A New Artificial Intelligence Approach Using Extreme Learning Machine as the Potentially Effective Model to Predict and Analyze the Diagnosis of Anemia

Machine learning-based detection and quantification of red blood cells in Cholistani cattle: A pilot study

This study presents the first account of using machine learning to detect and count normal and abnormal red blood cells (RBCs), including tear-drop cells and schistocytes, in Cholistani cattle from Pakistan. A Support Vector Machine (SVM) model was applied and compared with manual counting methods. Pre-annotated blood smear images were preprocessed using contrast stretching transformation, followed by segmentation and resizing. Labeled datasets were augmented, and Principal Component Analysis (PCA) was employed for feature reduction. The dataset was randomly split into training (80 %) and testing (20 %) subsets, and the SVM model was trained and evaluated accordingly. No statistically significant difference (P ≥ 0.05) was observed between manual and machine learning-based RBC counts for all the studied cell types. The highest classification probability was recorded for normal RBCs (87 %), followed by tear-drop cells (84 %) and schistocytes (73 %). Accuracy was highest for tear-drop cells (0.991), followed by normal RBCs (0.965) and schistocytes (0.707). Precision values followed a similar trend, with the highest for normal RBCs (0.932), followed by tear-drop cells (0.921) and schistocytes (0.855). These findings suggest that machine learning, particularly SVM-based models, can accurately and precisely detect and count normal RBCs and tear-drop cells in Cholistani cattle. However, further refinements are needed to improve RBC detection using convolution neural networks or other deep learning approaches. This study highlights the potential of artificial intelligence for hematological assessments in veterinary medicine.

Quantitative analysis of safflower seed oil adulteration based on near-infrared spectroscopy combined with improved sparrow algorithm optimization model ISSA-ELM

Safflower seed oil is expensive, and there are issues in the market such as the adulteration with cheaper edible oils, leading to inferior products. Near-infrared spectroscopy (NIR) is a non-destructive analytical method with the advantages of being fast, non-destructive, and pollution-free. However, in the case of low-concentration adulterated oil, the main components and their contents are nearly identical to those in pure oil, and the spectral feature peaks of the samples are similar, making it difficult for conventional analysis methods to select effective characteristic variables. Extreme Learning Machine (ELM), as a single-hidden-layer feedforward neural network model, possesses strong feature extraction and model representation capabilities. However, its random initialization of weights and biases leads to the issue of blind training. The Sparrow Search Algorithm (SSA) can effectively optimize the random initialization problem in ELM, but it is prone to issues such as being trapped in local optima and slower convergence. This study proposes a novel quantitative adulteration analysis model for safflower seed oil (ISSA-ELM), which combines ELM with an improved Sparrow Search Algorithm (ISSA). In the experiment, safflower seed oil was used as the base oil, and peanut oil, corn oil, and soybean oil were gradually added to prepare adulterated oil samples. To ensure the resolution of experimental samples and accurately capture spectral variations in the low-concentration range, a concentration gradient of 2% was used for adulteration levels between 2% and 70%, while a 5% gradient was applied for concentrations exceeding 70%. Each type of adulterant had 42 gradient concentrations, with 6 samples per concentration, totaling 252 samples. Original spectral data were collected, and the samples were randomly divided into a training set (70%) and a testing set (30%). Different preprocessing methods and feature extraction techniques were combined to discuss the results of three adulteration analysis models: PLS, SSA-ELM and ISSA-ELM. The final experimental results show that compared to the safflower seed oil adulteration prediction model based on SSA-ELM (R2 = 0.8938, RMSE = 0.0835, RPD = 2.9018)and PLS(RP2 = 0.9015, RMSEP = 0.0876, RPD = 3.0128), the model based on ISSA-ELM (RP2 = 0.9934, RMSEP = 0.0207, RPD = 10.1457) offers higher prediction accuracy and stability, and the error detection limit of ISSA-ELM can reach 2%. The ISSA optimized the SSA's tendency to reduce population diversity, slow convergence, and get stuck in local optima in the later stages of iteration, greatly enhancing the global optimization ability of the algorithm. Therefore, ISSA-ELM can effectively identify adulterated safflower seed oil, providing a technological pathway and basis for research into the adulteration of safflower seed oil.

Differential diagnosis of iron deficiency anemia from aplastic anemia using machine learning and explainable Artificial Intelligence utilizing blood attributes

As per world health organization, Anemia is a most prevalent blood disorder all over the world. Reduced number of Red Blood Cells or decrease in the number of healthy red blood cells is considered as Anemia. This condition also leads to the decrease in the oxygen carrying capacity of the blood. The main goal of this research is to develop a dependable method for diagnosing Aplastic Anemia and Iron Deficiency Anemia by examining the blood test attributes. As of today, there are no studies which use Interpretable Artificial Intelligence to perform the above differential diagnosis. The dataset used in this study is collected from Kasturba Medical College, Manipal. The dataset consisted of various blood test attributes such as Red Blood cell count, Hemoglobin level, Mean Corpuscular Volume, etc. One of the trending topics in Machine Learning is Explainable Artificial Intelligence. They are known to demystify the machine learning outputs to all its stakeholders. Hence, Five XAI tools including SHAP, LIME, Eli5, Qlattice and Anchor are used to understand the model's predictions. The importance characteristics according to XAI models are PLT, PCT, MCV, PDW, HGB, ABS LYMP, WBC, MCH, and MCHC. are employed to train and test the data. The goal of using data analytic techniques is to give medical professionals a useful tool that improves decision-making, enhances resource management, and eventually raises the standard of patient care. By considering the unique qualities of each patient, medical professionals who must rely on AI-assisted diagnosis and treatment suggestions, XAI offers arguments to strengthen their faith in the model outcomes.

Leveraging machine learning models for anemia severity detection among pregnant women following ANC: Ethiopian context

BACKGROUND

Anemia during pregnancy is a significant public health concern, particularly in resource-limited settings. Machine learning (ML) offers promising avenues for improved anemia detection and management. This study investigates the potential of ML models in predicting anemia severity among pregnant women attending Antenatal Care (ANC) visits in Ethiopia.

METHODS

Data from the Ethiopian Demographic Health Survey, specialized hospitals, and public hospitals were utilized. The dataset included individuals diagnosed with severe (65.12%), moderate (15.63%), mild (16.65%) anemia, and non-anemic (2.61%) cases. Feature selection employed filter methods based on mutual information, and F-score was used to assess anemia severity prediction across four classes. Six ML models (MLP-NN, XGBoost, GNB, Decision Tree, Random Forest, and KNN) were evaluated using accuracy, precision, recall, and F1-score.

RESULTS

The Random Forest classifier achieved the best overall performance across all categories, with an accuracy of 97%, precision of 93%, recall of 93%, and F1-score of 93%. This indicates high true positive rates and low false positive rates. While other models like XGBoost, MLP-NN, and Decision Tree showed good performance, they weren't quite as strong as Random Forest. Classifiers like KNN and GNB had lower overall accuracy and a tendency to misclassify some cases.

CONCLUSIONS

This study demonstrates the promising potential of Random Forest in predicting anemia severity among pregnant women in Ethiopia. The findings contribute to a more holistic understanding of anemia risk factors and pave the way for improved early detection and targeted interventions.

Multidisciplinary approaches to study anaemia with special mention on aplastic anaemia (Review)

Anaemia is a common health problem worldwide that disproportionately affects vulnerable groups, such as children and expectant mothers. It has a variety of underlying causes, some of which are genetic. A comprehensive strategy combining physical examination, laboratory testing (for example, a complete blood count), and molecular tools for accurate identification is required for diagnosis. With nearly 400 varieties of anaemia, accurate diagnosis remains a challenging task. Red blood cell abnormalities are largely caused by genetic factors, which means that a thorough understanding requires interpretation at the molecular level. As a result, precision medicine has become a key paradigm, utilising artificial intelligence (AI) techniques, such as deep learning and machine learning, to improve prognostic evaluation, treatment prediction, and diagnostic accuracy. Furthermore, exploring the immunomodulatory role of vitamin D along with biomarker‑based molecular techniques offers promising avenues for insight into anaemia's pathophysiology. The intricacy of aplastic anaemia makes it particularly noteworthy as a topic deserving of concentrated molecular research. Given the complexity of anaemia, an integrated strategy integrating clinical, laboratory, molecular, and AI techniques shows a great deal of promise. Such an approach holds promise for enhancing global anaemia management options in addition to advancing our understanding of the illness.

Assessment of iron status

PURPOSE OF REVIEW

Iron is an essential trace element in human health that can be harmful at abnormal levels such as iron overload or deficiency. Measured iron status in the body can depend on health outcomes experienced by the individual and this can complicate its accurate assessment. This review will highlight recent research on iron assessment in the literature.

RECENT FINDINGS

Research on iron assessment within the past 18 months included some common themes spanning new methods and biomarkers, as well as existing problems in assessing iron deficiency and overload. Heterogeneity in associations between inflammation and iron levels are reflected across different inflammatory biomarkers. New methods relevant to low- and high-resource settings may improve assessment in tissues with iron deficiency and overload. Consensus papers outlined best practices when using MRI to assess iron status. Outside of newer methods, traditional serum markers are the subject of a call for updated guidance when assessing iron status.

SUMMARY

Research continues on the topic of iron assessment, underlying its complex metabolism in the body and resulting challenges in assessment. Current literature underscores progress to make iron assessment more accessible, improve existing methods, and update current assessment methods so they correspond with recent research to improve human health.

BrainNet: a fusion assisted novel optimal framework of residual blocks and stacked autoencoders for multimodal brain tumor classification

A significant issue in computer-aided diagnosis (CAD) for medical applications is brain tumor classification. Radiologists could reliably detect tumors using machine learning algorithms without extensive surgery. However, a few important challenges arise, such as (i) the selection of the most important deep learning architecture for classification (ii) an expert in the field who can assess the output of deep learning models. These difficulties motivate us to propose an efficient and accurate system based on deep learning and evolutionary optimization for the classification of four types of brain modalities (t1 tumor, t1ce tumor, t2 tumor, and flair tumor) on a large-scale MRI database. Thus, a CNN architecture is modified based on domain knowledge and connected with an evolutionary optimization algorithm to select hyperparameters. In parallel, a Stack Encoder-Decoder network is designed with ten convolutional layers. The features of both models are extracted and optimized using an improved version of Grey Wolf with updated criteria of the Jaya algorithm. The improved version speeds up the learning process and improves the accuracy. Finally, the selected features are fused using a novel parallel pooling approach that is classified using machine learning and neural networks. Two datasets, BraTS2020 and BraTS2021, have been employed for the experimental tasks and obtained an improved average accuracy of 98% and a maximum single-classifier accuracy of 99%. Comparison is also conducted with several classifiers, techniques, and neural nets; the proposed method achieved improved performance.

Classification and Explanation of Iron Deficiency Anemia from Complete Blood Count Data Using Machine Learning

Background: Currently, discriminating Iron Deficiency Anemia (IDA) from other anemia requires an expensive test (serum ferritin). Complete Blood Count (CBC) tests are less costly and more widely available. Machine learning models have not yet been applied to discriminating IDA but do well for similar tasks. Methods: We constructed multiple machine learning methods to classify IDA from CBC data using a US NHANES dataset of over 19,000 instances, calculating accuracy, precision, recall, and precision AUC (PR AUC). We validated the results using an unseen dataset from Kenya, using the same model. We calculated ranked feature importance to explain the global behavior of the model. Results: Our model classifies IDA with a PR AUC of 0.87 and recall/sensitivity of 0.98 and 0.89 for the original dataset and an unseen Kenya dataset, respectively. The explanations indicate that low blood level of hemoglobin, higher age, and higher Red Blood Cell distribution width were most critical. We also found that optimization made only minor changes to the explanations and that the features used remained consistent with professional practice. Conclusions: The overall high performance and consistency of the results suggest that the approach would be acceptable to health professionals and would support enhancements to current automated CBC analyzers.

Laboratory tests for investigating anemia: From an expert system to artificial intelligence

Objective

To compare the laboratory tests conducted in real-life settings for patients with anemia with the expected prescriptions derived from an optimal checkup.

Methods

A panel of experts formulated an "optimal laboratory test assessment" specific to each anemia profile. A retrospective analysis was done of the laboratory tests conducted according to the type of anemia (microcytic, normocytic or macrocytic). Using an algorithmic system, the laboratory tests performed in real-life practice were compared with the recommendations suggested in the "optimal laboratory test assessment" and with seemingly "unnecessary" laboratory tests.

Results

In the analysis of the "optimal laboratory test assessment", of the 1179 patients with microcytic anemia, 269 (22.8%) had had one of the three tests recommended by the expert system, and only 33 (2.8%) had all three tests. For normocytic anemia, 1054 of 2313 patients (45.6%) had one of the eleven recommended tests, and none had all eleven. Of the 384 patients with macrocytic anemia, 196 (51%) had one of the four recommended tests, and none had all four. In the analysis of "unnecessary laboratory tests", one lab test was unnecessarily done in 727/3876 patients (18.8%), i.e. 339 of 1179 (28.8%) microcytic, 171 of 2313 (7.4%) normocytic, and 217 of 384 (56.5 %) macrocytic anemias.

Conclusion

Laboratory investigations of anemia remain imperfect as more than half of the cases did not receive the expected tests. Analyzing other diagnostic domains, the authors are currently developing an artificial intelligence system to assist physicians in enhancing the efficiency of their laboratory test prescriptions.

Diagnosis and screening of abnormal hemoglobins

Hemoglobin (Hb) abnormalities, such as thalassemia and structural Hb variants, are among the most prevalent inherited diseases and are associated with significant mortality and morbidity worldwide. However, there were not comprehensive reviews focusing on different clinical analytical techniques, research methods and artificial intelligence (AI) used in clinical screening and research on hemoglobinopathies. Hence the review offers a comprehensive summary of recent advancements and breakthroughs in the detection of aberrant Hbs, research methods and AI uses as well as the present restrictions anddifficulties in hemoglobinopathies. Recent advances in cation exchange high performance liquid chromatography (HPLC), capillary zone electrophoresis (CZE), isoelectric focusing (IEF), flow cytometry, mass spectrometry (MS) and polymerase chain reaction (PCR) etc have allowed for the definitive detection by using advanced AIand portable point of care tests (POCT) integrating with smartphone microscopic classification, machine learning (ML) model, complete blood counts (CBC), imaging-based method, speedy immunoassay, and electrochemical-, microfluidic- and sensing-related platforms. In addition, to confirm and validate unidentified and novel Hbs, highly specialized genetic based techniques like PCR, reverse transcribed (RT)-PCR, DNA microarray, sequencing of genomic DNA, and sequencing of RT-PCR amplified globin cDNA of the gene of interest have been used. Hence, adequate utilization and improvement of available diagnostic and screening technologies are important for the control and management of hemoglobinopathies.

Design of a Collaborative Knowledge Framework for Personalised Attention Deficit Hyperactivity Disorder (ADHD) Treatments

Attention deficit hyperactivity disorder (ADHD) is a neurodevelopmental disorder. From the data collected by the Ministry of Public Health, Thailand, it has been reported that more than one million Thai youths (6-12 years) have been diagnosed with ADHD (2012-2018) This disorder is more likely to occur in males (12%) than females (4.2%). If ADHD goes untreated, there might be problems for individuals in the long run. This research aims to design a collaborative knowledge framework for personalised ADHD treatment recommendations. The first objective is to design a framework and develop a screening tool for doctors, parents, and teachers for observing and recording behavioural symptoms in ADHD children. This screening tool is a combination of doctor-verified criteria and the ADHD standardised screening tool (Vanderbilt). The second objective is to introduce practical algorithms for classifying ADHD types and recommending appropriate individual behavioural therapies and activities. We applied and compared four well-known machine-learning methods for classifying ADHD types. The four algorithms include Decision Tree, Naïve Bayes, neural network, and k-nearest neighbour. Based on this experiment, the Decision Tree algorithm yielded the highest average accuracy, which was 99.60%, with F1 scores equal to or greater than 97% for classifying each type of ADHD.

Artificial Intelligence-Assisted Diagnostic Cytology and Genomic Testing for Hematologic Disorders

Artificial intelligence (AI) is a rapidly evolving field of computer science that involves the development of computational programs that can mimic human intelligence. In particular, machine learning and deep learning models have enabled the identification and grouping of patterns within data, leading to the development of AI systems that have been applied in various areas of hematology, including digital pathology, alpha thalassemia patient screening, cytogenetics, immunophenotyping, and sequencing. These AI-assisted methods have shown promise in improving diagnostic accuracy and efficiency, identifying novel biomarkers, and predicting treatment outcomes. However, limitations such as limited databases, lack of validation and standardization, systematic errors, and bias prevent AI from completely replacing manual diagnosis in hematology. In addition, the processing of large amounts of patient data and personal information by AI poses potential data privacy issues, necessitating the development of regulations to evaluate AI systems and address ethical concerns in clinical AI systems. Nonetheless, with continued research and development, AI has the potential to revolutionize the field of hematology and improve patient outcomes. To fully realize this potential, however, the challenges facing AI in hematology must be addressed and overcome.

The Clinical Researcher Journey in the Artificial Intelligence Era: The PAC-MAN’s Challenge

Artificial intelligence (AI) is a powerful tool that can assist researchers and clinicians in various settings. However, like any technology, it must be used with caution and awareness as there are numerous potential pitfalls. To provide a creative analogy, we have likened research to the PAC-MAN classic arcade video game. Just as the protagonist of the game is constantly seeking data, researchers are constantly seeking information that must be acquired and managed within the constraints of the research rules. In our analogy, the obstacles that researchers face are represented by “ghosts”, which symbolize major ethical concerns, low-quality data, legal issues, and educational challenges. In short, clinical researchers need to meticulously collect and analyze data from various sources, often navigating through intricate and nuanced challenges to ensure that the data they obtain are both precise and pertinent to their research inquiry. Reflecting on this analogy can foster a deeper comprehension of the significance of employing AI and other powerful technologies with heightened awareness and attentiveness.