Harmony

AI x Neurotech Entrepreneurship |  Explainable AI  |  VUI

16 Weeks, Jan 2024-May 2024 | 6 members | Berkeley, California




Team Members: Belinda Liu, Amelia Lubeska, Peter Laurance, Catherine, Patience, Sena
Instructor: Gail Gannon, Mike Chop
My Role: User Interview, Business Model Ideation, Audio Preprocessing Hackathon, User Experience Design, MVP Design, User Accessbility Testing

How Multimodal AI Interfaces Transform Brain Health?

Harmony is an entrepreneuship project I contributed to at UC Berkeley, where my developed machine learning models to predict dementia through voice biomarkers. As the sole designer in this interdisciplinary neurotechnology team, I was responsible for transforming AI technology into an empathetic voice user interface (VUI). This work focused on creating a context-aware, emotionally resonant interface that enabled more natural, compassionate interactions with the system, bridging the gap between advanced AI and human-centered design.



Cutting-edge AI Neurotech enables
early dementia detection just through voice

Our team has developed two machine learning-based regression models to screen for mild cognitive impairment, trained on eight voice biomarkers. These models have shown over 80% accuracy in initial testing, and we are continuing to collect data to refine and improve their precision.


Model 1: Regression
Techniques: XGBoost, KNN, Random Forest, VotingClassifier, Standard Scalar, Gaussian Weights, CrossValidation
 
Model 2: Neural Network
Techniques: LSTM, Sigmoid, GeLU, Dense, Dropout,Adam, SelectKBest, Gradient Clipping, Standard Scalar, EarlyStopping





Check Design Process


Analysis Proces Model 1: Regression
Techniques: XGBoost, KNN, Random Forest, VotingClassifier, Standard Scalar, Gaussian Weights, CrossValidation
Model 2: Neural Network
Techniques: LSTM, Sigmoid, GeLU, Dense, Dropout,Adam, SelectKBest, Gradient Clipping, Standard Scalar, EarlyStopping


More Projects.

Breathings
Chromatic Movement
Harmony
Iris