Evaluating the Effectiveness and Acceptability of a GPT-4o and RAG-Based Voice Chatbot for Depression Screening Using PHQ-9
This study is looking at a new voice chatbot designed to help people check for depression. This chatbot uses clever AI technology, called GPT-4o and RAG, to understand and respond to you in a natural, human-like way. It will ask you questions from a standard depression questionnaire (PHQ-9) to see how you've been feeling over the last two weeks. The goal is to see if this chatbot is easy to use and accurate in spotting depression symptoms. If successful, it could offer a new, private, and quick way for people to get an initial check on their mental health, potentially helping more people access support.
At a glance
What is this study about?
This study is testing a new way to help people check for depression using a special voice-based computer program, often called a chatbot. Many people find it hard to get help for mental health challenges due to things like feeling embarrassed, long waiting lists, or simply not knowing where to start. This new chatbot aims to make it easier for people to get an initial check on their mental health.
The chatbot uses advanced artificial intelligence (AI) to talk with you. It's built with technology that helps it understand what you say and respond in a helpful, friendly way. It also has access to up-to-date information, so its answers are reliable. The main task of the chatbot in this study is to ask you a set of nine questions from a well-known questionnaire called the PHQ-9. These questions are designed to get a picture of your mood and feelings over the last two weeks and help work out if you might be experiencing symptoms of depression. After you've answered the questions, the chatbot will tell you what your responses suggest about your mental well-being and offer some tailored advice.
Researchers want to see if this voice chatbot can be a helpful and accurate tool. They will compare the chatbot's results with a self-test you do, to make sure it's giving reliable information. If it works well, this kind of chatbot could offer a private and convenient way for many more people to check their mental health and potentially get early support.
Key takeaways
- This study tests an AI voice chatbot for depression screening.
- The chatbot uses a standard questionnaire (PHQ-9) to assess symptoms.
- It aims to provide confidential and convenient initial mental health checks.
- The study compares chatbot results with self-tests for accuracy.
- You'll get personalised suggestions for self-help after the interaction.
Who may be eligible?
This study is looking for adults aged between 18 and 65 years old. To take part, you need to be able to speak and understand English well, and have a device like a smartphone or computer with a good internet connection so the voice chatbot can work properly. You also need to be happy to chat with the chatbot and have a follow-up chat with the researchers later on.
However, some people won't be able to join. This includes anyone who has been diagnosed with serious mental health conditions like psychosis or bipolar disorder. Also, if you are currently seeing a psychiatrist for depression treatment, this study might not be right for you. Lastly, if you're not comfortable using voice technology or can't give your clear permission to take part, you won't be able to join.
Could this study suit you?
Answer these quick questions to see if you may be eligible. This is a guide only — the research team makes the final call.
- Are you between 18 and 65 years old?
- Can you speak and understand English well?
- Do you have a device with a microphone and internet access?
- Are you comfortable talking to a voice chatbot?
- Have you been diagnosed with a severe mental health condition like psychosis or bipolar disorder?
- Are you currently seeing a psychiatrist for depression?
What does participation involve?
If you join this study, your involvement will be in a few stages. First, you'll complete the standard PHQ-9 depression questionnaire on your own to set a baseline. Then, you'll interact with the voice chatbot. This interaction will start with some casual chat with the chatbot to get you comfortable. After that, the chatbot will guide you through the nine questions of the PHQ-9, just like talking to a person. It will ask you how often you've experienced certain feelings over the last two weeks. If you don't understand a question, the chatbot can help clarify it. Once you've answered all the questions, the chatbot will tell you what the results mean and offer some personalised suggestions for self-help. The total duration of the chatbot interaction is not specified, but it will be a single session. Your results from the chatbot will be compared with your initial self-test to check the chatbot's accuracy.
Potential risks and benefits
Locations (1)
- UCL Institute of Health InformaticsVerified postcodeLondon, United Kingdom
Common questions
What is a 'chatbot'?
A chatbot is a computer program designed to have conversations with people, usually through text or voice, like a digital assistant.
What is the PHQ-9 questionnaire?
The PHQ-9 is a set of nine simple questions about your mood and feelings over the last two weeks, commonly used by doctors to help check for depression.
Will my answers be kept private?
Yes, your answers will be used only for the study's research purposes and will be kept confidential. They won't be shared publicly.
What if the chatbot says I might have depression?
The chatbot will offer self-help suggestions. This study is not a substitute for seeing a doctor, and it's always a good idea to speak to your GP if you have concerns about your mental health.
Do I need any special equipment?
You just need a device like a smartphone or computer that has a microphone for voice interaction, and a stable internet connection.
How to find out more
Always speak to your GP or specialist before deciding to take part in a study.
Interested in taking part?
Discussion
Community discussion
Powered by our forum at community.patient.info. Please be respectful — this is not medical advice.