CogniSpeak

Project Overview: Early Dementia Detection Through Voice Analysis

Our project aims to leverage cutting-edge machine learning and clinical cognitive computing (MLCC) techniques to detect early signs of dementia using voice recordings. Dementia is a progressive condition that affects cognitive functions like memory, thinking, and communication, and early detection is crucial for effective intervention. More than 55 million people live with dementia, with 10 million new cases occurring each year. In addition more than 60% of dementia patients live in low and middle-income countries, highlighting the need for accessible monitoring services.

How It Works

• Voice Recording: Users are prompted to record their voice through a simple and user-friendly interface. The recordings can be short narratives, answering guided questions, or performing specific verbal tasks.
• Machine Learning Analysis: Once the recording is submitted, our system uses our own ML model we trained with advanced MLCC algorithms to analyze various aspects of the voice, including:
  • Speech Patterns: Speed, rhythm, and fluency of speech.
  • Voice Modulation: Changes in pitch, tone, and inflection.
  • Pauses and Hesitations: Frequency and duration of silent pauses or hesitations during speech.
  • Language Content: Analysis of word usage, sentence structure, and complexity of the language. All training data is open source! ### Purpose and Innovation The key goal of this project is to identify subtle changes in voice and speech patterns that may indicate early cognitive decline, which is often an early sign of dementia. By integrating MLCC with voice analysis, we aim to provide:
• Non-Invasive Screening: A simple, accessible way for users to check their cognitive health without the need for extensive clinical testing.
• Early Detection: Identifying signs of dementia at its earliest stages allows for timely medical consultation and intervention, which can significantly improve patient outcomes.
• Personalized Insights: Users receive tailored reports indicating whether their voice patterns show any signs of concern, along with recommendations for further action if needed.

Why Voice Analysis?

Research has shown that cognitive decline often manifests in subtle changes in speech patterns long before more noticeable symptoms appear. Our system focuses on detecting these changes using machine learning, enabling a quicker and non-intrusive method of early dementia detection. This innovative approach provides a low-cost, accessible tool that can help in proactive cognitive health management and potentially slow down the progression of dementia through early intervention.

Built With

• cloudflare
• defang
• fastapi
• javascript
• next
• python
• scikit-learn