SBIR/STTR Award attributes
1 Implicit bias (IB) negatively impacts the medical care and health outcomes of Black patients.2 Physician IB compromises the quality of patient-physician communication, eroding trust and3 confidence, impairing treatment decisions and adherence, and causing patient harm. Studies4 show that non-Black physicians, on average, have moderate levels of IB toward Black patients.5 Since most Black patients see non-Black physicians, this means that most Black patients see6 physicians with IB towards them. The main way in which IB is communicated and reinforced is7 via paraverbal behaviors (i.e., how people deliver speech, such as tone, pitch, volume) and non-8 verbal behaviors (i.e., how people use their body, such as eye gaze, hand gestures, and body9 leaning), as opposed to verbal behaviors (i.e., what people say) during interactions. Current IB 10 training is ineffective because they rely on relatively brief interventions that seek only to in- 11 crease physicians’ awareness about their having IB. While this is a necessary first step, by itself 12 it is insufficient. IB training must also show learners the behaviors they unwittingly display due to 13 their IB, and teach them concrete remediation strategies. 14 MCI’s work under this grant will produce the first-ever effective IB training system. MPathic- 15 IBCH is a new AI-based technology that will enhance two-way communication using adaptable 16 virtual human (VH) interactions, and present detailed, actionable, personalized feedback on 17 learners’ unwitting display of negative nonverbal and paraverbal behaviors. To recognize se- 18 lected facial expressions associated with learner IB, MCI will implement a proprietary system 19 based on the Emotion Facial Action Coding System (EMFACS) which analyzes combinations of 20 facial action units for real-time emotional state analysis. We will further enhance the responsive- 21 ness of VHs with an EMFACS-based facial rigging design that parallels the detection design. 22 This SBIR FastTrack will accomplish this by: (Ph1-1) Identifying key paraverbal and nonver- 23 bal communication behaviors systematically associated with pro-White/anti-Black IB. (PH1-2) 24 Developing and evaluating an MPathic-IBCH prototype to capture a key communication behav- 25 ior identified in prior research and to differentiate medical students based on their race IB levels; 26 (Ph2-1) Developing and integrating MPathic-IBCH into a blended IB/Cultural Humility curricu- 27 lum; (Ph2-2) Building MPathic-IBCH into a fully-featured, web-deployable application with im- 28 proved nonverbal and speech prosody detection, AI, 3 full scenarios, and integrating it into a 29 learning environment; and, (Ph2-3) Deploying blended curriculum with MPathic-IBCH and evalu- 30 ating the effects on student learning, communication behaviors, and attitudes as measured by 31 SP performance ratings, scores within MPathic-IBCH, and self-reflections and experiences.Project NarrativeImplicit Bias (IB) is a toxic and pervasive problem in healthcare that contributes to the preventable harm and death of Black patients. The work performed under this grant will produce the first-ever effective IB mitigation training system by virtue of its evidence-based focus on reducing providers’ nonverbal and paraverbal expressions of IB in race discordant medical encounters. It uses a new, distributable, AI-based Virtual Human (VH) simulation technology designed to heighten learners’ engagement and their awareness of IB, presenting them with detailed, personalized feedback and concrete strategies for mitigating IB by substituting positive nonverbal and paraverbal behaviors for negative behaviors and letting them reflect and practice their new skills over time.
