A megastudy on the predictability of personal information from facial images: Disentangling demographic and non-demographic signals.

While prior research has shown that facial images signal personal information, publications in this field tend to assess the predictability of a single variable or a small set of variables at a time, which is problematic. Reported prediction quality is hard to compare and generalize across studies due to different study conditions. Another issue is selection bias: researchers may choose to study variables intuitively expected to be predictable and underreport unpredictable variables (the 'file drawer' problem). Policy makers thus have an incomplete picture for a risk-benefit analysis of facial analysis technology. To address these limitations, we perform a megastudy-a survey-based study that reports the predictability of numerous personal attributes (349 binary variables) from 2646 distinct facial images of 969 individuals. Using deep learning, we find 82/349 personal attributes (23%) are predictable better than random from facial image pixels. Adding facial images substantially boosts prediction quality versus demographics-only benchmark model. Our unexpected finding of strong predictability of iPhone versus Galaxy preference variable shows how testing many hypotheses simultaneously can facilitate knowledge discovery. Our proposed L1-regularized image decomposition method and other techniques point to smartphone camera artifacts, BMI, skin properties, and facial hair as top candidate non-demographic signals in facial images.

Optic Neuritis After COVID-19 Vaccination: An Analysis of the Vaccine Adverse Event Reporting System.

BACKGROUND: To investigate the association of optic neuritis (ON) after the COVID-19 vaccines. METHODS: Cases of ON from Vaccine Adverse Event Reporting System (VAERS) were collected and divided into the prepandemic, COVID-19 pandemic, and COVID-19 vaccine periods. Reporting rates were calculated based on estimates of vaccines administered. Proportion tests and Pearson χ 2 test were used to determine significant differences in reporting rates of ON after vaccines within the 3 periods. Kruskal-Wallis testing with Bonferroni-corrected post hoc analysis and multivariable binary logistic regression was used to determine significant case factors such as age, sex, concurrent multiple sclerosis (MS) and vaccine manufacturer in predicting a worse outcome defined as permanent disability, emergency room (ER) or doctor visits, and hospitalizations. RESULTS: A significant increase in the reporting rate of ON after COVID-19 vaccination compared with influenza vaccination and all other vaccinations (18.6 vs 0.2 vs 0.4 per 10 million, P < 0.0001) was observed. However, the reporting rate was within the incidence range of ON in the general population. Using self-controlled and case-centered analyses, there was a significant difference in the reporting rate of ON after COVID-19 vaccination between the risk period and control period ( P < 0.0001). Multivariable binary regression with adjustment for confounding variables demonstrated that only male sex was significantly associated with permanent disability. CONCLUSIONS: Some cases of ON may be temporally associated with the COVID-19 vaccines; however, there is no significant increase in the reporting rate compared with the incidence. Limitations of this study include those inherent to any passive surveillance system. Controlled studies are needed to establish a clear causal relationship.