Evidence

Evidence behind what we build.

We evaluate our products through structured research, simulated clinical studies, user feedback and institutional implementation. This page brings together the evidence generated so far, alongside work that is still in progress.

400+ Clinician engagements across training, research and validation
1700+ Student and learner engagements
30,000+ De-identified clinical utterances in the language corpus
TRL 7 Integrated pilot system demonstrated

How we evaluate

Evidence is built in layers.

We look at evidence across four levels: whether the underlying technology performs as intended, whether the outputs align with clinically meaningful communication patterns, whether users can act on the feedback, and whether the system can be implemented consistently in real educational or healthcare settings.

Technical performance Clinical relevance User and educational utility Implementation evidence
Technology

Communication intelligence architecture

Evaluation of the core language-analysis engine, cue generation, transcript-level analysis and reporting workflows.

Clinical communication

Scenario-based consultation studies

Structured analysis of simulated consultations across communication domains including empathy, clarity, jargon, collaboration and person-centered language.

User perspective

Patient and learner feedback

Research examining how communication is understood, experienced and translated into readiness for action or self-management.

Implementation

Institutional pilots and multicenter research

Evaluation of feasibility, adoption, faculty use, longitudinal feedback and institutional reporting across training and clinical settings.

Featured evidence

Selected studies, analyses and validation work.

medRxiv, 2026
Publications Research

What Patients Reward and Punish in Physician Communication: A Framework-Based Analysis of 62,319 Indian Reviews

A framework-based analysis of 62,319 Indian patient reviews identified reassurance and empathy as the strongest positive communication predictors of recommendation, while excessive directiveness and rushedness were the strongest negative predictors.

View evidence
IDF Western Pacific Region Congress 2026
Conference research Publications Research

How People with Type 2 Diabetes Hear Insulin: Patient-Reported Communication Barriers During Insulin Initiation Discussions

A structured survey of 33 adults with type 2 diabetes examined how insulin-initiation conversations were framed, remembered and experienced, including clarity, emotional response and exploration of concerns.

View evidence
ATTD 2026
Conference research Medical Education Publications Research

Quantifying Empathy and Language Patterns in Mock Clinical Encounters Using ConversationAIly™

An analysis of 63 mock consultations examined empathy, tone, stigmatizing language and conversational structure, alongside clinician reflections after receiving ConversationAIly-generated feedback.

View evidence

Evidence archive

Explore the full evidence base.

Filter by evidence type to review published research, conference work, pilots, medical education studies and other evidence generated around SigmaMozak products and programs.

medRxiv, 2026
Publications Research

What Patients Reward and Punish in Physician Communication: A Framework-Based Analysis of 62,319 Indian Reviews

A framework-based analysis of 62,319 Indian patient reviews identified reassurance and empathy as the strongest positive communication predictors of recommendation, while excessive directiveness and rushedness were the strongest negative predictors.

View evidence
IDF Western Pacific Region Congress 2026
Conference research Publications Research

How People with Type 2 Diabetes Hear Insulin: Patient-Reported Communication Barriers During Insulin Initiation Discussions

A structured survey of 33 adults with type 2 diabetes examined how insulin-initiation conversations were framed, remembered and experienced, including clarity, emotional response and exploration of concerns.

View evidence
ATTD 2026
Conference research Medical Education Publications Research

Quantifying Empathy and Language Patterns in Mock Clinical Encounters Using ConversationAIly™

An analysis of 63 mock consultations examined empathy, tone, stigmatizing language and conversational structure, alongside clinician reflections after receiving ConversationAIly-generated feedback.

View evidence
ADA Scientific Sessions 2026
Conference research Medical Education Publications Research

Communication as a Modifiable Factor for Insulin Inertia: An Analysis of Mock Consultations Using ConversationAIly™

An analysis of 197 standardized mock consultations examined how empathy, jargon, stigmatizing language and directive framing varied across insulin initiation and titration scenarios in type 2 diabetes.

View evidence
IDF World Diabetes Congress 2025
Conference research Pilots Publications Research

ConversationAIly™: Using Person-Centered Language Cues to Improve Frequency of Follow-up Visits in Persons with Type 2 Diabetes Mellitus

A 90-day cohort evaluation in 15 persons with type 2 diabetes examined whether ConversationAIly-supported person-centered communication could shorten off-clinic time and improve follow-up attendance.

View evidence
ATTD Asia 2024
Conference research Publications Research

ConversationAIly™: Duration and Impact of Positive Conversation Cues on Compliance Behavior in an Intermittent Insulin User

A single-case evaluation examined the duration of behavior change after a ConversationAIly-supported insulin consultation, using CGM-derived postprandial glucose patterns for follow-up.

View evidence
ADA Scientific Sessions 2024
Conference research Pilots Publications Research

Optimizing In-Clinic Conversations using Data Science and AI to Promote Medication Adherence

A proof-of-concept study in 25 persons with diabetes evaluated AI-generated language cues intended to support more constructive in-clinic conversations and medication self-care.

View evidence
ESICON 2024
Conference research Pilots Research

Empowering Person-Centered In-Clinic Conversations in Obesity with ConversationAIly™: A Novel, AI-Powered HCP Support Tool

A proof-of-concept pilot in 15 persons living with obesity examined how ConversationAIly could support more person-centered, stigma-sensitive weight-related conversations in clinical practice.

View evidence
ADA Scientific Sessions 2025
Conference research Medical Education Pilots Publications Research

Changing the Narrative from Disease-Focused to Person-Centered Language in the Clinic: ConversationAIly™ as a Training Tool

A comparative pilot involving 20 endocrinology residents examined whether ConversationAIly™ could support training in person-centered language and reduce stigmatizing communication during clinical interactions.

View evidence
ATTD 2025
Conference research Pilots Publications Research

Impact of Word Choices on Person Satisfaction and Compliance Behavior Using AI-Powered HCP Tool ConversationAIly™

A pilot in 15 insulin-treated persons with type 2 diabetes examined how ConversationAIly-supported word choices influenced communication around insulin timing, administration technique, dose adjustment and hypoglycemia management.

View evidence
Evidence is presented according to its current stage of development. Findings from simulated studies, surveys, pilots, preprints and peer-reviewed publications are identified as such, and ongoing research is not presented as established clinical effectiveness.