying substantial guidance for the application of this technology.

AI-based prediction of androgen receptor expression and its prognostic significance in prostate cancer.

Biochemical recurrence (BCR) of prostate cancer (PCa) negatively impacts patients' post-surgery quality of life, and the traditional predictive models have shown limited accuracy. This study develops an AI-based prognostic model using deep learning that incorporates androgen receptor (AR) regional features from whole-slide images (WSIs). Data from 545 patients across two centres are used for training and validation. The model showed strong performances, with high accuracy in identifying regions with high AR expression and BCR prediction. This AI model may help identify high-risk patients, aiding in better treatment strategies, particularly in underdeveloped areas.

A hybrid deep learning model based on signal decomposition and dynamic feature selection for forecasting the influent parameters of wastewater treatment plants.

Accurate prediction of influent parameters such as chemical oxygen demand (COD) and biochemical oxygen demand over five days (BOD5) is crucial for optimizing wastewater treatment processes, enhancing efficiency, and reducing costs. Traditional prediction methods struggle to capture the dynamic variations of influent parameters. Mechanistic biochemical models are unable to predict these parameters, and conventional machine learning methods show limited accuracy in forecasting key water quality indicators such as COD and BOD5. This study proposes a hybrid model that combines signal decomposition and deep learning to improve the accuracy of COD and BOD5 predictions. Additionally, a new dynamic feature selection (DFS) mechanism is introduced to optimize feature selection in real-time, reducing model redundancy and enhancing prediction stability. The model achieved R2 values of 0.88 and 0.96 for COD, and 0.75 and 0.93 for BOD5 across two wastewater treatment plants. RMSE and MAE values were significantly reduced, with decreases of 14.93% and 12.55% for COD at WWTP No. 5, and 20.89% and 20.40% for COD at WWTP No. 7. For BOD5, RMSE and MAE decreased by 3.56% and 5.28% at WWTP No. 5, and by 10.06% and 10.20% at WWTP No. 7. These results highlight the effectiveness of the proposed model and DFS mechanism in improving prediction accuracy and model performance. This approach provides valuable insights for wastewater treatment optimization and broader time series forecasting applications.