• Extracted and analyzed a Multimodal dataset ( Facial, Linguistic & Contextual ) concerning People with Dementia

• Implemented machine learning algorithms, namely LSTM, SVM, & Random Forest to build multimodal classification model for engagement prediction

• Results demonstrated a significantly higher than chance accuracy of 88.1%

• Technologies used: Python, TensorFlow, Keras, Openface, LIWC

• Advised by Dr. Dennis Küster