# Enriching Neural Models with Targeted Features for Dementia Detection

## TL;DR

This paper presents a neural CNN-LSTM model that detects Alzheimer's disease from conversational transcripts, achieving state-of-the-art accuracy and potentially enabling earlier, less invasive diagnosis.

## Contribution

It introduces a novel neural architecture that combines targeted and implicit features for dementia detection from speech data.

## Key findings

- Achieved an F1 score of 0.929 on DementiaBank dataset.
- Outperformed previous models in dementia classification accuracy.
- Demonstrated effectiveness of combining targeted and learned features.

## Abstract

Alzheimer's disease (AD) is an irreversible brain disease that can dramatically reduce quality of life, most commonly manifesting in older adults and eventually leading to the need for full-time care. Early detection is fundamental to slowing its progression; however, diagnosis can be expensive, time-consuming, and invasive. In this work we develop a neural model based on a CNN-LSTM architecture that learns to detect AD and related dementias using targeted and implicitly-learned features from conversational transcripts. Our approach establishes the new state of the art on the DementiaBank dataset, achieving an F1 score of 0.929 when classifying participants into AD and control groups.