IoT For All recently published an article about how AI chatbots and other technologies can be useful in contributing to mental healthcare, a particularly relevant topic given the negative implications of the Covid-19 pandemic.
In particular, the US has taken steps towards improving mental healthcare resources such as investing $1.5 billion in mental-health related startups in 2020, establishing more than 10 mental health unicorns with a valuations of $1 billion, and closing over 124 mental health startup deals in 2021. Though these are great accomplishments, perhaps many of these ventures could be enhanced with the use of artificial intelligence.
Some of the benefits AI chatbot users have expressed are its anonymity, timely support, and affordability. When it comes to mental healthcare, anonymity is huge. Many people avoid utilizing resources for “fear of being judged, especially when sensitive issues are involved”. Chatbots can help patients to overcome this fear and provide the help and care for those who need it. In fact, many patients “find it easier to open up” to a chatbot as compared to a real person. Moreover, chatbots are available to give timely support, 24/7. It is critical for resources to be available at all hours of the day to accommodate those who work unconventional hours, or who may be dealing with a mental health crises at any hour of the day. As well, particularly when discussing healthcare access in the US, affordable and accessible healthcare is not universal to and easily available to all. For instance, in the US, a “one-hour consultation with a professional can cost between $65 and $250”. Chatbots can significantly reduce the price of consultation by reducing telephone and even transportation costs.
Despite the many benefits just discussed, there are also many limitation to using chatbots. As with many AI use cases, chatbots often “struggle to understand the nuances of human language,” making it challenging particularly when it comes to understanding, caring for, and identifying mental health challenges. Relying on solely “textual information” to perceive emotion can be difficult for chatbots. Another concern with the use of chatbots in healthcare is privacy and ensuring that data sharing will not expose users and patients to privacy risks.
Overall, providers can improve chatbots by utilizing “self-supervised learning” to help chatbots to better understand the patient by combining the benefits of both supervised and unsupervised learning. Furthermore, providers should collect “conversation success metrics” and continuously monitor them to keep track of the system drawbacks. Finally, it is important to provide users with the opportunity to connect with a real human if there are communication issues, or if they so choose.
Read the full article from IoT For All.
