Multi-Class Classification Framework for Early Detection of Liver Diseases Sandeep Singh Kang, Parvez Rahi Conference Proceedngs Wccst 2026 World Conference on Computational Science and Technology, 2026 Liver disease refers to an acute health issue affecting the world with the increasing rates of occurrence due to alcohol use, lifestyle and increasing age population. The comorbidity such as hypertension, liver failure and cancer risk should be preemptively diagnosed and correctly categorized to offer optimal patient outcome and minimise medical expenditure. The paper hypothesizes an XGBoost machine learning framework to predict five categories of hepatitis, cirrhosis, non-alcoholic fatty liver disease (NAFLD), alcoholic liver disease (ALD), and hemochromatosis of a Kaggle dataset that it has borrowed. Hyperparameter tuning and feature scaling were done using gridSearchCV, missing values were solved using imputation and SMOTE was used to solve the imbalance in the classes. Good prediction power was 0.928 with accuracy of model was 93%. Feature importance calculation determined the top predictors on additional validation through accuracy, precision, recall, F1-score, confusion matrix, and ROC curve confirmed the correctness of the above approach. Results show that the model offers a scalable technique to improve early diagnosis of liver disease with potential applications in clinical practice and health economy.
Framework for Stock Price Forecasting Using Fundamental Analysis and Fuzzy-Driven LSTM Parvez Rahi, Sandeep Singh Kang Conference Proceedngs Wccst 2026 World Conference on Computational Science and Technology, 2026 This paper proposes a hybrid method applying Fuzzy Logic together with LSTM neural networks to enhance the precision of forecast of stock price from underlying financial values. Fuzzy financial data is handled efficiently by Fuzzy Logic, and the LSTM network is working on such fuzzy data to provide the forecast. The model was run on historical data of Apple Inc. (AAPL), Microsoft Corporation (MSFT), Alphabet Inc. (GOOGL), Amazon.com Inc. (AMZN), and Tesla Inc. (TSLA) and produced Mean Squared Error (MSE) value of <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{0. 0 0 0 1 5}$</tex>, Root Mean Squared Error (RMSE) value of 0.038, and coefficient of determination (<tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{R}^{\wedge} \mathbf{2}$</tex>) value of 0.9834 with a staggering prediction accuracy of 92.10 %. These figures verify the model attained record gain in forecasting power and validity compared to traditional methods. This essay recognizes the promise of the union of fuzzy logic and sophisticated models of neural networks to enable smoother stock market analysis and logical decisionmaking under a condition of uncertainty within a market.
A Novel Random Forest-SMOTE Framework With Polynomial Feature Engineering for Early Detection of Gastrointestinal Disorders Parvez Rahi, Mohammad Rashid Hussain, Salem Alqahtani, Mohammad Husain, Ajay Pal Singh, Inderjeet Singh IEEE Access, 2025 About 40% of people around the world have gastrointestinal (GI) problems. Gastrointestinal issues negatively impact the lives of millions of people and place significant stress on healthcare systems. The wrong diagnosis might make treatment less effective because symptoms like diarrhea, bloating, stomach pain, and liver problems often happen at the same duration. For early intervention and personalized care, it is essential to correctly and quickly classify these symptoms. We used the Random Forest model to create a strong classification framework for this study. It can tell if someone has one of seven main symptom groups: bloating, diarrhea or constipation, blood in the stool, unexplained weight loss, liver problems, nausea or vomiting, or abdominal cramps or pain. In our model we have used a multi-modal dataset that includes genetic markers, clinical assessments, lifestyle factors, and diagnostic imaging data to give a full picture of the diagnosis, and then we added polynomial feature engineering and other methods to the model to make it better at finding hidden and small patterns and relations in the parameters. We have also used the Synthetic Minority Oversampling Technique (SMOTE) to fix the problem of uneven data and GridSearchCV to make the model robust against hyperparameters. Although past studies have mostly focused on the lesion-based imaging or targeted disease identification, our model is a step forward in terms of symptom-based early classification with multi-modal tabular data. The model is capable of combining clinically informed two-factor interactions (e.g., stress × appetite loss, inflammation × bowel habits), which significantly improve the predictive performance and interpretability. This design reduces the disparity between the algorithmic predictions and the clinician thinking, thus, the system has become more applicable to realistic diagnostic decision-making. In this way, it allows for personalized treatment plans that make the disease curable and save time and risk of severity of GI disease. Future work will focus on adding biomarker data and real-time patient tracking to the framework so that it can be used easily in clinical decision-making.
Clinically Interpretable and Statistically Robust EHR-Based Ensemble Model for Anxiety Screening Parvez Rahi, Sanjay Singla Proceedings of the IEEE International Conference Image Information Processing, 2025 Anxiety disorders affect over 301 million people worldwide, yet early detection remains challenging due to the lack of scalable and clinically interpretable screening solutions, particularly in primary care and telehealth settings. This study introduces an AI-driven framework for first-line anxiety screening using structured Electronic Health Record (EHR) data and a soft-voting ensemble combining Random Forest and XGBoost classifiers. Two benchmark datasets were used: the Enhanced Anxiety dataset for model training and internal validation, and the Family_Anxiety_14 dataset for external testing to ensure generalizability. Anxiety risk was classified based on Generalized Anxiety Disorder (GAD) thresholds (0–4: Not Anxious; 5–10: Anxious). The proposed model achieved 97.21% accuracy (95% CI: 96.84–97.58) and an F1-score of 0.972, significantly outperforming individual base learners (McNemar’s test, p < 0.001). Its light-weight, interpretable design makes it easy to integrate into clinical workflows, quickly triaging at-risk patients for further evaluation. This focused approach optimizes healthcare resources, reduces diagnostic delay, and allows timely treatment. Its expansion to a multimodal framework of physiological, behavioral, and textual cues for severity estimation will be the subject of future research, maximizing personalization and clinical decision support.
Stock Market Prediction with LSTM and Graph Neural Networks: A Multi-Feature Hybrid Approach Ajay Pal Singh, Parvez Rahi 2025 2nd International Conference on Advanced Computing and Emerging Technologies Acet 2025, 2025 Forecasting the stock market is a very challenging task because financial time series are volatile and nonlinear. This paper unleashes a strong hybrid architecture consisting of Long Short-Term Memory (LSTM) networks and Graph Neural Networks (GNNs) to both capture temporal relationships and structural correlations among stocks at the same time. The model uses historical stock prices and basic indicators like Price-to-Earnings (P/E) ratio, Market Capitalization, and Revenue Growth in order to help us understand the manner in which the market works in comparison. LSTM module can effectively model sequential patterns, whereas the GNN part of the model learns the manner in which firms and sectors are related. Ensemble integration further improves predictive stability and minimizes overfitting. The stability and accuracy of the model are ensured by a <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathrm{R}^{\mathrm{2}}$</tex> value of 0.987, a very low Mean Squared Error (MSE) and Root Mean Squared Error (RMSE), and 93.24% total accuracy on a multi-company data set. The hybrid architecture introduced is a decision-support system that has the potential to grow. It offers investors, researchers, and planners in algorithmic trading and stock market research with informative material that they can use.
Dual-Pipeline LSTM Screening for Anxiety and Depression via Video Interviews Parvez Rahi, Sanjay Singla 2025 IEEE 7th International Conference on Computing Communication and Automation Iccca 2025, 2025 More than 264 million people suffer from depression and more than 301 million have anxiety disorders globally, as stated by the World Health Organization, indicating the urgency for early diagnosis. Historically used diagnosis is typically in demand for specialized clinicians and long testing and has obstacles to rapid treatment. In order to undo this, we propose hierarchical Long Short-Term Memory (LSTM) architecture for AI-based binary screening of risk of mental illness based on pre-prepared interview responses. Our subjects undergo an interview in the form of a video whose questions have been pre-prepared according to the DAIC-WOZ protocol and PHQ-8 clinical guidelines. They present responses that are accepted as transcripts and fed into the analysis framework. The design consists of a word-level LSTM semantic speech encoding and an utterance-level LSTM for contextual flow encoding at the session level. Experiments were conducted on two benchmark datasets, the DAIC-WOZ corpus for depression and Simulated Face-to-Face Medical Consultation Corpus (SFMCM) for anxiety. Results indicate that the model attains 90.62% classification accuracy with an F1-Score of 0.91, which is indicative of its ability to flag high-risk individuals. This work demonstrates the promise of LSTM-based models for large-scale text-based mental health screening. Scaling this up to a dual-pipeline diagnosis model screening for depression and anxiety simultaneously is what our future research will involve, bringing us yet another step closer to reality with AI-enabled mental health solutions.
An Image Synthesis Using Progressive Generative Adversarial Networks (PGANs) Ajay Pal Singh, Parvez Rahi, Vinod Kumar Handbook of Intelligent Automation Systems Using Computer Vision and Artificial Intelligence, 2025 This chapter provides a summary of the utilization of generative adversarial networks (GANs) in the domain of computer vision—image synthesis and manipulation. GANs are composed of two intricate neural networks, specifically, two competitively trained components—a discriminator and a generator. Owing to the formidable capabilities of deep neural networks and their adversarial training approach, GANs exhibit the ability to generate realistic and plausible images, thereby demonstrating remarkable prowess across various applications in the realm of image synthesis and manipulation. This survey paper delves into recent GAN-related research. Working on the technique of progressive GANs model helps to find out how capable is it than other GANs algorithms. Two brain organizations make up GANs for image synthesis: a discriminator and a generator. The discriminator determines if a picture is real (from a dataset) or fake (created by the generator), whereas the generator creates artificial images from erratic noise. In preparation, the discriminator aims to improve its ability to distinguish authentic from counterfeit images, while the generator plans to provide images that are indistinct from real ones. The generator creates increasingly realistic images as a result of this adversarial process. In domains like PC vision and designs, GANs have been widely used to generate superb images, creative representations, and information expansion.
Reverse Image Search Using Deep Learning Ajay Pal Singh, Sahil Singla, Parvez Rahi 2025 International Conference on Engineering Innovations and Technologies Icoeit 2025, 2025 Reverse image search has become a valuable technique for identifying and retrieving visually comparable images from extensive databases, with uses spanning from content authentication to online retail. This study investigates the creation of an effective reverse image search system utilizing deep learning methods. We introduce a convolutional neural network (CNN)-based framework trained on a varied image collection to extract robust feature embeddings, facilitating precise image similarity assessments. By combining a feature extraction process with an enhanced indexing system, our method achieves high accuracy and scalability. Tests on benchmark datasets show that the suggested model surpasses conventional approaches like SIFT and SURF in terms of retrieval precision and computational efficiency. Additionally, we tackle issues related to managing noisy data and diverse image resolutions, offering insights into the real-world implementation of deep learning-based reverse image search systems. This study enhances image retrieval technologies and highlights the potential of deep learning in solving complex visual search problems.
An Efficient ADA Boost and CNN Hybrid Model for Weed Detection and Removal Proceedings of the 17th Indiacom 2023 10th International Conference on Computing for Sustainable Global Development Indiacom 2023, 2023
RECENT SCHOLAR PUBLICATIONS
Multi-Class Classification Framework for Early Detection of Liver Diseases SS Kang, P Rahi 2026 World Conference on Computational Science and Technology (WcCST), 752-757 , 2026 2026
Framework for Stock Price Forecasting Using Fundamental Analysis and Fuzzy-Driven LSTM P Rahi, SS Kang 2026 World Conference on Computational Science and Technology (WcCST), 758-763 , 2026 2026
Autoencoder–XGBoost Ensemble Framework for Smart EV Charging Duration Prediction D Mishra, S Kumar, P Dwivedi, P Rahi, S Negi 2026 3rd International Conference on Advancements and Key Challenges in … , 2026 2026
Dual-Pipeline LSTM Screening for Anxiety and Depression via Video Interviews P Rahi, S Singla 2025 IEEE 7th International Conference on Computing, Communication and … , 2025 2025
Clinically Interpretable and Statistically Robust EHR-Based Ensemble Model for Anxiety Screening P Rahi, S Singla 2025 Eighth International Conference on Image Information Processing (ICIIP … , 2025 2025
A novel paradigm in cardiovascular disease risk prediction through hybrid machine learning P Rahi, SS Kang Системная инженерия и информационные технологии 7 (3 (22)), 48-65 , 2025 2025 Citations: 6
Stock Market Prediction with LSTM and Graph Neural Networks: A Multi-Feature Hybrid Approach AP Singh, P Rahi 2025 2nd International Conference on Advanced Computing and Emerging … , 2025 2025 Citations: 1
A Novel Random Forest-SMOTE Framework With Polynomial Feature Engineering for Early Detection of Gastrointestinal Disorders P Rahi, MR Hussain, S Alqahtani, M Husain, AP Singh, I Singh IEEE Access 13, 188577-188604 , 2025 2025 Citations: 1
An Image Synthesis Using Progressive Generative Adversarial Networks (PGANs) AP Singh, P Rahi, V Kumar Handbook of Intelligent Automation Systems Using Computer Vision and … , 2025 2025
Stock Price Prediction Using ARIMA with Option Chain Data and Technical P Rahi, MT Siddiqi, KK Komal, I Singh, AP Singh Innovations in Data Analytics: Selected Papers of ICIDA 2024, Volume 3 3, 63 , 2025 2025
Models Using LSTM P Rahi, AP Singh, I Singh Innovations in Data Analytics: Selected Papers of ICIDA 2024, Volume 2 2, 197 , 2025 2025
Reverse Image Search Using Deep Learning AP Singh, S Singla, P Rahi 2025 International Conference on Engineering Innovations and Technologies … , 2025 2025 Citations: 1
From Genes to Symptoms: Predicting Huntington’s Disease Progression with DNN P Rahi, I Singh, S Mahajan, R Chauhan, T Kaur, J Singh 2025 International Conference on Electronics, AI and Computing (EAIC), 1-9 , 2025 2025 Citations: 4
Alzheimer’s and AI Transforming Diagnostics with Regularized Neural Models A Walia, SS Kang, P Rahi, I Singh 2025 4th OPJU International Technology Conference (OTCON) on Smart Computing … , 2025 2025 Citations: 2
Towards Accurate Skin Cancer Screening: A CNN Model for Dermatological Images I Singh, P Rahi, D Sharma, Y Garg, Y Narayan 2025 International Conference on Computing for Sustainability and … , 2025 2025 Citations: 1
Predictive Modeling Techniques for Data Science I Singh, P Rahi, S Jain, D Sanghi, J Singh, A Agarwal, D Sharma International Conference On Innovative Computing And Communication, 49-67 , 2025 2025
Advancements in Image Synthesis with Progressive Generative Adversarial Networks AP Singh, N Basu, P Rahi, B Goyal, V Yadav 2025 2nd International Conference on Computational Intelligence … , 2025 2025 Citations: 1
Advanced Anxiety Risk Stratification Using Regularized Deep Neural Networks P Rahi, SS Kang, AP Singh, I Singh 2024 International Conference on Progressive Innovations in Intelligent … , 2024 2024 Citations: 6
Predictive Analytics for Stock Markets Using Graph Neural Networks (GNNs) P Rahi, D Patel, S Prabhakar, R Srivastava, P Singh 2024 International Conference on Progressive Innovations in Intelligent … , 2024 2024 Citations: 2
Decentralized Application (DApp) Implementation for an Efficient and Secure Employee Management System AP Singh, P Rahi, Bharti, V Kumar International Conference on Innovations in Data Analytics, 469-488 , 2024 2024
MOST CITED SCHOLAR PUBLICATIONS
Heart disease prediction using machine learning J Patel, AA Khaked, J Patel, J Patel Proceedings of Second International Conference on Computing, Communications … , 2021 2021 Citations: 41
Image registration concept and techniques: a review SS Bisht, B Gupta, P Rahi Int. J. Eng. Res. Appl 4 (4), 30-35 , 2014 2014 Citations: 23
Use of blockchain technology in electronic voting systems: an overview from computer security P Rahi, AP Singh, P Badoni, I Singh, R Khan 2023 International Conference on Communication, Security and Artificial … , 2023 2023 Citations: 12
An Efficient ADA Boost and CNN Hybrid Model for Weed Detection and Removal MN Ahmed, G Singh, P Badoni, R Walia, P Rahi, AT Saddiqui 2023 10th International Conference on Computing for Sustainable Global … , 2023 2023 Citations: 8
A novel paradigm in cardiovascular disease risk prediction through hybrid machine learning P Rahi, SS Kang Системная инженерия и информационные технологии 7 (3 (22)), 48-65 , 2025 2025 Citations: 6
Advanced Anxiety Risk Stratification Using Regularized Deep Neural Networks P Rahi, SS Kang, AP Singh, I Singh 2024 International Conference on Progressive Innovations in Intelligent … , 2024 2024 Citations: 6
RNN-Driven Prognosis of Heart Disease Based on Health Parameters P Rahi, I Singh, AP Singh, A Kumar 2024 Second International Conference on Advanced Computing & Communication … , 2024 2024 Citations: 6
A Random Forest Framework for Predicting Cardiovascular Disease in Diverse Populations “ P Rahi, SS Kang, AP Singh, I Singh 2024 Second International Conference on Advanced Computing & Communication … , 2024 2024 Citations: 6
Liver Disease Risk Prediction Using Regularized Deep Learning: A Novel Approach P Rahi, SS Kang, AP Singh, I Singh 2024 2nd International Conference on Advancements and Key Challenges in … , 2024 2024 Citations: 5
Forecasting mobile prices: Harnessing the power of machine learning algorithms P Badoni, R Kumar, P Rahi, APS Yadav, SK Singh Applied Data Science and Smart Systems, 348-362 , 2024 2024 Citations: 5
Performance Enhancement in Public key Cryptosystems for Security using RSA Algorithm AP Singh, P Rahi IJARCCE 5 (11), 359-362 , 2016 2016 Citations: 5
From Genes to Symptoms: Predicting Huntington’s Disease Progression with DNN P Rahi, I Singh, S Mahajan, R Chauhan, T Kaur, J Singh 2025 International Conference on Electronics, AI and Computing (EAIC), 1-9 , 2025 2025 Citations: 4
Boosting tuberculosis classification accuracy with polynomial features and random forests P Rahi, I Singh, A Kumar, T Baliyan 2024 Second International Conference on Advanced Computing & Communication … , 2024 2024 Citations: 4
Deep learning-based road sign detection: the ultimate road technology AP Singh, P Rahi, B Sahu, N Basu, H Fida, V Yadav 2024 1st International Conference on Advanced Computing and Emerging … , 2024 2024 Citations: 4
Recognition System: Detection of License Plate EI Singh, P Rahi, V Jain, N Sharma, Y Anand, AK Shukla 2024 7th International Conference on Circuit Power and Computing … , 2024 2024 Citations: 4
Advanced Stock Market Prediction Models Using LSTM P Rahi, AP Singh, I Singh, AK Singh, K Gupta, P Singla International Conference on Innovations in Data Analytics, 197-209 , 2024 2024 Citations: 3
Harnessing machine learning for precise rainfall forecasting: a comparative study A Pathania, P Rahi, A Khan, T Singh, B Singh 2024 International Conference on Advances in Computing Research on Science … , 2024 2024 Citations: 3
Cell Counting Based on Image Processing for the Detection of Cancer Clumps AP Singh, P Rahi, SP Sethi 2023 International Conference on Computing, Communication, and Intelligent … , 2024 2024 Citations: 3
Revolutionizing remote collaboration: a comprehensive review of cloud-based real-time platforms to secure teams AP Singh, P Rahi, I Singh, V Yadav online] SNSFAIT, 120-129 , 2024 2024 Citations: 3
Alzheimer’s and AI Transforming Diagnostics with Regularized Neural Models A Walia, SS Kang, P Rahi, I Singh 2025 4th OPJU International Technology Conference (OTCON) on Smart Computing … , 2025 2025 Citations: 2
