A medication recommender system that predicts sentiment based on patient reviews using vectorization processes including Bow, TF-IDF, Word2Vec, and Manual Feature Analysis. Multiple classification algorithms were applied to recommend the best drug for a given disease. The Linear SVC classifier using TF-IDF vectorization achieved the highest accuracy of 93%, demonstrating the power of machine learning in pharmaceutical scrutiny.

Linear SVC on TF-IDF → 93% accuracy · Perceptron on Bow → 91% · LGBM on Word2Vec → 91% · Random Forest on manual features → 88%

A new computer-aided design (CAD) technique for classifying brain tumours in MRI data. The proposed approach collects characteristics from brain MRI images by exploiting the Discrete Wavelet Transform's significant energy compactness property (DWT). The Wavelet characteristics are combined with a Convolutional Neural Network (CNN) to categorise input MRI images. The proposed strategy exceeds other commonly used methods with an overall accuracy of 99.3%.

Overall accuracy: 99.3% · Sensitivity: 97.4–98.2% · Specificity: 95.5–96.1% · Dataset: Figshare (Cheng) — 233 patients, 3 tumour types

An accident detection and alert system using Arduino, GPS Receiver, and GSM module to address the critical issue of individuals being left without assistance following a road accident. The GPS Receiver identifies the vehicle's location and direction, while the GSM module sends an SMS with Google Maps coordinates to the assigned contact. The system detects severe accidents using a vibration sensor and can also identify rollovers, providing immediate location data to rescue services.

Real-time accident detection via vibration sensor · Precise GPS location transmitted via GSM · Google Maps link sent to emergency contact · Rollover detection included · Effective in high-accident regions — India, Nepal, Bangladesh