Search

Depressive illness, influenced by social, psychological, and biological factors, is a significant public health concern that necessitates accurate and prompt diagnosis for effective treatment. This study explores the multifaceted nature of depression by investigating its correlation with various social factors and employing machine learning, natural language processing, and explainable AI to analyze depression assessment scales. Data from a survey of 520 Bangladeshi university students, encompassing socio-personal and clinical questions, was utilized in this study. Eight machine learning algorithms with optimized hyperparameters were applied to evaluate eight depression assessment scales, identifying the most effective one. Additionally, ten machine learning models, including five BERT-based and two generative large language models, were tested using three prompting approaches and assessed across four categories of social factors: relationship dynamics, parental pressure, academic contentment, and exposure to violence. The results showed that support vector machines achieved a remarkable 99.14% accuracy with the PHQ9 scale. While considering the social factors, the stacking ensemble classifier demonstrated the best results. Among NLP approaches, BioBERT outperformed other BERT-based models with 90.34% accuracy when considering all social aspects. In prompting approaches, the Tree of Thought prompting on Claude Sonnet surpassed other prompting techniques with 75.00% accuracy. However, traditional machine learning models outshined NLP methods in tabular data analysis, with the stacking ensemble model achieving the highest accuracy of 97.88%. The interpretability of the top-performing classifier was ensured using LIME.

Alzheimer’s disease, characterized by progressive and irreversible deterioration of cognitive functions, represents a significant health concern, particularly among older adults, as it stands as the foremost cause of dementia. Despite its debilitating nature, early detection of Alzheimer’s disease holds considerable advantages for affected individuals. This study investigates machine-learning methodologies for the early diagnosis of Alzheimer’s disease, utilizing datasets sourced from OASIS and ADNI. The initial classification methods consist of a 5-class ADNI classification and a 3-class OASIS classification. Three unique methodologies encompass binary-class inter-dataset models, which involve training on a single dataset and subsequently testing on another dataset for both ADNI and OASIS datasets. Additionally, a hybrid dataset model is also considered. The proposed methodology entails the concatenation of both datasets, followed by shuffling and subsequently conducting training and testing on the amalgamated dataset. The findings demonstrate impressive levels of accuracy, as Light Gradient Boosting Machine (LGBM) achieved a 99.63% accuracy rate for 5-class ADNI classification and a 95.75% accuracy rate by Multilayer Perceptron (MLP) for 3-class OASIS classification, both when hyperparameter tweaking was implemented. The K-nearest neighbor algorithm demonstrated exceptional performance, achieving an accuracy of 87.50% in ADNI-OASIS (2 Class) when utilizing the Select K Best method. The Gaussian Naive Bayes algorithm demonstrated exceptional performance in the OASIS-ADNI approach, attaining an accuracy of 77.97% using Chi-squared feature selection. The accuracy achieved by the Hybrid method, which utilized LGBM with hyperparameter optimization, was 99.21%. Furthermore, the utilization of Explainable AI approaches, particularly Lime, was implemented in order to augment the interpretability of the model.

Rice is one of the foremost food grains that dispenses sustenance to about half of the world’s population. It is cultivated all over the world. The leaf disease detection of this crop is one of the chronic agricultural obstacles that farmers and planting experts have been struggling with for a long time. As a result of the leaf diseases, producing the amount of rice required to feed the world’s rising population has become very challenging. Hence, automatically detecting rice leaf diseases is an inevitable task to increase productivity. Numerous deep learning based methods have been proposed for rice leaf disease detection, which we found rather inefficient considering the size of the models. In this article, we introduce a lightweight deep Convolutional Neural Network (dCNN) based method for rice leaf disease detection, that outperforms contemporary state-of-the-art methods and showcases competitive performance against 21 established benchmark architectures, including AlexNet, MobileNet, ResNet50, DenseNet121, ResNeXt50, ShuffleNet, ConvNext, EfficientNet, GogoleNet, SwinTransformer, VisionTransformer, and MaxVit, to name a few, with significantly lower trainable parameters. Notably, our method achieves an accuracy score of 99.81%, a precision score of 0.99828, a recall score of 0.99826, and an f1-score of 0.99827. Moreover, we enhance the rice leaf disease dataset by merging two existing datasets and supplemented them with an additional 95 manually annotated images gathered from publicly available sources on the internet. We also develop a comprehensive crop health monitoring system for farmers, and develop an open API for the automatic annotation of new instances, benefiting the research community at large.

Life satisfaction is a crucial facet of human well-being. Hence, research on life satisfaction is incumbent for understanding how individuals experience their lives and influencing interventions targeted at enhancing mental health and well-being. Life satisfaction has traditionally been measured using analog, complicated, and frequently error-prone methods. These methods raise questions concerning validation and propagation. However, this study demonstrates the potential for machine learning algorithms to predict life satisfaction with a high accuracy of 93.80% and a 73.00% macro F1-score. The dataset comes from a government survey of 19000 people aged 16-64 years in Denmark. Using feature learning techniques, 27 significant questions for assessing contentment were extracted, making the study highly reproducible, simple, and easily interpretable. Furthermore, clinical and biomedical large language models (LLMs) were explored for predicting life satisfaction by converting tabular data into natural language sentences through mapping and adding meaningful counterparts, achieving an accuracy of 93.74% and macro F1-score of 73.21%. It was found that life satisfaction prediction is more closely related to the biomedical domain than the clinical domain. Ablation studies were also conducted to understand the impact of data resampling and feature selection techniques on model performance. Moreover, the correlation between primary determinants with different age brackets was analyzed, and it was found that health condition is the most important determinant across all ages. The best performing Machine Learning model trained in this study is deployed on a public server, ensuring unrestricted usage of the model. We highlight the advantages of machine learning methods for predicting life satisfaction and the significance of XAI for interpreting and validating these predictions. This study demonstrates how machine learning, large language models and XAI can jointly contribute to building trust and understanding in using AI to investigate human behavior, with significant ramifications for academics and professionals working to quantify and comprehend subjective well-being.

Depression is a common psychiatric disorder that is becoming more prevalent in developing countries like Bangladesh. Depression has been found to be prevalent among youths and influences a person’s lifestyle and thought process. Unfortunately, due to the public and social stigma attached to this disease, the mental health issue of individuals are often overlooked. Early diagnosis of patients who may have depression often helps to provide effective treatment. This research aims to develop mechanisms to detect and predict depression levels and was applied to university students in Bangladesh. In this work, a questionnaire containing 106 questions has been constructed. The questions in the questionnaire are primarily of two kinds – (i) personal, and (ii) clinical. The questionnaire was distributed amongst Bangladeshi students and a total of 684 responses (aged between 19 and 35) were obtained. After appropriate consents from the participants, they were allowed to take the survey. After carefully scrutinizing the responses, 520 samples were taken into final consideration. A hybrid depression assessment scale was developed using a voting algorithm that employs eight well-known existing scales to assess the depression level of an individual. This hybrid scale was then applied to the collected samples that comprise personal information and questions from various familiar depression measuring scales. In addition, ten machine learning and two deep learning models were applied to predict the three classes of depression (normal, moderate and extreme). Five hyperparameter optimizers and nine feature selection methods were employed to improve the predictability. Accuracies of 98.08%, 94.23%, and 92.31% were obtained using Random Forest, Gradient Boosting, and CNN models, respectively. Random Forest accomplished the lowest false negatives and highest F Measure with its optimized hyperparameters. Finally, LIME, an explainable AI framework, was applied to interpret and retrace the prediction output of the machine learning models.

Deep learning techniques have recently demonstrated remarkable success in numerous domains. Typically, the success of these deep learning models is measured in terms of performance metrics such as accuracy and mean average precision (mAP). Generally, a model's high performance is highly valued, but it frequently comes at the expense of substantial energy costs and carbon footprint emissions during the model building step. Massive emission of CO2 has a deleterious impact on life on earth in general and is a serious ethical concern that is largely ignored in deep learning research. In this article, we mainly focus on environmental costs and the means of mitigating carbon footprints in deep learning models, with a particular focus on models created using knowledge distillation (KD). Deep learning models typically contain a large number of parameters, resulting in a 'heavy' model. A heavy model scores high on performance metrics but is incompatible with mobile and edge computing devices. Model compression techniques such as knowledge distillation enable the creation of lightweight, deployable models for these low-resource devices. KD generates lighter models and typically performs with slightly less accuracy than the heavier teacher model (model accuracy by the teacher model on CIFAR 10, CIFAR 100, and TinyImageNet is 95.04%, 76.03%, and 63.39%; model accuracy by KD is 91.78%, 69.7%, and 60.49%). Although the distillation process makes models deployable on low-resource devices, they were found to consume an exorbitant amount of energy and have a substantial carbon footprint (15.8, 17.9, and 13.5 times more carbon compared to the corresponding teacher model). The enormous environmental cost is primarily attributable to the tuning of the hyperparameter, Temperature (τ). In this article, we propose measuring the environmental costs of deep learning work (in terms of GFLOPS in millions, energy consumption in kWh, and CO2 equivalent in grams). In order to create lightweight models with low environmental costs, we propose a straightforward yet effective method for selecting a hyperparameter (τ) using a stochastic approach for each training batch fed into the models. We applied knowledge distillation (including its data-free variant) to problems involving image classification and object detection. To evaluate the robustness of our method, we ran experiments on various datasets (CIFAR 10, CIFAR 100, Tiny ImageNet, and PASCAL VOC) and models (ResNet18, MobileNetV2, Wrn-40-2). Our novel approach reduces the environmental costs by a large margin by eliminating the requirement of expensive hyperparameter tuning without sacrificing performance. Empirical results on the CIFAR 10 dataset show that the stochastic technique achieves an accuracy of 91.67%, whereas tuning achieves an accuracy of 91.78%-however, the stochastic approach reduces the energy consumption and CO2 equivalent each by a factor of 19. Similar results have been obtained with CIFAR 100 and TinyImageNet dataset. This pattern is also observed in object detection classification on the PASCAL VOC dataset, where the tuning technique performs similarly to the stochastic technique, with a difference of 0.03% mAP favoring the stochastic technique while reducing the energy consumptions and CO2 emission each by a factor of 18.5.

Named Entity Recognition (NER) is a task in Natural Language Processing (NLP) that aims to classify words into a predetermined list of Named Entities (NE). Many architectures have produced good results on high resourced languages like English and Chinese. However, the NER task has not yet achieved much progress for Bangla, a low resource Language. In this paper, we perform the NER task on Bangla Language using Word2Vec and contextual Bidirectional Encoder Representations from Transformers (BERT) embeddings. We propose multiple BERT-based deep learning models that use the contextualized embedding from BERT as inputs and a simple statistical approach for class weight cost sensitive learning. The modified cost-sensitive loss function was used to address the class imbalance of the data. In our modified cost-sensitive loss function, we penalize the dominant classes by taking the ratio concerning the maximum sample in a class instead of the whole dataset. This penalty is made so that the learner learns slowly for the dominant class. In addition, we experiment by adding a Conditional Random Field (CRF) layer and incorporating Focal Loss to the training process. We found the best F1 Macro score to be 65.96%, F1 Micro score of 90.64%, and F1 Message Understanding Coreference (MUC) score of 72.04%, which were calculated at Named Entity level. Our experimental results demonstrate that one of the proposed models, which jointly optimizes for the CRF loss and class weighted cost-sensitive loss according to our proposed statistical approach, achieve an improvement of over 8% F1 MUC score on a recently introduced Bangla NER dataset when compared to previously published work.

Pre-owned cars (i.e., cars with one or more previous retail owners) are extremely popular in Bangladesh. Customers who plan to purchase a pre-owned car often struggle to find a car within a budget as well as to predict the price of a particular pre-owned car. Currently, Bangladesh lacks online services that can provide assistance to customers purchasing pre-owned cars. A good prediction of prices of pre-owned cars can help customers greatly in making an informed decision about buying a pre-owned car. In this article, we look into this problem and develop a forecasting system (using machine learning techniques) that helps a potential buyer to estimate the price of a pre-owned car he is interested in. A dataset is collected and pre-processed. Exploratory data analysis has been performed. Following that, various machine learning regression algorithms, including linear regression, LASSO (Least Absolute Shrinkage and Selection Operator) regression, decision tree, random forest, and extreme gradient boosting have been applied. After evaluating the performance of each method, the best-performing model (XGBoost) was chosen. This model is capable of properly predicting prices more than 91% of the time. Finally, the model has been deployed as a web application in a local machine so that this can be later made available to end users.

BanglaLekha-Isolated, a Bangla handwritten isolated character dataset is presented in this article. This dataset contains 84 different characters comprising of 50 Bangla basic characters, 10 Bangla numerals and 24 selected compound characters. 2000 handwriting samples for each of the 84 characters were collected, digitized and pre-processed. After discarding mistakes and scribbles, 1,66,105 handwritten character images were included in the final dataset. The dataset also includes labels indicating the age and the gender of the subjects from whom the samples were collected. This dataset could be used not only for optical handwriting recognition research but also to explore the influence of gender and age on handwriting. The dataset is publicly available at https://data.mendeley.com/datasets/hf6sf8zrkc/2.

Machine Learning has a significant impact on different aspects of science and technology including that of medical researches and life sciences. Diabetes Mellitus, more commonly known as diabetes, is a chronic disease that involves abnormally high levels of glucose sugar in blood cells and the usage of insulin in the human body. This article has focused on analyzing diabetes patients as well as detection of diabetes using different Machine Learning techniques to build up a model with a few dependencies based on the PIMA dataset. The model has been tested on an unseen portion of PIMA and also on the dataset collected from Kurmitola General Hospital, Dhaka, Bangladesh. The research is conducted to demonstrate the performance of several classifiers trained on a particular country’s diabetes dataset and tested on patients from a different country. We have evaluated decision tree, K-nearest neighbor, random forest, and Naïve Bayes in this research and the results show that both random forest and Naïve Bayes classifier performed well on both datasets.

Malaria is a life-threatening disease that is spread by the Plasmodium parasites. It is detected by trained microscopists who analyze microscopic blood smear images. Modern deep learning techniques may be used to do this analysis automatically. The need for the trained personnel can be greatly reduced with the development of an automatic accurate and efficient model. In this article, we propose an entirely automated Convolutional Neural Network (CNN) based model for the diagnosis of malaria from the microscopic blood smear images. A variety of techniques including knowledge distillation, data augmentation, Autoencoder, feature extraction by a CNN model and classified by Support Vector Machine (SVM) or K-Nearest Neighbors (KNN) are performed under three training procedures named general training, distillation training and autoencoder training to optimize and improve the model accuracy and inference performance. Our deep learning-based model can detect malarial parasites from microscopic images with an accuracy of 99.23% while requiring just over 4600 floating point operations. For practical validation of model efficiency, we have deployed the miniaturized model in different mobile phones and a server-backed web application. Data gathered from these environments show that the model can be used to perform inference under 1 s per sample in both offline (mobile only) and online (web application) mode, thus engendering confidence that such models may be deployed for efficient practical inferential systems.