Revolutionary agentic AI-powered cognitive assessment that uses multiple collaborating AI agents to detect patterns invisible to traditional scoring — delivering up to 3× better sensitivity for early detection of cognitive changes.

Starter repository for learning how to make machine learning models with voice data.

Detecting Parkinson's Disease using Machine Learning classifiers on voiceprint recordings. Features comparative analysis of XGBoost, KNN, and SVM algorithms with >95% accuracy.

Threshold-based voice biomarker analysis using Mel Filterbank Magnitudes (MFBM)

🏆 Voice AI Hack London 2026, Overall Winner. Patients minimise. Voices don't. Real-time clinical voice intelligence that flags when what a patient says diverges from what their voice reveals. Built on Speechmatics medical STT, Thymia voice biomarkers, and Claude.

BreathySync is an AI-powered respiratory health platform that uses 6-second voice samples to generate lung metrics, predicts risks 24–48 hours ahead with a Digital Twin model, and integrates air quality tracking, sleep monitoring, Lung Gym, and ABHA support.

Python toolkit for character-level Markov modeling and linguistic biomarker discovery in dementia

Explainable Feature-Attention MLP for continuous Parkinson's UPDRS motor score regression from voice biomarkers, with SHAP interpretability and subject-wise data splitting