AIdeas: NeuroVoice - Multimodal AI for Early Screening of Neurological Diseases
What if a 5-minute conversation could reveal early neurological risk? NeuroVoice turns voice and behavioral signals into continuous Parkinson’s monitoring - accessible to anyone, anywhere.
Team Members: @shambhvi2006, @yash4428
App Category: Social Impact
Our Vision:
What if neurological decline could be detected before it becomes visible?
Many neurological diseases, including Parkinson’s, start subtly. Changes in speech and facial movements or daily behaviours often appear long before a medical diagnosis can be made. Yet these signals are not tracked continuously.
NeuroVoice is here to change that.
Designed as an AI companion, available through WhatsApp or a mobile app, NeuroVoice converts a 5-minute daily interaction into an actionable and structured medical insight.
Through natural conversation, the app analyses vocal biomarkers like jitters, shimmer, pauses and other patterns along with facial analysis including facial rigidity, facial asymmetry and blink rate to give a combined risk score every day. These features are tracked regularly to give a risk analysis or track progression over time along with easy-to-read graphs.
For individuals without a diagnosis, NeuroVoice encourages early medical intervention. For diagnosed patients, it provides structured progression tracking between appointments for doctors and caregivers.
Our vision is to shift neurological care from occasional and episodic to continuous awareness integrated into everyday life. Our intent is not to replace doctors but to strengthen monitoring and informed conversations that happen between clinical visits.
Why this matters?
Neurological diseases often grow silently making it even harder to identify them in time.
By the time visible symptoms show up, window for early intervention and prevention is lost. Even though many families notice these changes, but they dismiss them as natural aging because of limited awareness and the absence of simple tools that can tell whether these changes require medical attention or not.
At the same time, for people who have already been diagnosed with Parkinsons don’t have any way to track their condition, whether their symptoms are stabilizing or progressing between clinical visits.
In regions like India where awareness is low, consistent follow-up is limited but even in advanced healthcare systems, continuous monitoring is still uncommon.
NeuroVoice addresses this by bringing neurological monitoring into our everyday life, for individuals this enables awareness, for caregivers it provides reassurance and for doctors it gives a way to track patterns of all their patients longitudinally between visits.
Accessibility is central to this idea therefore we have integrated NeuroVoice into WhatsApp a platform already familiar to millions of elderly users and caregivers, we aim to remove the adoption barriers and expand access to everyone.
How We Built This?
NeuroVoice starts with a basic interaction where the user sends a voice message through WhatsApp or the app. This process is securely received through Amazon API Gateway and processed through AWS Lambda with authentication provided by Amazon Cognito. The audio files are stored in Amazon S3, and the history and trends of the user are stored in Amazon DynamoDB, thus enabling the tracking of a user’s progression over time.
The workflow of AI occurs through AWS Step Functions. Amazon Transcribe converts audio files into text, and a feature layer extracts features such as jitter, shimmer, speech rate, pitch variation, and tremors from the audio files and processes them through a machine learning model hosted on Amazon SageMaker, generating neurological risk or progression insights. Amazon Bedrock powers the adaptive conversational layer, enabling contextual follow-up questions and explanations, while Amazon Polly delivers responses back in natural voice. A second-level interaction allows short facial recordings to capture additional biomarkers for richer monitoring.
In parallel, the conversational layer securely captures structured health context- including sleep quality, mood changes, motor difficulty, and sensory indicators-strengthening longitudinal analysis rather than relying on isolated data points.
Caregiver and doctor dashboards are built using AWS Amplify, with automated monthly summaries generated through Amazon EventBridge, AWS Lambda, anddelivered via Amazon SES.
The current models were trained using publicly available neurological datasets from the UCI Machine Learning Repository, demonstrating feasibility for Parkinson’s-focused modelling, with improved performance when clinical indicators such as UPDRS (Unified Parkinson's Disease Rating Scale) are incorporated.
Demo:
App
App
Whatsapp
What we are currently working on:
For the next stage of NeuroVoice, we are taking feedback directly from the individuals suffering from Parkinson’s disease. Monitoring voice and facial signals is powerful, but to take our project to the next level, we intend to track additional subtle patterns that the body reveals.
We are currently working on integrating the following elements:
- Whole-Body Movement Analysis : Short movement recordings of the whole body to analyse arm swing symmetry, stiffness, and early motor imbalance.
- Periodic Motor Assessments : Weekly and monthly motor assessments like the finger tapping test, handwriting analysis, and simple drawing tasks to track over time.
- Sleep and Physiological Monitoring: Integrating ways to track sleep through wearables and health apps so that we can analyse REM Behaviour Disorder, and also combine it with heart-rate data and symptoms to understand the daily occurrence of dystonia (disorder where muscles contract involuntarily).
- Patient Support Network: Developing a support network that connects patients at similar stages of Parkinson’s progression based on the Hoehn and Yahr scale, helping them learn from and support one another.
- Medication Response Tracking : Medication is fuel for Parkinson’s patients, so we want to implement medication response tracking where we can help in deciding when to take their medicines according to their schedule or what suits them best. We are going to integrate their calendars into our app so we can learn and suggest when to take their medication while providing smart insights for doctors.
Gamified Physiotherapy: Physiotherapy is a critical part of managing Parkinson’s disease in day to day life, as structured movements help in maintain mobility and balance. Rehabilitation focuses on exercises which help in increasing the amplitude of movement of the patient and encourages patients to perform exaggerated movements such as larger steps, wider arm swings and full body stretches to counter the small slow movements caused by Parkinson’s.
To make physiotherapy more engaging, we aim to integrate Gamified Physiotherapy into NeuroVoice by adding movement based challenges, balance games, and guided motion exercises to make physiotherapy more engaging while still delivering meaningful rehabilitation benefits.
What We Learned?
Building NeuroVoice emphasized that healthcare innovation is not about accuracy but about the accessibility and responsibility of the people suffering from that disease. One of the most important insights for us was that how subtle the early signals for Parkinson’s were and how their tracking was only meaningful when done consistently over time.
A single interaction says little, but trends tell the real story.
Another thing we learnt was how important conversation is and how using it to our favor could make it a powerful way in which we could help people. A natural AI interaction allows us to collect all the biomarkers we need along with health context while making the experience feel fun and engaging instead of clinical or intimidating.
Accessibility shaped the design decisions, particularly the integration with WhatsApp to reduce adoption barriers for aging populations.
Most importantly, we learned that responsible healthcare technology must be transparent and grounded. NeuroVoice does not claim any diagnosis, it supports early awareness and more informed conversations between patients, caregivers, and doctors. With this it represents a shift from episodic neurological assessment to continuous, accessible monitoring built for scale and inclusion.
Enjoyed reading this content? Let the author know!
Your likes, comments, shares, and saves help creators reach more builders.
Loading recommendations
Loading article