Ever since Alan Turing’s seminal work on how computers can mimic human beings, AI or Artificial Intelligence has seen an exponential development in the field of technology and cognitive science. Artificial Intelligence at its core refers to the simulation of human intelligence and actions in computers and machines so that they can mimic our activities and think like us. These days we move about in our day-to-day lives with AI being there like an omnipresent entity. Be it SIRI on your iPhone or Alexa on your smart home device, AI is there to help us out in our day-to-day activities. You can even have conversations at length with these AI, albeit the range of responses is limited and banks on the availability of the Internet. Smart home devices now let us control all our indoor appliances using AI-driven technology, making our lives so much easier and comfortable. Online websites also use AI-driven user interface and chat box protocols to understand and study end-user better and make their experience hassle-free and smooth. Thus, it seems only natural that such a groundbreaking technology would be adapted in the field of mental health to assess or screen the users to understand their mental health needs.
The basic approach to AI is to simulate human cognitive functions so that technologies can perform the same. These functions typically involve the ability to reasoning and problem solving, recognize patterns, generalization, and predictive inferences. In the field of mental health, the clinicians engage in a series of questioning and meticulous observation to understand the symptom pattern of the client to navigate the core psychopathology, and predict the possible outcomes and how to manage the same. Often it has been noted that the clinicians, being human, tend to slip up and may at times be erroneous in their case conceptualizations. Valuable information may be lost due to slips of concentration, or delays in observation. And with Covid wreaking havoc globally, it has reduced the doctor-client interaction much more. This has further fueled the research into the application of AI in the field so that this gap can be effectively bridged.
Ways How AI Is Being Used
There are three main ways how AI is now being used in the field of mental health:
Personal Sensing or Digital Phenotyping: Digital phenotyping or personal sensing involves using sensor and usage data from personal gadgets, particularly smartphones, to infer background and behavioural information that is then used as input for machine learning to predict psychological outcomes and mental health conditions. Keystroke input methods like clicking, scrolling, and tapping have given insight regarding depression and anxiety. Reductions in the number and duration of outgoing calls and text messages have been found to be associated with relapses of schizophrenia. However, the way ahead for include context-driven and symptom-relevant personalized therapy recommendations.
Natural Language Processing of Clinical Text and Social Media Content: It is common knowledge that our voice and intonations provide valuable information regarding the psychological states of individuals. AI-driven systems can engage language processing and audio analysis to understand telling signs of mental health concerns. Transcripts of interview sessions and even social media written/audio/video content can be used in the same way to understand and predict the presence and course of mental health issues. Research into the same has indicated impoverished vocabulary, semantic incoherence, and syntactic complexity as signs of serious mental illness like schizophrenia or other forms of psychosis.
A chatbot technically tries to mimic a real-life conversation using rule-based responses to text or voice messages and may also incorporate sophisticated audio analysis protocols for the same. The simple chatbots often are used to navigate and find out information regarding therapy and therapist recommendations. Modern work on AI is now focused on incorporating sophisticated language processing techniques to provide therapeutic conversations. While they are not intended to replace the human therapist, at least they can be a bulwark in times of need. The three most prominent mental health chatbots that have emerged in recent times are Woebot, Wysa, and Tess. They primarily incorporate cognitive-behavioural and mindfulness protocols.
Pros, Cons and the Road Ahead
AI-driven protocols have the advantage of providing convenient timings (removing the traditional appointment-taking concept), an instant connection that aids the healing process, and at the same time ensuring anonymity. However, the most pertinent thing to understand here is that although AI provides a robust help in the field of detecting and providing preliminary solution protocols in clinical cases, it can only serve as a complement or supplement, rather than being a replacement for traditional therapists. In emergency situations, their response may be grossly inadequate and inappropriate. Also, the clinical manifestations of mental health concerns can be multifarious, and thus at times requires the ‘clinical eye’ which is a human ability that AI has not yet achieved. Nevertheless, the field of AI has shown improvement in leaps and bounds in the last few decades. And one can only hope that mental health treatment protocols using AI in a growing digital world would be effectively bridging the gap between the client and the treatment.
(Disclaimer: Dr JPS Bakshi, Founder, Dr Bakshi’s Healthcare)