The Application of Artificial Intelligence (AI) in Mental Health

 

 

 Dr. Ruwan M Jayatunge, M.D. PhD

Artificial intelligence (AI) is used in a vast and rapidly expanding range of areas. AI entered healthcare in the 1960s and 1970s with the development of early expert systems. In 1950, Alan Turing published his influential work, titled "Computing Machinery and Intelligence," laying the foundational concepts for the field. The phrase "artificial intelligence" was coined by computer scientist John McCarthy in 1955, marking a significant milestone in the discipline.

In 1956, the term "artificial intelligence" was officially introduced at the Dartmouth Conference, marking the establishment of the field of study. Subsequent advancements, such as IBM's Deep Blue in 1997 and the emergence of machine learning in the 1990s, further boosted the development of AI technologies.

As humanity progresses, we find ourselves in the era of artificial intelligence, a reality envisioned by thinkers such as Jules Verne, Isaac Asimov, Arthur C. Clarke, and Carl Sagan.

Artificial Intelligence (AI) is the field of computer science focused on creating systems that can perform tasks demanding human intelligence, like learning, reasoning, problem-solving, perception, and decision-making, often by analyzing vast data to recognize patterns and act independently or with minimal human input.

The integration of artificial intelligence (AI) into mental health care has evolved from early theoretical work in the mid-20th century to a range of modern applications, largely to improve accessibility, efficiency, and diagnostic support for mental health conditions. AI has become an essential operational tool in mental healthcare.

AI supports mental health by improving early detection, providing accessible 24/7 support, personalizing treatment plans with real-time data, and offering tools for self-management, all through analyzing speech, text, and behaviour to identify patterns, suggest coping mechanisms, and reduce barriers like stigma or cost. 

As mental health professionals, we recognize that stigma significantly impedes individuals with mental health challenges from seeking the treatment they need. Artificial intelligence has the potential to eliminate this barrier, facilitating a more supportive environment for those in need of care.

Inability to maintain the patient's records is another hindrance we faced. I worked for 16 years in Sri Lanka as a medical doctor serving in various hospitals. There, I observed a significant lack of an accurate data management system in our government hospitals regarding patient information, their conditions, and treatment progress. This deficiency forced us to rely on traditional methods, such as maintaining paper records, which were often inefficient and prone to errors. It would have been great if we'd had AI technology back then.

The integration of artificial intelligence in mental health care enhances speed, precision, and complete effectiveness. By utilizing electronic health records, healthcare providers can prioritize individuals at high risk, enabling early detection of conditions such as depression, psychosis, and suicidal thoughts.

AI tools help predict patients' behaviour patterns and risks associated with them. We can forecast potential suicides, self-harm, or homicidal tendencies. Here, I recall a special case study. This particular patient was referred to me by Dr. Neil Fernando for a psychological assessment. He was a combatant with a traumatic brain injury and drastic personality changes. We found that this combatant had unstable moods and a potential risk for violence. Therefore, we advised the authorities to place him under observation and to refrain from issuing any weapons to him. However, these recommendations were not taken into consideration. The time passed, and within 8 months, we heard that this person committed several murders, and eventually the police arrested him. While in custody, he took his own life in the remand prison. If we had potential AI tools, we could put more pressure on the authorities and convince them. Moreover, we could have evaded a major disaster.

Some AI systems are capable of forecasting declines in mental health up to a year ahead with an impressive accuracy rate of 84%. Additionally, these systems offer personalized treatment recommendations while ensuring accessibility and confidentiality for those hesitant to seek traditional in-person care due to stigma.  

As I mentioned earlier, stigma creates fear of judgment, leading to shame, isolation, and discrimination, which delays or prevents people from seeking help, reduces treatment adherence, worsens symptoms, and leads to poorer health outcomes. Artificial intelligence (AI) helps eliminate the barrier of mental health stigma by providing anonymous, non-judgmental, and accessible platforms for seeking help.

I am delighted to say that I am now integrated into the AI-based health care monitoring system. My family physician in Toronto utilizes AI technology to provide more precise and insightful predictions regarding my health. With access to my comprehensive blood work and medical history, he is well-equipped to alert me to any emerging health risks.

AI enhances our ability to utilize psychometrics with greater effectiveness and efficiency. It allows high-precision screening tools, particularly for conditions like Depression PTSD, ADHD, Schizophrenia, etc., to achieve accuracy rates of up to 89%, and importantly, they eliminate racial and gender biases in diagnosis and treatment. We know that racial and gender biases in mental health lead to misdiagnosis, under-treatment, and mistrust. AI can help eliminate racial and gender biases in mental health by standardizing diagnostic processes, analyzing large and diverse datasets to identify and correct disparities, and offering a neutral, non-judgmental digital interface for initial screenings.

AI-driven tele-therapy and mobile applications help dismantle geographical and logistical barriers, allowing mental health services to manage millions of interactions simultaneously. Triage tools powered by AI have been shown to cut wait times by as much as 50% by effectively prioritizing high-risk patients for immediate clinical intervention.

AI has greatly enhanced the efficacy of Virtual Reality (VR) therapy by establishing secure and controlled settings for diverse therapeutic methods. This AI-driven Virtual Reality technology supports exposure therapy, effectively addressing various phobias. Additionally, it integrates Eye Movement Desensitization and Reprocessing (EMDR) to enhance trauma processing and offers modules for Cognitive Behavioural Therapy (CBT), resulting in improved treatment effectiveness.

AI-based mindfulness and stress management apps reduce stress by offering guided practices (breathing, meditation, body scans) that build present-moment awareness, helping users observe thoughts non-judgmentally to shift from reacting to responding. They improve emotional regulation, increase self-awareness of triggers, and foster self-compassion, making it easier to manage challenging situations, improve focus, and promote calmer states, thereby lowering cortisol and enhancing overall mental resilience. AI can support and enhance aspects of spiritual practice and personal growth.

Although there are new advancements associated with AI, many individuals harbour concerns that artificial intelligence may replace the human element. However, this notion is not entirely accurate; AI serves as a co-pilot, with humans maintaining leadership. Rather than replacing people, AI is designed to enhance their abilities and support their decision-making processes. While humans are prone to errors and may overlook certain blind spots in their work, AI acts as a corrective measure, positioning itself as a tool for empowerment. Although fears of a dystopian future, reminiscent of the "rise of the machines," may lead us to seek a saviour figure like John Connor, it is essential to recognize that AI is fundamentally projected to assist, not to dominate. AI is to augment human factors.

While the benefits of AI are numerous, it is important to recognize that it is not a magic bullet. AI comes with its own set of drawbacks and limitations. Therefore, I want to clarify that I do not idolize AI.  It is not a divine or superior entity.

The use of AI in mental health care presents several significant downsides. One major concern is the absence of genuine human empathy; while AI can mimic empathetic responses, it cannot grasp emotional cues or establish the therapeutic rapport that human clinicians naturally develop, which is essential for effective therapy. AI cannot establish a genuine therapeutic relationship.  

Today, many individuals rely on AI-driven virtual assistants like Siri and Alexa for their convenience. But Siri and Alexa cannot give us the human touch. Siri and Alexa do not love you.

Here, I remember one incident that occurred in February 2006 in Philadelphia. I was on my way to California, and my flight was cancelled due to a snowstorm. The blizzard grounded all the airplanes. I had to find a way to go to LA, and I was looking for possible flight options. When I called United Airlines, a young female answered me. I explained my dire situation, and she gave me several options. However, while I was talking to her, I realized that I was not talking to a human but to a robotic machine, and I became disappointed. I wanted a human connection. Despite the heavy snowfall, I went to the Philadelphia airport to seek human assistance. This indicates how we crave a human connection.

In the realm of mental health, the significance of emotional connection and trust is imperative. However, artificial intelligence lacks the capacity for empathy, compassion, and moral responsibility, which are crucial elements in fostering genuine human relationships.

Additionally, there are safety issues associated with AI-powered software that simulates human conversation, as unregulated usage can unintentionally reinforce harmful thoughts or worsen symptoms, particularly in vulnerable populations.

Privacy and data security also pose critical challenges, given that mental health data is highly sensitive, and the reliance on extensive personal information raises ethical concerns regarding misuse and breaches.  For instance, there was a significant breach of former Toronto Mayor Rob Ford's health records in 2014 when staff at multiple hospitals, including Mount Sinai, inappropriately accessed his confidential medical information while he was being treated for cancer

Furthermore, algorithmic bias is a risk, as AI models trained on non-representative data may produce biased outcomes, perpetuating inequalities for marginalized groups. Algorithms trained on Western data may fail to recognize cultural variations in symptom expression. For example, a model might flag outward sadness as the primary indicator for depression while missing "somatic" expressions (like fatigue or pain) more common in non-Western cultures.

The unregulated nature of many AI tools means they often lack clinical validation, leading to potentially inaccurate or unsafe advice. As a matter of fact, AI is ill-equipped to handle critical emergencies, such as suicidal ideation, where immediate human intervention is vital. In response to these issues, some regions, like Illinois, have begun to impose restrictions on AI use in mental health therapy, emphasizing the need for professional oversight.

There were some instances where AI failed to recognize complex and serious mental health situations. AI cannot intervene in real time, and AI cannot be held morally or legally accountable like humans. AI cannot replace trained professionals. AI can support mental health services, but it cannot replace human judgment, empathy, or responsibility.

The use of AI in the mental health field does have its limitations; however, completely discarding it in favour of traditional approaches is not a viable option. Embracing a balanced integration of both AI and conventional methods may yield more effective outcomes for mental health care.

There are shortcomings in using AI in the mental health field. But we cannot totally remove AI and go back to the old system. We cannot "throw the baby out with the bathwater. “We can't go back to the old school method.

 AI is still a developing tool, and whenever there are glitches, it has to be rectified and modified. AI would play a greater role in the mental health field, and it will be an essential and helpful tool. The future of AI in mental health involves transformative improvements in personalized, pre-emptive, and accessible care.