Conversational AI Chatbot for the Elderly and Disabled using NLP

Challenge Background

Due to the world's aging population, a critical issue is the lack of nursing facilities and caretakers. The number of potential caregivers between the ages of 45 and 64, which is the most prevalent caregiving age range, is known as the caregiver support ratio (CSR). People over the age of 80 are the subset of older persons who are most at risk of needing long-term services and support. According to research based on 2011 data from the CENSUS HUB database which includes France among other Mediterranean countries. France was found to have a CSR of 6:1. Which means it has around 6 prospective caregivers for every elderly person. This number has been drastically reduced owing to several negative influences in recent years; world wide pandemic has discouraged much physical interaction especially with the fragile elderly people. It is both economical and more effective to allow artificial intelligence systems to manage the basic aspects of interaction with the elderly.

The Problem

In this project, our goal is to create a virtual caregiver system that extracts the expression of mental and physical health states through dialogue-based human-computer interaction to support tailored treatment for elderly people and disabled. Natural language processing libraries like Natural Language Toolkit or Microsoft DialoGPT etc. makes it possible to model a conversational AI chatbot capable of assistance with daily living activities, providing advice on managing complex care, providing emotional support, participating in decision-making, and communicating with healthcare providers.

Goal of the Project

• Data cleaning: The process of data understanding, removal of outliers, missing data, and investigating the data flow for proper alignment with the set goals

• Data Intent: The goal of an intent can be to extract data from the user or process acquired data. The data may be categorical data in variables like social, mental, physical, which will be converted to formats suitable for NLP algorithms.

• Data Normalization: Prioritized compatibility of data to units or arrangements in order to avoid problems during modeling. Due to the different types of data (e.g., continuous, discrete, and categorical) present in the dataset, it is essential to normalize the data to eliminate the influence of the dimension and avoid difficulties during the model development phase.

• Contextualisation: The goal of the personalization intent is to extract personal demographic data from the user to provide a personalized experience throughout the application. therefore, it uses a question-answer format to extract information about the user (such as name, age, height, weight, etc) and populate slots in the Personalisation template

• Goal Setting: The Goal Setting intent extracts a user’s physical activity goals for the upcoming week. Following behavior change theory (mic ), the idea is to enable users to be conscious (e.g. voicing it) about the specific goals being set as a form of commitment to positive behavior change.

• The Reporting Intent: is aimed at enabling self-reporting of activities for the purpose of self-management. Depending on the type of goal being set the conversational agent must initiate a contextually relevant dialogue with the intent of extracting activities the user has undertaken during the day. For this purpose, the data obtained from the Goal Setting template is retrieved to form the context of the conversation. For instance if the user had indicated that she was playing golf on Thursday, then the agent would be able to ask how she got on with that activity on a Thursday.

Project Timeline

What you'll learn

Data Pre-processing & Data Insights. An interactive chart/plot on sample conversations. Exploratory Data Analysis (EDA) NLP: Conversation Modeling