Comparison of clinical geneticist and computer visual attention in assessing genetic conditions.

Artificial intelligence (AI) for facial diagnostics is increasingly used in the genetics clinic to evaluate patients with potential genetic conditions. Current approaches focus on one type of AI called Deep Learning (DL). While DL- based facial diagnostic platforms have a high accuracy rate for many conditions, less is understood about how this technology assesses and classifies (categorizes) images, and how this compares to humans. To compare human and computer attention, we performed eye-tracking analyses of geneticist clinicians (n = 22) and non-clinicians (n = 22) who viewed images of people with 10 different genetic conditions, as well as images of unaffected individuals. We calculated the Intersection-over-Union (IoU) and Kullback-Leibler divergence (KL) to compare the visual attentions of the two participant groups, and then the clinician group against the saliency maps of our deep learning classifier. We found that human visual attention differs greatly from DL model's saliency results. Averaging over all the test images, IoU and KL metric for the successful (accurate) clinician visual attentions versus the saliency maps were 0.15 and 11.15, respectively. Individuals also tend to have a specific pattern of image inspection, and clinicians demonstrate different visual attention patterns than non-clinicians (IoU and KL of clinicians versus non-clinicians were 0.47 and 2.73, respectively). This study shows that humans (at different levels of expertise) and a computer vision model examine images differently. Understanding these differences can improve the design and use of AI tools, and lead to more meaningful interactions between clinicians and AI technologies.

Educational considerations based on medical student use of polygenic risk information and apparent race in a simulated consultation.

PURPOSE: To craft evidence-based educational approaches related to polygenic risk score (PRS) implementation, it is crucial to forecast issues and biases that may arise when PRS are introduced in clinical care. METHODS: Medical students (N = 84) were randomized to a simulated primary care encounter with a Black or White virtual reality-based patient and received either a direct-to-consumer-style PRS report for 5 common complex conditions or control information. The virtual patient inquired about 2 health concerns and her genetic report in the encounter. Data sources included participants' verbalizations in the simulation, care plan recommendations, and self-report outcomes. RESULTS: When medical students received PRSs, they rated the patient as less healthy and requiring more strict advice. Patterns suggest that PRSs influenced specific medical recommendations related to the patient's concerns, despite student reports that participants did not use it for that purpose. We observed complex patterns regarding the effect of patient race on recommendations and behaviors. CONCLUSION: Educational approaches should consider potential unintentional influences of PRSs on decision-making and evaluate ways that they may be applied inconsistently across patients from different racial groups.

Teaching gene-environment interaction concepts with narrative vignettes: Effects on knowledge, stigma, and behavior motivation.

Gene-environment interaction (GxE) concepts underlie a proper understanding of complex disease risk and risk-reducing behavior. Communicating GxE concepts is a challenge. This study designed an educational intervention that communicated GxE concepts in the context of eating behavior and its impact on weight, and tested its efficacy in changing knowledge, stigma, and behavior motivation. The study also explored whether different framings of GxE education and matching frames with individual eating tendencies would result in stronger intervention impact. The experiment included four GxE education conditions and a control condition unrelated to GxE concepts. In the education conditions, participants watched a video introducing GxE concepts then one of four narrative vignettes depicting how a character's experience with eating hyperpalatable or bitter tasting food (reward-based eating drive vs. bitter taste perception scenario) is influenced by genetic or environmental variations (genetic vs. environmental framings). The education intervention increased GxE knowledge, genetic causal attributions, and empathetic concern. Mediation analyses suggest that causal attributions, particularly to genetics and willpower, are key factors that drive downstream stigma and eating behavior outcomes and could be targeted in future interventions. Tailoring GxE education frames to individual traits may lead to more meaningful outcomes. For example, genetic (vs. environmental) framed GxE education may reduce stigma toward individuals with certain eating tendencies among individuals without such tendencies. GxE education interventions would be most likely to achieve desired outcomes such as reducing stigma if they target certain causal beliefs and are strategically tailored to individual attributes.

Using Educational Videos and Perspective-Taking to Communicate Gene-By-Environment Interaction Concepts about Eating Behavior: Effects on Empathy and Weight Stigma.

OBJECTIVE: This study investigated whether education about gene-by-environment interaction (G × E) concepts could improve G × E knowledge and positively affect empathy and weight stigma. DESIGN: We conducted a randomized trial using a 2 × 2 between-subjects design. SETTING: Online. PARTICIPANTS: Five hundred eighty-two American participants from the Prolific platform. INTERVENTION: Participants were randomly assigned to watch an educational or a control video. Participants then watched a set of vignette scenarios that depicted what it is like to have a predisposition toward obesogenic eating behaviors from either a first-person or third-person perspective. MAIN OUTCOME MEASURE(S): Participants completed questionnaires measuring G × E knowledge, causal attributions, weight stigma, and empathy postintervention. ANALYSIS: Two-by-two between-subjects ANOVAs and exploratory mediation analyses were conducted. RESULTS: Participants who watched the educational video demonstrated greater G × E knowledge, reported higher empathy toward the characters in the vignette scenarios and held fewer stigmatizing attitudes (notably blame) toward individuals with higher weight. Exploratory mediation analyses indicated that the educational video led to these positive downstream effects by increasing the extent to which participants attributed genetic causes to eating behaviors. CONCLUSIONS AND IMPLICATIONS: Education about G × E causes of eating behaviors can have beneficial downstream effects on attitudes toward people with higher weight.