Our Research and Science
Building the future of behavioral health AI—with clinical insight, transparent benchmarking, and peer-reviewed validation.
At mpathic, our scientific foundation is rooted in decades of research across behavioral health, communication science, and machine learning. We collaborate with clinicians, researchers, and technical experts to develop AI systems that are not only high-performing—but measurable, interpretable, and clinically trustworthy. Our benchmarking process compares AI predictions to gold-standard human judgments, helping ensure that our models reflect real-world clinical reasoning and uphold the highest standards of safety and quality.
How We Benchmark Our Models
mpathic’s models are benchmarked against human expert raters using established coding systems and safety frameworks. Coders will establish interrater reliability with each other prior to establishing a ground-truth dataset or other evaluation framework to measure a model’s performance. We evaluate performance using multiple metrics to ensure consistency, reliability, and alignment with human judgment:
These evaluations span multiple domains including motivational interviewing, empathy and soft skills in medical settings, clinical risk and misconduct, and patient adherence.
Key Research Themes
Our work bridges applied science and technical rigor. The following peer-reviewed studies demonstrate our contributions across fidelity evaluation, AI ethics, and health system impact:
1. Fidelity & Skill Evaluation in Clinical Settings
- More than Reflections: Empathy in motivational interviewing includes language style synchrony between therapist and client [Lord et al., 2015]
- Common Factors in Psychotherapy: Enhancing Provider-to-Patient Dynamics to Improve Patient Outcomes. [Cerezo, et al., 2024]
- AI-based COA quality oversight in psychedelic trials. [Cerezo et al., 2025]
2. Safety, Equity & Cultural Responsiveness
- Leveraging AI to deliver culturally responsive mental health care at scale. [Cerezo et al., 2024]
- Critical behavioral traits foster peer engagement in online mental health communities. [Srivastava et al., 2023]
3. LLM Behavior, Summarization & Clinical Relevance
- Counseling summarization using knowledge-guided utterance filtering. [Srivastava et al., 2022]
- Sentiment-guided commonsense-aware response generation for mental health counseling. [Srivastava et al., 2025]
4. Multimodal & Cross-Context Use
- Enhancing non-technical performance in robotic surgery using AI. [Fernandez et al., 2025]
- System and method for increasing effective communication through evaluation of multimodal data, auto-correction and behavioral suggestions based on models from evidence-based counseling, motivational interviewing, and empathy [Lord & Bertagnolli, 2025]
- Benchmarking Commercial and Open-Source Speech AI for Speaker Attribution in Real-World Clinical Conversations [Bruzinski et al., 2025]
Publications
Learn how mpathic’s evidence-based approach is grounded in rigorous research and continuous validation. Our studies demonstrate the measurable impact of empathic communication across industries.
View full list of publications
Collaborate with our research team
Our research team actively engages in collaborations that advance the scientific understanding of empathic communication and its applications across diverse fields. We welcome inquiries from scholars, institutions, and organizations interested in joint studies, data partnerships, or applied research initiatives.