COVID19, Mental Health & Artificial Intelligence — Part2

Advances in natural language processing and the popularity of smartphones have enabled the use of AI for mental health care.

Why use AI for mental health?

Mental disorders are the costliest health condition in the United States.

AI based solutions can provide some relief to people suffering from mental health issues by providing access to faster and easier help. We are still in the initial stages of utilizing AI for mental health. However, if initial results are any indication, AI can bring some major advancements to the field of mental health.

Advantages

Using AI based solutions could help reduce some of the reasons why many people don’t receive timely intervention.

• Virtual Counselling can help alleviate shortage of therapists, social stigma associated with mental health & high costs by bringing psychotherapy to more people who need it in a more accessible and affordable fashion.
• Automated chatbots are especially useful during the current pandemic (Covid-19) with increasing anxiety and depression in addition to disruption of regular help resources due to social distancing guidelines and lockdowns.

Limitations

• Chatbots cannot replace the human element. Some people might need more detailed and individualized treatment than a virtual counselling session can provide.
• AI algorithms are trained using existing data and information, and this can introduce some implicit or explicit bias in the solutions based on them. This bias can be reduced by ensuring that the data used to train the models incorporates a wide range of cases. It should include actual clinical data in addition to research data.
• AI based chatbots and virtual agents are fairly new, and its important to evaluate and iterate to continually better the solutions.
• Privacy and safety of patient digital data is also a concern.

How is AI helping?

The biggest advantages of using AI to provide therapy is scalability (the ability to scale easily to larger groups of people) and availability (instant access 24/7 online through smartphone app or internet).

Let’s look at some existing solutions based on AI.

• Woebot

Woebot is a chatbot integrated with Facebook that resembles an instant messaging service. It tries to replicate the conversation a person might have with their therapist by asking questions to learn about them, their mood, their emotions, their thoughts. It then suggests help using cognitive behavioral therapy (CBT) techniques. The Woebot tailors to the individual’s situation. It offers private sessions that are available 24/7 instantly.

• Tess

Silicon Valley startup X2AI has created a chatbot Tess. Tess provides CBT help with anxiety and depression, by having text message conversations, similar to texting with a friend or a mental health coach.

• Ellie

Ellie is a virtual therapist created by University of Southern California . It was first launched to treat depression and post-traumatic stress syndrome in veterans. Ellie combines nonverbal signs such as facial expressions and gestures with verbal detection.

• Ginger

Ginger provides on demand mental health support. It combines machine learning and a clinical network to provide immediate, real-time support. This platform integrates clinicians with AI and offers 24/7 online CBT, where a patient can seek online help via texts, and meet with a licensed therapist or psychiatrist via video within hours.

• Lumen

Lumen is a virtual agent being developed by researchers at University of Illinois Chicago. It will be based on the same technology as Amazon’s Alexa. The app will act as a virtual mental health agent that will talk patients through problem-solving steps and provide strategies to improve their emotional health.

• Wysa

Wysa is a mental health app that coaches users on how to cope with daily stresses. It is an AI enabled life coach that provides early intervention using 3 methods — AI chatbot, a library of evidence-based self-help tools, and messaging based support from human psychologists.

There are other chatbots that provide help in different languages too.

• Emma — a chatbot that speaks Dutch, and helps with mild anxiety.
• Karim — a chatbot that speaks Arabic and has been helping Syrian refugees cope with the trauma of war.

How does a chatbot work?

Mental health chatbots use natural language processing along with Sentiment Analysis, machine learning and neural networks to analyze their conversations with the patients. They are usually based on Cognitive Behavioral Therapy methodology (CBT) which is a form of therapy to help patients manage their emotional health.

A key design component in mental health chatbots is empathetic engagement. Empathetic engagement in conversation means to act as a credible and trustworthy conversation partner that patiently hears you out and offer a detached point of view on things.

A chatbot typically kickstarts the conversation, asks direct questions to assess mood & emotions, and suggests solutions based on the responses.

It’s NLP algorithm along with Sentiment Analysis recognizes the words a patient uses. Each trigger word has a decision tree, based on its training from behavioral data samples, clinical data and chat messages with other patients. It relies on this to provide treatment recommendations, which can include either CBT, a piece of advice or contacting professional help.

Divya Sikka is a Student Ambassador in the Inspirit AI Student Ambassadors Program. Inspirit AI is a pre-collegiate enrichment program that exposes curious high school students globally to AI through live online classes. Learn more at https://www.inspiritai.com/.