Hybrid CNN-LSTM Architecture For Automated Diabetic Retinopathy Detection With Clinical Explainability
Authors: Jayraj Patil, Yash Pavnekar, Siddhesh Nikam, Pratik More, Prof. Dr. Jyoti Chavan
Abstract: Every year, diabetic retinopathy (DR) threatens the vision of millions, but the screening process just can’t keep up. The current system moves slowly—specialists are overworked, results change from one doctor to the next, and too many patients learn they have DR only after their sight is already at risk. We wanted a fix. So, our team created an automated deep learning platform—a hybrid that stacks a Convolutional Neural Network (CNN) with Long Short-Term Memory (LSTM) layers. This model doesn’t just detect DR; it also grades its severity from plain retinal fundus photos. We didn’t stop after building one model. We tried three hybrid approaches—Custom CNN+LSTM, MobileNetV2+LSTM, and InceptionResNetV2+LSTM—and compared them to seven standard CNN-only baselines. To make sure even the subtle signs stand out, we used CLAHE (Contrast Limited Adaptive Histogram Equalization) on every image. Medical datasets are imbalanced by nature, so we rebalanced things through loss weighting, giving serious DR cases the extra attention they deserve. And for transparency, we turned to Grad-CAM, producing heatmaps so doctors can see exactly what our AI focused on. When it came down to results, the InceptionResNetV2+LSTM beat the rest: 91.4% accuracy, a Quadratic Cohen’s Kappa of 0.89, and a Macro F1-Score of 0.86 for multi-class DR grading. More than just numbers—two ophthalmologists validated our Grad-CAM maps and agreed with the AI’s focus 91% of the time. To make everything practical, we built a Streamlit web app layered with secure roles, live predictions, explainability, and instant PDF reports. This project pushes DR screening closer to where it needs to be. With smarter AI, clear explainability, and a clinic-ready platform, screening can be faster, fairer, and more dependable—catching DR cases that used to slip by, and backing up doctors with real confidence.
Artificial Intelligence Assisted Drug Discovery Of Noncommunicable Disease: Predictive Modelling And Optimization
Authors: Ayush Patel, Sangeeta Vhatkar, Namdeo Badhe
Abstract: AI and machine learning are shaping up drug discovery and it is about time. The old way- slow, expensive and full of dead-ends- are outdated. Tools like deep learning, graph neural networks, GANs and reinforcement learning are stepping up. These tools actually help scientists spot new targets, sift through virtual libraries for promising compounds, predict how molecules will behave, dream up brand new drug designs, find fresh uses for old drugs and even streamline clinical trials. Graph models, in particular, shine because they get the complicated shape and connections in molecules. These all let researchers simulate how tiny structures interact in the messy reality of biology. Generative AI pushes boundaries even further by designing all sorts of molecules- each tailored for certain properties- across an almost endless chemical universe. Technology is making and creating waves everywhere: cancer, heart conditions, brain disorders, infections-you name it. Across the board, the results are better predictions, smarter trade-offs, more molecular variety and a smoother path from lab to clinic. Of course, it’s not all smooth sailing. Challenges remain like messy data, black-box designing making, regulatory headaches and the tricky business of converting code into medicine. But even with those bumps, AI-powered drug discovery isn’t another upgrade. It is a real-shift: more data-driven, more scalable and a lot more personal. The evidence keeps piling up-AI is speeding up therapeutic breakthroughs and rewriting the future position of medicine, one algorithm at a time.
DOI: https://doi.org/10.5281/zenodo.19663484
Determination Of Varicose Veins Problems Using Concurrent Sensor Network With Heat Treatment Module
Authors: Karthikeyan D, Dhanush D, Harikrishnan S, Jagadesh J, Jagan K
Abstract: Varicose veins are a prevalent vascular disorder caused by weakened vein walls and malfunctioning valves, resulting in improper blood circulation and vein enlargement in the lower extremities. Early identification and timely intervention are essential to prevent complications such as venous ulcers and chronic discomfort. This paper presents a wearable healthcare system designed to detect and manage varicose vein conditions using a concurrent sensor network integrated with a heat treatment module. The proposed system employs multiple sensors, including photoplethysmography (PPG), temperature, infrared, and pressure sensors, to acquire physiological data related to blood circulation and skin temperature variations. The collected signals are processed using a microcontroller-based system that performs real-time analysis and identifies abnormal vascular patterns. Upon detecting irregularities, the system activates a controlled heat therapy module to improve blood flow and reduce discomfort. The integration of sensing and therapeutic functionality enables continuous monitoring and immediate intervention, enhancing patient convenience and reducing dependency on hospital visits. The proposed framework demonstrates the effectiveness of IoT-based wearable systems in improving vascular health monitoring and providing automated therapeutic response for varicose vein management.