Our system level approach to AI innovation utilizes multiple AI models which focus on understanding the user, identifying the user needs (short term and long term) and matching the user to culturally appropriate coaching.
Feature extraction from the conversations is a core part of the workflow for classifying conversations into appropriate buckets using custom trained AI models.
Downstream from the classifiers, LLMs are utilized to generate appropriate coaching (CBTs) for presenting to the user.
We utilize Natural Language Processing and sentiment analysis as a means of adding guard rails to outgoing messages to the chatbot and incoming feedback from the chatbot. Any outgoing messages that violate the core subject matter of the chatbot and attempt to jailbreak the conversations are politely steered back to the mental health focus of the conversation.
We also implement guard rails to prevent some of the LLM models from hallucination.
We utilize randomize control trial governed by IRBs (Institutional Review Boards) to provide datapoints on the efficacy of the Tibb chatbot and our SAAS platform.
We utilize clinically validated and time tested cognitive behavioral therapy approaches to provide coaching. Depending on the subject matter of the conversation best approaches to coaching like dialectical behavioral therapy and interpersonal behavioral therapy is used where appropriate.
Citations
Brown, Richard P., and Patricia L. Gerbarg. "Sudarshan Kriya Yogic Breathing in the Treatment of Stress, Anxiety, and Depression: Part I—Neurophysiologic Model." Journal of Alternative and Complementary Medicine 14, no. 1 (2008): 11-18.
Cunha, Lúcia F., Lúcia G. Pellanda, and Christian H. Reppold. "Positive Psychology and Gratitude Interventions: A Randomized Clinical Trial." Frontiers in Psychology 10 (2019): 584.
Emmons, Robert A., and Michael E. McCullough. "Counting Blessings Versus Burdens: An Experimental Study of Gratitude and Subjective Well-Being in Daily Life." Journal of Personality and Social Psychology 84, no. 2 (2003): 377-389.
Goyal, Madhav, Sonal Singh, Erica M. S. Sibinga, Neda F. Gould, Anastasia Rowland-Seymour, Ritu Sharma, Zackary Berger, et al. "Meditation Programs for Psychological Stress and Well-being: A Systematic Review and Meta-analysis." JAMA Internal Medicine 174, no. 3 (2014): 357-368.
Koenig, Harold G. "Religion, Spirituality, and Health: The Research and Clinical Implications." ISRN Psychiatry 2012 (2012): 278730.
Laborde, Sylvain, Markus Raab, and Emma Mosley. "Can Slow-Paced Breathing Reduce Pain? An Experimental Investigation." Frontiers in Psychology 12 (2021): 663838.
Neff, Kristin D. "Self-Compassion: An Alternative Conceptualization of Positive Self-Regard." Psychological Inquiry 22, no. 4 (2011): 1-7.
O'Connell, Brenda H., Deirdre O'Shea, and Stephen Gallagher. "Enhancing Social Relationships Through Positive Psychology Activities: A Randomised Controlled Trial." The Journal of Positive Psychology 11, no. 2 (2016): 149-162.
Seppälä, E. M., et al. "Breathing-Based Meditation Decreases Posttraumatic Stress Disorder Symptoms in U.S. Military Veterans: A Randomized Controlled Longitudinal Study." Journal of Traumatic Stress 27, no. 4 (2014): 397-405.
Warber, Sara L., Katherine R. Ingerman, Stephanie L. Moura, Kelly M. Wunder, Alyson Northrop, Jon Gillespie, Kristin Durda, Katherine Smith, Paul S. Rhodes, and Katherine J. Rubenfire. "Healing Effects of Qigong Practice and Therapeutic Touch Energy Healing on Health-Related Quality of Life in Adult Patients with Chronic Pain." Explore 7, no. 4 (2011): 234-243.
Wong, Y. Joel, Jesse Owen, Nicole T. Gabana, Joshua W. Brown, Sydney McInnis, Paul Toth, and Lynn Gilman. "Does Gratitude Writing Improve the Mental Health of Psychotherapy Clients? Evidence from a Randomized Controlled Trial." Psychotherapy Research 28, no. 2 (2018): 192-202.
Zaccaro, Andrea, Andrea Piarulli, Marco Laurino, Erika Garbella, Danilo Menicucci, Bruno Neri, and Angelo Gemignani. "How Breath-Control Can Change Your Life: A Systematic Review on Psycho-Physiological Correlates of Slow Breathing." Frontiers in Human Neuroscience 12 (2018): 353. https://doi.org/10.3389/fnhum.2018.00353.
Our multiple AI models are specially trained using proprietary and open source training and test data taking care to ensure the core data is free of bias and is representative of the user population.
Our AI models performance is continuously evaluated and interpreted to understand and explain the decision making criteria of the model.